DETAILED ACTION
This action is in response to the application filed 11/01/2022. Claims 1-20 are pending and have been examined.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claim Interpretation
The following is a quotation of MPEP 2111.04 II:
The broadest reasonable interpretation of a method (or process) claim having contingent limitations requires only those steps that must be performed and does not include steps that are not required to be performed because the condition(s) precedent are not met. For example, assume a method claim requires step A if a first condition happens and step B if a second condition happens. If the claimed invention may be practiced without either the first or second condition happening, then neither step A or B is required by the broadest reasonable interpretation of the claim. If the claimed invention requires the first condition to occur, then the broadest reasonable interpretation of the claim requires step A. If the claimed invention requires both the first and second conditions to occur, then the broadest reasonable interpretation of the claim requires both steps A and B.
The broadest reasonable interpretation of a system (or apparatus or product) claim having structure that performs a function, which only needs to occur if a condition precedent is met, requires structure for performing the function should the condition occur. The system claim interpretation differs from a method claim interpretation because the claimed structure must be present in the system regardless of whether the condition is met and the function is actually performed.
Claims 15-20 refer to a “computer program product comprising a computer readable storage medium”. Paragraph [0066] of the instant Specification states, “A computer readable storage medium, as that term is used in the present disclosure, is not to be construed as storage in the form of transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide, light pulses passing through a fiber optic cable, electrical signals communicated through a wire, and/or other transmission media”. Accordingly, the computer readable storage media is not interpreted to include transitory signals per se.
The following is a quotation of 35 U.S.C. 112(f):
(f) Element in Claim for a Combination. – An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The following is a quotation of pre-AIA 35 U.S.C. 112, sixth paragraph:
An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The claims in this application are given their broadest reasonable interpretation using the plain meaning of the claim language in light of the specification as it would be understood by one of ordinary skill in the art. The broadest reasonable interpretation of a claim element (also commonly referred to as a claim limitation) is limited by the description in the specification when 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is invoked.
As explained in MPEP § 2181, subsection I, claim limitations that meet the following three-prong test will be interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph:
(A) the claim limitation uses the term “means” or “step” or a term used as a substitute for “means” that is a generic placeholder (also called a nonce term or a non-structural term having no specific structural meaning) for performing the claimed function;
(B) the term “means” or “step” or the generic placeholder is modified by functional language, typically, but not always linked by the transition word “for” (e.g., “means for”) or another linking word or phrase, such as “configured to” or “so that”; and
(C) the term “means” or “step” or the generic placeholder is not modified by sufficient structure, material, or acts for performing the claimed function.
Use of the word “means” (or “step”) in a claim with functional language creates a rebuttable presumption that the claim limitation is to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites sufficient structure, material, or acts to entirely perform the recited function.
Absence of the word “means” (or “step”) in a claim creates a rebuttable presumption that the claim limitation is not to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is not interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites function without reciting sufficient structure, material or acts to entirely perform the recited function.
Claim limitations in this application that use the word “means” (or “step”) are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action. Conversely, claim limitations in this application that do not use the word “means” (or “step”) are not being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action.
This application includes one or more claim limitations that do not use the word “means,” but are nonetheless being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, because the claim limitation(s) uses a generic placeholder that is coupled with functional language without reciting sufficient structure to perform the recited function and the generic placeholder is not preceded by a structural modifier. Such claim limitation(s) is/are:
Claim 1
Limitation 3: “an object detection component that generates a knowledge graph with vectors corresponding with nodes representative of a page layout of a document”
Limitation 4: “an evaluation component that compares the vectors to identify whether an edge is present between the vectors of the knowledge graph”
Limitation 5: “an encoder component that re-encodes the knowledge graph”
Limitation 6: “a comparison component that compares a structure of the knowledge graph with one or more other knowledge graphs corresponding to one or more other documents to determine if the document is abnormal”
Claim 2
Limitation 1: “wherein the object detection component employs layout analysis based on object detection to segment one or more pages of the document into page elements”
Claim 3
Limitation 1: “wherein the object detection component employs multimodal embedding such that the page elements of the document are represented by the nodes in the knowledge graph”
Claim 4
Limitation 1: “wherein the evaluation component determines whether the edge is present between the vectors if a similarity between two of the vectors is greater than a predetermined edge threshold”
Claim 5
Limitation 1: “wherein the evaluation component determines whether two of the vectors include the edge by comparing the multimodal embedding manner of the page elements”
Claim 6
Limitation 1: “wherein the encoder component employs pairwise comparisons based on a graph attention algorithm for the re-encoded knowledge graph and the one or more other knowledge graphs”
Because this/these claim limitation(s) is/are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, it/they is/are being interpreted to cover the corresponding structure described in the specification as performing the claimed function, and equivalents thereof.
If applicant does not intend to have this/these limitation(s) interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, applicant may: (1) amend the claim limitation(s) to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (e.g., by reciting sufficient structure to perform the claimed function); or (2) present a sufficient showing that the claim limitation(s) recite(s) sufficient structure to perform the claimed function so as to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-7 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Limitations reciting the use of a “means” or equivalent generic placeholder that is modified by functional language, and not modified by sufficient structure within the claim, is interpreted as a means-plus-function limitation under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (MPEP 2181(I) A.). For limitations interpreted under 35 U.S.C. 112(f) using means-plus-function language, the structure of the “means” or the equivalent generic placeholder substitute must be disclosed in the specification itself in a way that one skilled in the art will understand what structure will perform the recited function (MPEP 2181 (II.) A.). Additionally, for a computer-implemented means-plus-function limitation interpreted under 35 U.S.C. 112(f), the specification must disclose an algorithm for performing the claimed specific computer function (MPEP 2181 (II.) A.). Failure to adequately disclose either the structure or algorithm in sufficient detail in the specification for a computer-implemented means-plus-function limitation renders the claim indefinite under 35 U.S.C. 112(b).
As noted in the claim interpretation section above, claims 1-6 recite computer-implemented means-plus-function limitations incorporating the use of an “object detection component”, “evaluation component”, “encoder component”, and “comparison component“, generic placeholders substituting “means”. The instant specification fails to disclose any structural limitations for these generic placeholders, and would be insufficient for one of ordinary skill in the art to understand what structures could perform the recited functions. Thus, claims 1-6 are considered indefinite and are rejected under 35 U.S.C. 112(b) or pre-AIA 35 U.S.C. 112, second paragraph. This deficiencies are inherited by dependent claim 7.
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 1-20 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention.
Claim 1 discloses:
“a knowledge graph with vectors corresponding with nodes associated with respective identified page elements and their spatial layout relationships” in its third limitation. While the instant specification discloses nodes representative of a page layout of a document ([0003]) and representing page elements by nodes in a knowledge graph ([0036]), it does not disclose nodes associated with spatial layout relationships, thus this appears to be new matter.
“an encoder component that re-encodes the knowledge graph while preserving relational information derived from the page layout relationships between the nodes” in its fifth limitation. While the instant specification discloses re-encoding the knowledge graph with an encoder ([0003]), it does not disclose preserving relational information derived from the page layout between the nodes during the re-encoding. Thus, this appears to be new matter.
“a comparison component that compares a structural topology of the knowledge graph with one or more other knowledge graphs corresponding to one or more other documents to identify structural divergence indicative of abnormality to determine if the document is abnormal” in its sixth limitation. While the instant specification discloses comparing a structure of the knowledge graph with other knowledge graphs to determine if the document is abnormal ([0003]), it does not disclose comparing structural topologies or identifying structural divergence indicative of abnormalities. Thus, this appears to be new matter.
These deficiencies are inherited by dependent claims 2-7
Claim 8 discloses:
“a knowledge graph with vectors corresponding with nodes associated with the extracted page elements and their relative layout positions representative of a page layout of a document” in its second limitation. While the instant specification discloses nodes representative of a page layout of a document ([0003]) and representing page elements by nodes in a knowledge graph ([0036]), it does not disclose nodes associated with relative layout positions, thus this appears to be new matter.
“re-encoding, using the processor, the knowledge graph to produce a transformed representation that maintains layout-dependent relationships among the nodes” in its fourth limitation. While the instant specification discloses re-encoding the knowledge graph with an encoder ([0003]), it does not disclose maintaining layout-dependent relationships among the nodes. Thus, this appears to be new matter.
“comparing, using the processor, graph-level relational characteristics of the knowledge graph corresponding to one or more other documents to detect deviations from a learned or observed document layout pattern to determine if the document is abnormal” in its fifth limitation. While the instant specification discloses comparing a structure of the knowledge graph with other knowledge graphs to determine if the document is abnormal ([0003]), it does not disclose comparing graph-level relational characteristics or detecting deviations from a learned or observed document layout pattern. Thus, this appears to be new matter.
These deficiencies are inherited by dependent claims 9-14.
Claim 15 discloses:
“a knowledge graph with vectors corresponding with nodes mapped to the page element representations and the encoded positional relationships representative of a page layout of a document” in its second limitation. While the instant specification discloses nodes representative of a page layout of a document ([0003]) and representing page elements by nodes in a knowledge graph ([0036]), it does not disclose nodes associated with encoded positional relationships representative of page layout, thus this appears to be new matter.
“re-encode, using the processor, the knowledge graph into an alternative graph representation that retains layout-derived relational information” in its fourth limitation. While the instant specification discloses re-encoding the knowledge graph with an encoder ([0003]), it does not disclose retaining layout-derived relational information during the re-encoding. Thus, this appears to be new matter.
“compare, using the processor, relational patterns of the alternative graph representation with one or more other knowledge graphs corresponding to one or more other documents to identify anomalous document structures to determine if the document is abnormal” in its fifth limitation. While the instant specification discloses comparing a structure of the knowledge graph with other knowledge graphs to determine if the document is abnormal ([0003]), it does not disclose comparing relational patterns of an encoded graph with other graphs to identify anomalous document structures. Thus, this appears to be new matter.
These deficiencies are inherited by dependent claims 16-20.
Limitations reciting the use of a “means” or equivalent generic placeholder that is modified by functional language, and not modified by sufficient structure within the claim, is interpreted as a means-plus-function limitation under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (MPEP 2181(I) A.). For limitations interpreted under 35 U.S.C. 112(f) using means-plus-function language, the written description under 35 U.S.C. 112(a) must adequately link or associate particular structure, material, or acts to perform the function or it must be clear based on the facts of the application that one skilled in the art would have known what structure, material, or acts disclosed in the specification perform the recited function (MPEP 2163(II) A. (3)).
Claims 1-7 recite computer-implemented means-plus-function limitations incorporating the use of an “object detection component”, “evaluation component”, “encoder component”, and “comparison component“, generic placeholders substituting “means”. As noted above, these claims are rejected under 35 U.S.C. 112(b) as being indefinite for failing to adequately disclose the corresponding structures or algorithms in sufficient detail in the specification. When a claim containing a computer-implemented 35 U.S.C. 112(f) claim limitation is found to be indefinite under 35 U.S.C. 112(b) for failure to disclose sufficient corresponding structure in the specification that performs the entire claimed function, it will also lack written description under 35 U.S.C. 112(a). See MPEP § 2163.03, subsection VI. Thus, these claims are rejected under 35 U.S.C. 112(a) for lack of written description. These deficiencies are inherited by dependent claim 7.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed inventions are directed to non-statutory subject matter without significantly more.
Claim 1
Step 1: The claim recites “A system”, and is therefore directed to the statutory category of machine
Step 2A Prong 1: The claim recites the following judicial exception(s)
an object detection component that identifies page elements of the document based on a page layout analysis and generates a knowledge graph with vectors corresponding with nodes associated with respective identified page elements and their spatial layout relationships representative of a page layout of a document: This can be performed as a mental process. One can merely scrutinize a document and imagine a graph representing it, each node corresponding to a page element and containing a vector of spatial layout information about it.
an evaluation component that explicitly determines whether an edge is present between the vectors of the knowledge graph based on a similarity between the vectors exceeding a defined threshold that reflects a relationship between the corresponding page elements: This can be performed as a mental process. One can merely imagine a threshold value, calculate the difference between one dimension of every pair of vectors, and assign edges for all pair differences over the threshold value.
an encoder component that re-encodes the knowledge graph while preserving relational information derived from the page layout relationships between the nodes: This can be performed as a mental process. One can merely change the order of dimensions of vectors associated with the knowledge graph.
a comparison component that compares a structural topology of the knowledge graph with one or more other knowledge graphs corresponding to one or more other documents to identify structural divergence indicative of abnormality to determine if the document is abnormal: This can be performed as a mental process. One can merely contrast the structure of the knowledge graph with other knowledge graphs, deciding the other graph is abnormal if it differs significantly.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the following additional element(s)
a memory that stores computer executable components; and a processor, operably coupled to the memory, and that executes the computer executable components stored in the memory, wherein the computer executable components comprise: This is mere instruction to execute the judicial exception(s) with generic computer hardware (MPEP 2106.05(f)).
an object detection component that identifies page elements of the document based on a page layout analysis and generates a knowledge graph with vectors corresponding with nodes associated with respective identified page elements and their spatial layout relationships representative of a page layout of a document: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
an evaluation component that explicitly determines whether an edge is present between the vectors of the knowledge graph based on a similarity between the vectors exceeding a defined threshold that reflects a relationship between the corresponding page elements: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
an encoder component that re-encodes the knowledge graph while preserving relational information derived from the page layout relationships between the nodes: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
a comparison component that compares a structural topology of the knowledge graph with one or more other knowledge graphs corresponding to one or more other documents to identify structural divergence indicative of abnormality to determine if the document is abnormal: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
Step 2B: The following additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
a memory that stores computer executable components; and a processor, operably coupled to the memory, and that executes the computer executable components stored in the memory, wherein the computer executable components comprise: This is mere instruction to execute the judicial exception(s) with generic computer hardware (MPEP 2106.05(f)).
an object detection component that identifies page elements of the document based on a page layout analysis and generates a knowledge graph with vectors corresponding with nodes associated with respective identified page elements and their spatial layout relationships representative of a page layout of a document: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
an evaluation component that explicitly determines whether an edge is present between the vectors of the knowledge graph based on a similarity between the vectors exceeding a defined threshold that reflects a relationship between the corresponding page elements: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
an encoder component that re-encodes the knowledge graph while preserving relational information derived from the page layout relationships between the nodes: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
a comparison component that compares a structural topology of the knowledge graph with one or more other knowledge graphs corresponding to one or more other documents to identify structural divergence indicative of abnormality to determine if the document is abnormal: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
Claim 2
Step 1: The claim recites a machine, as in claim 1
Step 2A Prong 1: The claim recites the following further judicial exception(s)
wherein the object detection component employs layout analysis based on object detection to segment one or more pages of the document into page elements: Generating a knowledge graph can still be performed as a mental process. One can merely imagine a graph representing a document, each node corresponding to a page element and containing a vector of information about it.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the further additional element(s)
wherein the object detection component employs layout analysis based on object detection to segment one or more pages of the document into page elements: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
Step 2B: The further additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
wherein the object detection component employs layout analysis based on object detection to segment one or more pages of the document into page elements: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
Claim 3
Step 1: The claim recites a machine, as in claim 2
Step 2A Prong 1: The claim recites the following further judicial exception(s)
wherein the object detection component employs multimodal embedding such that the page elements of the document are represented by the nodes in the knowledge graph: Generating a knowledge graph can still be performed as a mental process. One merely has to assign nodes to a document containing multiple forms of media (images, text).
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the further additional element(s)
wherein the object detection component employs multimodal embedding such that the page elements of the document are represented by the nodes in the knowledge graph: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
Step 2B: The further additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
wherein the object detection component employs multimodal embedding such that the page elements of the document are represented by the nodes in the knowledge graph: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
Claim 4
Step 1: The claim recites a machine, as in claim 3
Step 2A Prong 1: The claim recites the following further judicial exception(s)
wherein the evaluation component determines whether the edge is present between the vectors if a similarity between two of the vectors is greater than a predetermined edge threshold: Identifying present edges can still be performed as a mental process. One can merely imagine a threshold value, calculate the difference between one dimension of every pair of vectors, and assign edges for all pair differences over the threshold value.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the further additional element(s)
wherein the evaluation component determines whether the edge is present between the vectors if a similarity between two of the vectors is greater than a predetermined edge threshold: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
Step 2B: The further additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
wherein the evaluation component determines whether the edge is present between the vectors if a similarity between two of the vectors is greater than a predetermined edge threshold: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
Claim 5
Step 1: The claim recites a machine, as in claim 4
Step 2A Prong 1: The claim recites the following further judicial exception(s)
wherein the evaluation component determines whether two of the vectors include the edge by comparing the multimodal embedding manner of the page elements: Identifying present edges can still be performed as a mental process. If the vectors correspond to document elements of different mediums, one merely needs to mentally compare the differences as per the mental process described for claim 4.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the further additional element(s)
wherein the evaluation component determines whether two of the vectors include the edge by comparing the multimodal embedding manner of the page elements: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
Step 2B: The further additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
wherein the evaluation component determines whether two of the vectors include the edge by comparing the multimodal embedding manner of the page elements: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
Claim 6
Step 1: The claim recites a machine, as in claim 4
Step 2A Prong 1: The claim recites no further judicial exception(s)
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the further additional element(s)
wherein the encoder component employs pairwise comparisons based on a graph attention algorithm for the re-encoded knowledge graph and the one or more other knowledge graphs: This is mere instruction to apply a graph attention algorithm to the graph in a generic manner (MPEP 2106.05(f)).
Step 2B: The further additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
wherein the encoder component employs pairwise comparisons based on a graph attention algorithm for the re-encoded knowledge graph and the one or more other knowledge graphs: This is mere instruction to apply a graph attention algorithm to the graph in a generic manner (MPEP 2106.05(f)).
Claim 7
Step 1: The claim recites a machine, as in claim 6
Step 2A Prong 1: The claim recites the following further judicial exception(s)
wherein if an abnormal score of a node of the document is significantly higher than one or more other nodes of the one or more other documents, the document is abnormal: This can be performed as a mental process. One need only mentally assign documents with nodes that have abnormal scores higher than other documents as “abnormal”.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the additional element(s)
Step 2B: The additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
Claim 8
Step 1: The claim recites “A computer implemented method”, and is therefore directed to the statutory category of process
Step 2A Prong 1: The claim recites the following judicial exception(s)
extracting, using a processor coupled to memory, page elements from a document based on a page layout analysis: This can be performed as a mental process. One can merely observe a document’s page layout.
generating, using a processor coupled to memory, a knowledge graph with vectors corresponding with nodes associated with the extracted page elements and their relative layout positions representative of a page layout of a document: This can be performed as a mental process. One can merely scrutinize a document and imagine a graph representing it, each node corresponding to a page element and containing a vector of spatial layout information about it.
evaluating, using the processor, pairwise similarities between the vectors and explicitly establishing edges in the knowledge graph when a similarity criterion indicative of a relationship between corresponding page elements is satisfied: This can be performed as a mental process. One can merely imagine a threshold value, calculate the difference between one dimension of every pair of vectors, and assign edges for all pair differences over the threshold value.
re-encoding, using the processor, the knowledge graph to produce a transformed representation that maintains layout-dependent relationships among the nodes: This can be performed as a mental process. One can merely change the order of dimensions of vectors associated with the knowledge graph.
comparing, using the processor, the knowledge graph to produce a transformed representation that maintains layout-dependent relationships among the nodes: This can be performed as a mental process. One can merely contrast the structure of the knowledge graph with other knowledge graphs, deciding the other graph is abnormal if it differs significantly.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the following additional element(s)
extracting, using a processor coupled to memory, page elements from a document based on a page layout analysis: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
generating, using a processor coupled to memory, a knowledge graph with vectors corresponding with nodes associated with the extracted page elements and their relative layout positions representative of a page layout of a document: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
evaluating, using the processor, pairwise similarities between the vectors and explicitly establishing edges in the knowledge graph when a similarity criterion indicative of a relationship between corresponding page elements is satisfied: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
re-encoding, using the processor, the knowledge graph to produce a transformed representation that maintains layout-dependent relationships among the nodes: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
comparing, using the processor, the knowledge graph to produce a transformed representation that maintains layout-dependent relationships among the nodes: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
Step 2B: The following additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
extracting, using a processor coupled to memory, page elements from a document based on a page layout analysis: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
generating, using a processor coupled to memory, a knowledge graph with vectors corresponding with nodes associated with the extracted page elements and their relative layout positions representative of a page layout of a document: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
evaluating, using the processor, pairwise similarities between the vectors and explicitly establishing edges in the knowledge graph when a similarity criterion indicative of a relationship between corresponding page elements is satisfied: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
re-encoding, using the processor, the knowledge graph to produce a transformed representation that maintains layout-dependent relationships among the nodes: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
comparing, using the processor, the knowledge graph to produce a transformed representation that maintains layout-dependent relationships among the nodes: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
Claims 9-14
Step 1: Claims 9-14 recite a process, as in claim 8.
Step 2A Prong 1: Claims 9-14 recite the same judicial exception(s) as claims 2-7, respectively.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through any additional elements. The analysis of claims 9-14 at this step mirrors that of claims 2-7. Claims 9-14 comprise all limitations of claims 2-7, respectively.
Step 2B: The additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s). The analysis of claims 9-14 at this step mirrors that of claims 2-7. Claims 9-14 comprise all limitations of claims 2-7, respectively.
Claim 15
Step 1: The claim recites “A computer program product”, and is therefore directed to the statutory category of article of manufacture
Step 2A Prong 1: The claim recites the following judicial exception(s)
derive page element representations that encode positional relationships within a page layout of the document: This can be performed as a mental process. One can merely observe and imagine a document’s page layout.
generate, using the processor coupled to memory, a knowledge graph with vectors corresponding with nodes mapped to the page element representations and the encoded positional relationships representative of a page layout of a document: This can be performed as a mental process. One can merely scrutinize a document and imagine a graph representing it, each node corresponding to a page element and containing a vector of spatial layout information about it.
determine edge connectivity within the knowledge graph by applying a similarity evaluation to the vectors to selectively form edges indicative of relationships between corresponding page elements: This can be performed as a mental process. One can merely imagine a threshold value, calculate the difference between one dimension of every pair of vectors, and assign edges for all pair differences over the threshold value.
re-encode, using the processor, the knowledge graph into an alternative graph representation that retains layout-derived relational information: This can be performed as a mental process. One can merely change the order of dimensions of vectors associated with the knowledge graph.
compare, using the processor, relational patterns of the alternative graph with one or more other knowledge graphs corresponding to one or more other documents to determine if the document is abnormal: This can be performed as a mental process. One can merely contrast the structure of the knowledge graph with other knowledge graphs, deciding the other graph is abnormal if it differs significantly.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through the following additional element(s)
obtain a digital document: This amounts to mere reception of data and is insignificant extra-solution activity (MPEP 2106.05(g)).
generate, using the processor coupled to memory, a knowledge graph with vectors corresponding with nodes mapped to the page element representations and the encoded positional relationships representative of a page layout of a document: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
re-encode, using the processor, the knowledge graph into an alternative graph representation that retains layout-derived relational information: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
compare, using the processor, relational patterns of the alternative graph with one or more other knowledge graphs corresponding to one or more other documents to determine if the document is abnormal: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
Step 2B: The following additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s)
obtain a digital document: This is an instance of retrieving information from memory, a limitation known to be well-understood, routine, and conventional (MPEP 2106.05(d) II. iv.).
generate, using the processor coupled to memory, a knowledge graph with vectors corresponding with nodes mapped to the page element representations and the encoded positional relationships representative of a page layout of a document: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
re-encode, using the processor, the knowledge graph into an alternative graph representation that retains layout-derived relational information: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
compare, using the processor, relational patterns of the alternative graph with one or more other knowledge graphs corresponding to one or more other documents to determine if the document is abnormal: This is mere instruction to apply a judicial exception with a generic computing component (MPEP 2106.05(f)).
Claims 16-20
Step 1: Claims 16-20 recite an article of manufacture, as in claim 15.
Step 2A Prong 1: Claims 16-20 recite the same judicial exception(s) as claims 2-6, respectively.
Step 2A Prong 2: The judicial exception(s) are not integrated into a practical application through any additional elements. The analysis of claims 16-20 at this step mirrors that of claims 2-6, respectively, with the exception that claims 16-20 are directed to “A computer program product abnormal document self-discovery, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to”, performing operations mirroring those of claims 2-6. This is a mere instruction to apply the exceptions using generic computer equipment (MPEP 2106.05(f)).
Step 2B: The additional element(s) of the claim, taken alone or in combination, do not amount to significantly more than the recited judicial exception(s). The analysis of claims 16-20 at this step mirrors that of claims 2-6, with the exception that claims 16-20 are directed to “A computer program product abnormal document self-discovery, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to”, performing operations mirroring those of claims 2-6. This is a mere instruction to apply the exceptions using generic computer equipment (MPEP 2106.05(f)).
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Liang et al. (Logical Labeling of Document Images Using Layout Graph Matching with Adaptive Learning, published 2002, DAS 2002, LNCS 2423, pp. 224–235), hereafter referred to as Liang, in view of Peng et al. (ANOMALOUS: A Joint Modeling Approach for Anomaly Detection on Attributed Networks, published 2018, IJCAI’18: Proceedings of the 27th International Joint Conference on Artificial Intelligence Pages 3513-3519), hereafter referred to as Peng, and further in view of Fan et al. (ANOMALYDAE: DUAL AUTOENCODER FOR ANOMALY DETECTION ON ATTRIBUTED NETWORKS, published 2020, arXiv:2002.03665v2), hereafter referred to as Fan, and Belligundu (METHOD AND SYSTEM FOR ANOMALY DETECTION USING MULTIMODAL KNOWLEDGE GRAPH, filed 11/2/2020, US 2024/0153059 A1), hereafter referred to as Belligundu.
Regarding claim 1, Liang discloses [a] system comprising:
an object detection component that identifies page elements of the document based on a page layout analysis and generates a knowledge graph with vectors corresponding with nodes associated with respective identified page elements and their spatial layout relationships representative of a page layout of a document:
PNG
media_image1.png
400
772
media_image1.png
Greyscale
”System overview” (Liang, page 225, Fig. 1)
“Fig. 1 shows an overview of our document analysis system. First, document images are processed by a segmentation-and-OCR engine (object detection component)” (Liang, page 225, paragraph 1). It should be understood for following limitations that all functions performed by this system are being executed by some component(s) of this system.
PNG
media_image2.png
200
400
media_image2.png
Greyscale
PNG
media_image3.png
200
400
media_image3.png
Greyscale
“Original document page and converted HTML result” (Liang, page 227, Fig. 3)
PNG
media_image4.png
198
708
media_image4.png
Greyscale
”Example layout and layout graphs” (Liang, page 227, Fig. 4)
“A layout graph (knowledge graph) is a fully connected attributed relational graph. Each node corresponds to a segmented block on a page. The attributes (vector) of a node are the position (spatial layout position) and size of the bounding box, and the normalized font size (small, middle, or large as compared to the average font size over the whole page). An edge between a pair of nodes reflects the spatial relationship between two corresponding blocks in the image.” (Liang, page 227, paragraph 1)
an evaluation component that explicitly determines whether an edge is present between the vectors of the knowledge graph based on a similarity between the vectors exceeding a defined threshold that reflects a relationship between the corresponding page elements:
“An edge between a pair of nodes reflects the spatial relationship between two corresponding blocks in the image” (Liang, page 227, paragraph 1)
“The attributes of edge AB in the left graph are shown in Fig. 5” (Liang, page 228, paragraph 1)
PNG
media_image5.png
373
767
media_image5.png
Greyscale
”Example edge attributes” (Liang, page 228, Fig. 5). To store these attributes, the block corresponding to node A inherently must be compared to the block corresponding to node B.
an encoder component that re-encodes the knowledge graph while preserving relational information derived from the page layout relationships between the nodes:
“Our approach is a two-step approximate solution that aims at sub-optimal N-1 match. First, we search for the best 1-1 match from U to M, where U is the candidate graph, and M is the model graph” (Liang, page 229, paragraph 7). The candidate knowledge graph is re-encoded into the model knowledge graph.
“We need a metric to measure which mapping is the best. For a given mapping, an intermediate layout graph, T, is first constructed based on U such that the mapping between T and M is 1-1. Then a cost is computed for the 1-1 mapping and defined as the quality measurement of the mapping between U and M. The best match is the one with minimal cost.” (Liang, page 228, paragraph 4)
“For a pair of mapped nodes, the cost is defined as the sum of differences between corresponding attributes (positional relationships), weighted by the weight factors in model node.” (Liang, page 228, paragraph 5); “A cost is similarly defined for a pair of edges (spatial relationships)” (Liang, page 229, paragraph 2); “The graph match cost is the sum of all node pair costs and edge pair costs” (Liang, page 229, paragraph 3). Graph match cost is minimized to encode the final model graph, in doing so, preserving relational information between nodes and edges by minimizing their differences.
While Liang fails to disclose the further limitations of the claim, Peng, in combination with Liang, discloses a system comprising an evaluation component that explicitly determines whether an edge is present between the vectors of the knowledge graph based on a similarity between the vectors exceeding a defined threshold that reflects a relationship between the corresponding page elements: (Peng)
“We first give the formal definition of anomaly detection on attributed networks : suppose U = {u1, u2, … , un} indicates a set of n instances (nodes), each instance is affiliated with a set of d-dimensional attributes F = {f1, f2, … , fd} (vectors) … these instances are interconnected with each other to form a network, and we use the adjacency matrix
A
∈
R
n
×
n
to describe the link relationships (edge[s]) between them, where A(i, j) = 1 indicates ui and uj is connected with each other” (Peng, page 3514, right column, paragraph 2).
“Based on Homophily, we require that if two instances are connected in the network, their attribute patterns in the residual matrix
R
~
ought to be similar after attribute reconstruction. Formally, we achieve network structure modeling by minimizing
PNG
media_image6.png
41
564
media_image6.png
Greyscale
” (Peng, page 3515, left column, paragraph 1). By minimizing this expression, a degree of similarity between the attributes of connected nodes in the graph is enforced. This enforced degree of similarity can be described as a defined threshold.
Peng relates to anomaly detection in knowledge graphs and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Liang to enforce attribute similarity between graph-connected nodes, as disclosed by Peng. The vast majority of graph anomaly detection work relies on Homophily as an assumption, requiring this strong correlation between graph connections and attribute similarity. Non-Homophilic structures can lead to adverse effects on anomaly detection. See Peng, page 3513, right column, paragraph 2.
While Peng fails to disclose the further limitations of the claim, Fan, in combination with Liang, discloses a comparison component that compares a structural topology of the knowledge graph with one or more other knowledge graphs corresponding to one or more other documents to identify structural divergence indicative of abnormality to determine if the document is abnormal:
(Fan) “Given an attributed network G = {V, E, X} (knowledge graph), our goal is to detect the nodes that are rare and differ significantly from the majority of the reference nodes in terms both the structure (structural topology (divergence)) and attribute information of the nodes. More formally, we aim to learn a score function
f
:
V
i
→
y
i
∈
R
, to classify sample
x
i
based on the threshold
λ
:
PNG
media_image7.png
78
544
media_image7.png
Greyscale
where
y
i
denotes the label of sample
x
i
, with 0 being the normal class and 1 the anomalous (abnormal) class“ (Fan, page 2, left column, paragraph 2)
(Fan) “the anomaly score
S
V
i
of node
V
i
is defined as the reconstruction error from both network structure and node attribute perspective
PNG
media_image8.png
35
565
media_image8.png
Greyscale
Based on the measured anomaly scores, the threshold
λ
in Eq. 1 can be determined according to distribution of scores, e.g. the nodes of top-k scores are classified as anomalous nodes” (Fan, page 3, right column, paragraph 3). The abnormality threshold can be determined based on the distribution of scores measured from the graph(s).
“Three commonly used real-world datasets [14] are used in this paper to evaluate the proposed method, including Blog- Catalog, Flickr, and ACM” (Fan, page 4, left column, paragraph 1). This method can be executed on multiple graphs (one or more other knowledge graphs).
Liang and Fan relate to analysis of knowledge graphs and are analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Liang and Peng to detect graph abnormalities with Fan’s method. Fan’s method enables the detection of nodes (and by extension, graphs) that deviate significantly from a majority of reference nodes / graphs, while more effectively capturing useful cross-modality interactions between graph structures and node attributes often neglected in similar art. Fan’s method significantly outperforms many contemporary state-of-the-art methods on multiple datasets. See Fan, page 1, Abstract; and page 4, left column, paragraph 3.
While Fan fails to disclose the further limitations of the claim, Belligundu discloses [a] system comprising: a memory that stores computer executable components; and a processor, operably coupled to the memory, and that executes the computer executable components stored in the memory, wherein the computer executable components comprise: “According to a fourth aspect, the object of the disclosure is achieved by a computer program product (non-transitory computer readable storage medium (memory) having instructions (component[s]), which when executed by a processor, perform actions) for detecting anomalies associated with a plurality of data objects of a technical installation” (Belligundu, [0017]).
Belligundu relates to anomaly detection of knowledge graphs and is analogous to the claimed invention. The combination of Liang, Peng, and Fan teaches a system for detecting anomalies in document knowledge graphs. The claimed invention improves upon this method by storing it in the form of instructions on computer hardware. Belligundu teaches hardware for storing and executing methods for detecting anomalies in knowledge graphs, applicable to the combination of Liang, Peng, and Fan. A person of ordinary skill in the art would have recognized that storing the system of Liang, Peng, and Fan as computer instructions on Belligundu’s hardware would lead to the predictable result of the method being executable by a computing system, and would improve the known device by allowing it to be performed with real data (MPEP 2143 I. (D) Applying a known technique to a known device (method, or product) ready for improvement to yield predictable results).
Regarding claim 2, the rejection of claim 1 in view of Liang, Peng, Fan, and Belligundu is incorporated. Liang further discloses a method, wherein the object detection component employs layout analysis based on object detection to segment one or more pages of the document into page elements: “The task of logical labeling is to label segmented blocks (page elements) on a document image as title, author, header, text column, etc. The set of labels will depend on document classes and/or applications“ (Liang, page 224, paragraph 2); “Fig. 1 shows an overview of our document analysis system. First, document images are processed by a segmentation-and-OCR engine (object detection component). We assume that the results are reasonably good. In particular, mild over-segmentation is acceptable, while under-segmentation which crosses logical content is not welcome. The outcome XML file (PHY-XML in the figure) contains information about the physical layout and text content of the original document page. The LOG-XML in the figure stands for logical structure XML file, which contains information about document class and logical labels corresponding to the PHY-XML file” (Liang, page 225, paragraph 1).
Regarding claim 3, the rejection of claim 2 in view of Liang, Peng, Fan, and Belligundu is incorporated. Liang further discloses a system, wherein the object detection component employs multimodal embedding such that the page elements of the document are represented by the nodes in the knowledge graph:
PNG
media_image9.png
1475
1070
media_image9.png
Greyscale
“Example image model and labeling result (II)” (Liang, page 234, Fig. 11). A multimodal input (a document with text and images) is segmented into blocks for a knowledge graph.
Regarding claim 4, the rejection of claim 3 in view of Liang, Peng, Fan, and Belligundu is incorporated. Peng, in combination with Liang, further discloses a system, wherein the evaluation component determines whether the edge is present between the vectors if a similarity between two of the vectors is greater than a predetermined edge threshold: (Peng)
“We first give the formal definition of anomaly detection on attributed networks : suppose U = {u1, u2, … , un} indicates a set of n instances (nodes), each instance is affiliated with a set of d-dimensional attributes F = {f1, f2, … , fd} (vectors) … these instances are interconnected with each other to form a network, and we use the adjacency matrix
A
∈
R
n
×
n
to describe the link relationships (edge[s]) between them, where A(i, j) = 1 indicates ui and uj is connected with each other” (Peng, page 3514, right column, paragraph 2).
“Based on Homophily, we require that if two instances are connected in the network, their attribute patterns in the residual matrix
R
~
ought to be similar after attribute reconstruction. Formally, we achieve network structure modeling by minimizing
PNG
media_image6.png
41
564
media_image6.png
Greyscale
” (Peng, page 3515, left column, paragraph 1). By minimizing this expression, a degree of similarity between the attributes of connected nodes in the graph is enforced. This enforced degree of similarity can be described as a predetermined edge threshold.
Peng relates to anomaly detection in knowledge graphs and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Liang, Peng, Fan, and Belligundu to enforce attribute similarity between graph-connected nodes, as disclosed by Peng. The vast majority of graph anomaly detection work relies on Homophily as an assumption, requiring this strong correlation between graph connections and attribute similarity. Non-Homophilic structures can lead to adverse effects on anomaly detection. See Peng, page 3513, right column, paragraph 2.
Regarding claim 5, the rejection of claim 4 in view of Liang, Fan, Belligundu, and Peng is incorporated. Peng, in combination with Liang, further discloses a system, wherein the evaluation component determines whether two of the vectors include the edge by comparing the multimodal embedding manner of the page elements: (Peng) “Formally, we achieve network structure modeling by minimizing
PNG
media_image6.png
41
564
media_image6.png
Greyscale
” (Peng, page 3515, left column, paragraph 1). In this expression, the attributes (embeddings) of two nodes, i and j, are compared to enforce edge connections between similar nodes. In the combination with Liang, the embeddings would be of multi-modal page elements.
Peng relates to anomaly detection in knowledge graphs and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Liang, Peng, Fan, and Belligundu to enforce attribute similarity between graph-connected nodes, as disclosed by Peng. The vast majority of graph anomaly detection work relies on Homophily as an assumption, requiring this strong correlation between graph connections and attribute similarity. Non-Homophilic structures can lead to adverse effects on anomaly detection. See Peng, page 3513, right column, paragraph 2.
Regarding claim 6, the rejection of claim 4 in view of Liang, Fan, Belligundu, and Peng is incorporated. Fan further discloses a system, wherein the encoder component employs pairwise comparisons based on a graph attention algorithm for the re-encoded knowledge graph and the one or more other knowledge graphs:
“In order to obtain sufficient representative high-level node features, structure encoder firstly transforms the original observed node attribute X into the low-dimentional latent representation
Z
~
V
” (Fan, page 2, right column, paragraph 2)
“Given the transformed node embedding
Z
~
V
, a graph attention layer [16] is then employed to aggregate the representation from neighbor nodes, by performing a shared (pairwise) attentional mechanism on the nodes:
PNG
media_image10.png
40
532
media_image10.png
Greyscale
… its final embedding
Z
~
i
V
can be obtained by weighted sum based on the learned importance weights” (Fan, page 2, right column, paragraph 3).
“Finally, structure decoder takes the final node embeddings
Z
V
as inputs to decode them for reconstruction of the original network structure (re-encoded knowledge graph)” (Fan, page 2, right column, paragraph 3)
Fan relates to graph anomaly detection and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Liang, Fan, Belligundu, and Peng to re-encode graphs with an attention mechanism, as disclosed by Fan. Attention mechanisms can capture complex interactions between nodes, a feature important for anomaly detection, as well as reduce computational overhead caused by shallow learning mechanisms. Fan’s method overcomes these problems, and outperforms contemporary state-of-the-art methods. See Fan, page 1, Abstract; page 1, right column, paragraph 2; and page 4, right column, paragraph 3.
Regarding claim 7, the rejection of claim 6 in view of Liang, Fan, Belligundu, and Peng is incorporated. Fan, in combination with Liang, further discloses a system, wherein if an abnormal score of a node of the document is significantly higher than one or more other nodes of the one or more other documents, the document is abnormal: (Fan) “the anomaly score
S
V
i
(abnormal score) of node
V
i
is defined as the reconstruction error from both network structure and node attribute perspective
PNG
media_image8.png
35
565
media_image8.png
Greyscale
Based on the measured anomaly scores, the threshold
λ
in Eq. 1 can be determined according to distribution of scores, e.g. the nodes of top-k scores are classified as anomalous nodes” (Fan, page 3, right column, paragraph 3). If the score of a node in one graph is higher than the threshold value, while all nodes of another graph fall below the threshold, the former graph will be categorized as abnormal, while the latter won’t.
Fan relates to anomaly detection in graphs and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Liang, Fan, Belligundu, and Peng to detect graph abnormalities with Fan’s method. Fan’s method enables the detection of nodes (and by extension, graphs) that deviate significantly from a majority of reference nodes / graphs, while more effectively capturing useful cross-modality interactions between graph structures and node attributes often neglected in similar art. Fan’s method significantly outperforms many contemporary state-of-the-art methods on multiple datasets. See Fan, page 1, Abstract; and page 4, left column, paragraph 3.
Regarding claim 8, Liang discloses [a] computer implemented method of abnormal document self-discovery, comprising:
extracting, using a processor coupled to memory, page elements from a document based on a page layout analysis: “Once there is a model in the model base, a new document image (represented by a PHY-XML file after segmentation and OCR) can be first converted into a candidate layout graph, then matched to the model in order to assign logical labels to each block, resulting in a LOG-XML file. The PHY-XML and LOG-XML files can be used by downstream applications. If there is any error, the user verifies the PHY-XML and LOG-XML files. The verification results, along with the model, are handed over to the model learning module to improve the model.” (Liang, page 225, paragraph 2)
generating, using a processor coupled to memory, a knowledge graph with vectors corresponding with nodes associated with the extracted page elements and their relative layout positions representative of a page layout of a document
generating, using a processor coupled to memory, a knowledge graph with vectors corresponding with nodes representative of a page layout of a document:
PNG
media_image1.png
400
772
media_image1.png
Greyscale
”System overview” (Liang, page 225, Fig. 1)
“Fig. 1 shows an overview of our document analysis system. First, document images are processed by a segmentation-and-OCR engine” (Liang, page 225, paragraph 1). It should be understood for following limitations that all functions performed by this system are being executed by some component(s) of this system.
PNG
media_image2.png
200
400
media_image2.png
Greyscale
PNG
media_image3.png
200
400
media_image3.png
Greyscale
“Original document page and converted HTML result” (Liang, page 227, Fig. 3)
PNG
media_image4.png
198
708
media_image4.png
Greyscale
”Example layout and layout graphs” (Liang, page 227, Fig. 4)
“A layout graph (knowledge graph) is a fully connected attributed relational graph. Each node corresponds to a segmented block on a page. The attributes (vector) of a node are the position (relative layout position) and size of the bounding box, and the normalized font size (small, middle, or large as compared to the average font size over the whole page). An edge between a pair of nodes reflects the spatial relationship (layout relationships) between two corresponding blocks in the image.” (Liang, page 227, paragraph 1)
evaluating, using the processor, pairwise similarities between the vectors and explicitly establishing edges in the knowledge graph when a similarity criterion indicative of a relationship between corresponding page elements is satisfied:
“An edge between a pair of nodes reflects the spatial relationship between two corresponding blocks in the image” (Liang, page 227, paragraph 1)
“The attributes of edge AB in the left graph are shown in Fig. 5” (Liang, page 228, paragraph 1)
PNG
media_image5.png
373
767
media_image5.png
Greyscale
”Example edge attributes” (Liang, page 228, Fig. 5). To store these attributes, the block corresponding to node A inherently must be compared to the block corresponding to node B.
re-encoding, using the processor, the knowledge graph to produce a transformed representation that maintains layout-dependent relationships among the nodes:
“Our approach is a two-step approximate solution that aims at sub-optimal N-1 match. First, we search for the best 1-1 match from U to M, where U is the candidate graph, and M is the model graph” (Liang, page 229, paragraph 7). The candidate knowledge graph is re-encoded into the model knowledge graph.
“We need a metric to measure which mapping is the best. For a given mapping, an intermediate layout graph, T, is first constructed based on U such that the mapping between T and M is 1-1. Then a cost is computed for the 1-1 mapping and defined as the quality measurement of the mapping between U and M. The best match is the one with minimal cost.” (Liang, page 228, paragraph 4)
“For a pair of mapped nodes, the cost is defined as the sum of differences between corresponding attributes (positional relationships), weighted by the weight factors in model node.” (Liang, page 228, paragraph 5); “A cost is similarly defined for a pair of edges (spatial relationships)” (Liang, page 229, paragraph 2); “The graph match cost is the sum of all node pair costs and edge pair costs” (Liang, page 229, paragraph 3). Graph match cost is minimized to encode the final model graph, in doing so, preserving layout-dependent relationships between nodes and edges by minimizing their differences.
While Liang fails to disclose the further limitations of the claim, Peng, in combination with Liang, discloses a system comprising evaluating, using the processor, pairwise similarities between the vectors and explicitly establishing edges in the knowledge graph when a similarity criterion indicative of a relationship between corresponding page elements is satisfied: (Peng)
“We first give the formal definition of anomaly detection on attributed networks : suppose U = {u1, u2, … , un} indicates a set of n instances (nodes), each instance is affiliated with a set of d-dimensional attributes F = {f1, f2, … , fd} (vectors) … these instances are interconnected with each other to form a network, and we use the adjacency matrix
A
∈
R
n
×
n
to describe the link relationships (edge[s]) between them, where A(i, j) = 1 indicates ui and uj is connected with each other” (Peng, page 3514, right column, paragraph 2).
“Based on Homophily, we require that if two instances are connected in the network, their attribute patterns in the residual matrix
R
~
ought to be similar after attribute reconstruction. Formally, we achieve network structure modeling by minimizing
PNG
media_image6.png
41
564
media_image6.png
Greyscale
” (Peng, page 3515, left column, paragraph 1). By minimizing this expression, a degree of similarity between the attributes of connected nodes in the graph is enforced. This enforced degree of similarity can be described as a similarity criterion.
Peng relates to anomaly detection in knowledge graphs and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Liang to enforce attribute similarity between graph-connected nodes, as disclosed by Peng. The vast majority of graph anomaly detection work relies on Homophily as an assumption, requiring this strong correlation between graph connections and attribute similarity. Non-Homophilic structures can lead to adverse effects on anomaly detection. See Peng, page 3513, right column, paragraph 2.
While Liang fails to disclose the further limitations of the claim, Fan, in combination with Liang, discloses [a] computer implemented method of abnormal document self-discovery, comprising: … comparing, using the processor, graph-level relational characteristics a structure
of the knowledge graph with one or more other knowledge graphs corresponding to one or more
other documents to detect deviations from a learned or observed document layout pattern to
determine if the document is abnormal:
(Fan) “Given an attributed network G = {V, E, X} (knowledge graph), our goal is to detect the nodes that are rare and differ significantly from the majority of the reference nodes in terms both the structure and attribute (graph-level relational characteristics) information of the nodes. More formally, we aim to learn a score function
f
:
V
i
→
y
i
∈
R
, to classify sample
x
i
based on the threshold
λ
:
PNG
media_image7.png
78
544
media_image7.png
Greyscale
where
y
i
denotes the label of sample
x
i
, with 0 being the normal class and 1 the anomalous (abnormal / deviat[e]) class“ (Fan, page 2, left column, paragraph 2)
(Fan) “the anomaly score
S
V
i
of node
V
i
is defined as the reconstruction error from both network structure and node attribute perspective
PNG
media_image8.png
35
565
media_image8.png
Greyscale
Based on the measured anomaly scores, the threshold
λ
in Eq. 1 can be determined according to distribution of scores, e.g. the nodes of top-k scores are classified as anomalous nodes” (Fan, page 3, right column, paragraph 3). The abnormality threshold can be determined based on the distribution of scores measured from the graph(s).
“Three commonly used real-world datasets [14] are used in this paper to evaluate the proposed method, including Blog- Catalog, Flickr, and ACM” (Fan, page 4, left column, paragraph 1). This method can be executed on multiple graphs (one or more other knowledge graphs).
Liang and Fan relate to analysis of knowledge graphs and are analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Liang and Peng to detect graph abnormalities with Fan’s method. Fan’s method enables the detection of nodes (and by extension, graphs) that deviate significantly from a majority of reference nodes / graphs, while more effectively capturing useful cross-modality interactions between graph structures and node attributes often neglected in similar art. Fan’s method significantly outperforms many contemporary state-of-the-art methods on multiple datasets. See Fan, page 1, Abstract; and page 4, left column, paragraph 3.
While Fan fails to disclose the further limitations of the claim, Belligundu discloses [a] computer implemented method of abnormal document self-discovery, comprising: … using a processor coupled to memory: “According to a fourth aspect, the object of the disclosure is achieved by a computer program product (non-transitory computer readable storage medium (memory) having instructions, which when executed by a processor, perform actions) for detecting anomalies associated with a plurality of data objects of a technical installation” (Belligundu, [0017]).
Belligundu relates to anomaly detection of knowledge graphs and is analogous to the claimed invention. The combination of Liang, Peng, and Fan teaches a system for detecting anomalies in document knowledge graphs. The claimed invention improves upon this method by storing it in the form of instructions on computer hardware. Belligundu teaches hardware for storing and executing methods for detecting anomalies in knowledge graphs, applicable to the combination of Liang, Peng, and Fan. A person of ordinary skill in the art would have recognized that storing the system of Liang, Peng, and Fan as computer instructions on Belligundu’s hardware would lead to the predictable result of the method being executable by a computing system, and would improve the known device by allowing it to be performed with real data (MPEP 2143 I. (D) Applying a known technique to a known device (method, or product) ready for improvement to yield predictable results).
The analysis of claims 9-10 & 12-14 mirrors that of claims 2-3 & 5-7, with the exception that claims 9-10 & 12-14 are directed to extracting page elements from a document. Liang discloses the method of extracting page elements from a document, as discussed regarding claim 8. Thus, claims 9-10 & 12-14 are rejected under the same rationales used for claims 2-3 & 5-7, respectively.
Regarding claim 11, the rejection of claim 10 in view of Liang, Peng, Fan, and Belligundu is incorporated. Peng, in combination with Liang and Belligundu, further discloses a method, comprising: determining, using the processor, whether the edge is present between the vectors by determining whether a similarity between two of the vectors is greater than a predetermined edge threshold: (Peng)
“We first give the formal definition of anomaly detection on attributed networks : suppose U = {u1, u2, … , un} indicates a set of n instances (nodes), each instance is affiliated with a set of d-dimensional attributes F = {f1, f2, … , fd} (vectors) … these instances are interconnected with each other to form a network, and we use the adjacency matrix
A
∈
R
n
×
n
to describe the link relationships (edge[s]) between them, where A(i, j) = 1 indicates ui and uj is connected with each other” (Peng, page 3514, right column, paragraph 2).
“Based on Homophily, we require that if two instances are connected in the network, their attribute patterns in the residual matrix
R
~
ought to be similar after attribute reconstruction. Formally, we achieve network structure modeling by minimizing
PNG
media_image6.png
41
564
media_image6.png
Greyscale
” (Peng, page 3515, left column, paragraph 1). By minimizing this expression, a degree of similarity between the attributes of connected nodes in the graph is enforced. This enforced degree of similarity can be described as a predetermined edge threshold.
Peng relates to anomaly detection in knowledge graphs and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Liang, Peng, Fan, and Belligundu to enforce attribute similarity between graph-connected nodes, as disclosed by Peng. The vast majority of graph anomaly detection work relies on Homophily as an assumption, requiring this strong correlation between graph connections and attribute similarity. Non-Homophilic structures can lead to adverse effects on anomaly detection. See Peng, page 3513, right column, paragraph 2.
Regarding claim 15, Liang discloses a system, able to:
obtain a digital document and derive page element representations that encode positional relationships within a page layout of the document: “Once there is a model in the model base, a new document image (represented by a PHY-XML file after segmentation and OCR) can be first converted into a candidate layout graph, then matched to the model in order to assign logical labels to each block, resulting in a LOG-XML file. The PHY-XML and LOG-XML files can be used by downstream applications. If there is any error, the user verifies the PHY-XML and LOG-XML files. The verification results, along with the model, are handed over to the model learning module to improve the model.” (Liang, page 225, paragraph 2)
generate, using the processor coupled to memory, a knowledge graph with vectors corresponding with nodes mapped to the page element representations and the encoded positional relationships representative of a page layout of a document:
PNG
media_image1.png
400
772
media_image1.png
Greyscale
”System overview” (Liang, page 225, Fig. 1)
“Fig. 1 shows an overview of our document analysis system. First, document images are processed by a segmentation-and-OCR engine (object detection component)” (Liang, page 225, paragraph 1). It should be understood for following limitations that all functions performed by this system are being executed by some component(s) of this system.
PNG
media_image2.png
200
400
media_image2.png
Greyscale
PNG
media_image3.png
200
400
media_image3.png
Greyscale
“Original document page and converted HTML result” (Liang, page 227, Fig. 3)
PNG
media_image4.png
198
708
media_image4.png
Greyscale
”Example layout and layout graphs” (Liang, page 227, Fig. 4)
“A layout graph (knowledge graph) is a fully connected attributed relational graph. Each node corresponds to a segmented block on a page. The attributes (vector) of a node are the position (page layout position) and size of the bounding box, and the normalized font size (small, middle, or large as compared to the average font size over the whole page). An edge between a pair of nodes reflects the spatial relationship between two corresponding blocks in the image.” (Liang, page 227, paragraph 1)
determine edge connectivity within the knowledge graph by applying a similarity evaluation to the vectors to selectively form edges indicative of relationships between corresponding page elements:
“An edge between a pair of nodes reflects the spatial relationship between two corresponding blocks in the image” (Liang, page 227, paragraph 1)
“The attributes of edge AB in the left graph are shown in Fig. 5” (Liang, page 228, paragraph 1)
PNG
media_image5.png
373
767
media_image5.png
Greyscale
”Example edge attributes” (Liang, page 228, Fig. 5). To store these attributes, the block corresponding to node A inherently must be compared to the block corresponding to node B.
re-encode, using the processor, the knowledge graph into an alternative graph representation that retains layout-derived relational information: :
“Our approach is a two-step approximate solution that aims at sub-optimal N-1 match. First, we search for the best 1-1 match from U to M, where U is the candidate graph, and M is the model graph (alternative graph representation)” (Liang, page 229, paragraph 7). The candidate knowledge graph is re-encoded into the model knowledge graph.
“We need a metric to measure which mapping is the best. For a given mapping, an intermediate layout graph, T, is first constructed based on U such that the mapping between T and M is 1-1. Then a cost is computed for the 1-1 mapping and defined as the quality measurement of the mapping between U and M. The best match is the one with minimal cost.” (Liang, page 228, paragraph 4)
“For a pair of mapped nodes, the cost is defined as the sum of differences between corresponding attributes (positional relational information), weighted by the weight factors in model node.” (Liang, page 228, paragraph 5); “A cost is similarly defined for a pair of edges (spatial relational information)” (Liang, page 229, paragraph 2); “The graph match cost is the sum of all node pair costs and edge pair costs” (Liang, page 229, paragraph 3). Graph match cost is minimized to encode the final model graph, in doing so, preserving layout-derived relational information between nodes and edges by minimizing their differences.
While Liang fails to disclose the further limitations of the claim, Peng, in combination with Liang, discloses a method to determine edge connectivity within the knowledge graph by applying a similarity evaluation to the vectors to selectively form edges indicative of relationships between
corresponding page elements: (Peng)
“We first give the formal definition of anomaly detection on attributed networks : suppose U = {u1, u2, … , un} indicates a set of n instances (nodes), each instance is affiliated with a set of d-dimensional attributes F = {f1, f2, … , fd} (vectors) … these instances are interconnected with each other to form a network, and we use the adjacency matrix
A
∈
R
n
×
n
to describe the link relationships (edge[s]) between them, where A(i, j) = 1 indicates ui and uj is connected with each other” (Peng, page 3514, right column, paragraph 2).
“Based on Homophily, we require that if two instances are connected in the network, their attribute patterns in the residual matrix
R
~
ought to be similar after attribute reconstruction. Formally, we achieve network structure modeling by minimizing
PNG
media_image6.png
41
564
media_image6.png
Greyscale
” (Peng, page 3515, left column, paragraph 1). By minimizing this expression, a degree of similarity between the attributes of connected nodes in the graph is enforced. This enforced degree of similarity can be described as a similarity evaluation.
Peng relates to anomaly detection in knowledge graphs and is analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Liang to enforce attribute similarity between graph-connected nodes, as disclosed by Peng. The vast majority of graph anomaly detection work relies on Homophily as an assumption, requiring this strong correlation between graph connections and attribute similarity. Non-Homophilic structures can lead to adverse effects on anomaly detection. See Peng, page 3513, right column, paragraph 2.
While Peng fails to disclose the further limitations of the claim, Fan, in combination with Liang, discloses compare, using the processor, relational patterns of the alternative graph representation with one or more other knowledge graphs corresponding to one or more other documents to identify anomalous document structures to determine if the document is abnormal:
(Fan) “Given an attributed network G = {V, E, X} (knowledge graph), our goal is to detect the nodes that are rare and differ significantly from the majority of the reference nodes in terms both the structure and attribute (relational patterns) information of the nodes. More formally, we aim to learn a score function
f
:
V
i
→
y
i
∈
R
, to classify sample
x
i
based on the threshold
λ
:
PNG
media_image7.png
78
544
media_image7.png
Greyscale
where
y
i
denotes the label of sample
x
i
, with 0 being the normal class and 1 the anomalous (anomalous) class“ (Fan, page 2, left column, paragraph 2)
(Fan) “the anomaly score
S
V
i
of node
V
i
is defined as the reconstruction error from both network structure and node attribute perspective
PNG
media_image8.png
35
565
media_image8.png
Greyscale
Based on the measured anomaly scores, the threshold
λ
in Eq. 1 can be determined according to distribution of scores, e.g. the nodes of top-k scores are classified as anomalous nodes” (Fan, page 3, right column, paragraph 3). The abnormality threshold can be determined based on the distribution of scores measured from the graph(s).
“Three commonly used real-world datasets [14] are used in this paper to evaluate the proposed method, including Blog- Catalog, Flickr, and ACM” (Fan, page 4, left column, paragraph 1). This method can be executed on multiple graphs (one or more other knowledge graphs).
Fan relates to analysis of knowledge graphs and are analogous to the claimed invention. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Liang and Peng to detect graph abnormalities with Fan’s method. Fan’s method enables the detection of nodes (and by extension, graphs) that deviate significantly from a majority of reference nodes / graphs, while more effectively capturing useful cross-modality interactions between graph structures and node attributes often neglected in similar art. Fan’s method significantly outperforms many contemporary state-of-the-art methods on multiple datasets. See Fan, page 1, Abstract; and page 4, left column, paragraph 3.
While Fan fails to disclose the further limitations of the claim, Belligundu discloses [a] computer program product abnormal document self-discovery, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to: “According to a fourth aspect, the object of the disclosure is achieved by a computer program product (non-transitory computer readable storage medium having instructions, which when executed by a processor, perform actions) for detecting anomalies associated with a plurality of data objects of a technical installation” (Belligundu, [0017]).
Belligundu relates to anomaly detection of knowledge graphs and is analogous to the claimed invention. The combination of Liang and Fan teaches a system for detecting anomalies in document knowledge graphs. The claimed invention improves upon this method by storing it in the form of instructions on computer hardware. Belligundu teaches hardware for storing and executing methods for detecting anomalies in knowledge graphs, applicable to the combination of Liang, Peng, and Fan. A person of ordinary skill in the art would have recognized that storing the system of Liang and Fan as computer instructions on Belligundu’s hardware would lead to the predictable result of the method being executable by a computing system, and would improve the known device by allowing it to be performed with real data (MPEP 2143 I. (D) Applying a known technique to a known device (method, or product) ready for improvement to yield predictable results).
The analysis of claims 16-20 mirrors that of claims 2-6, with the exception that claims 16-20 are directed to obtaining and deriving element representations from a digital document and executing the methods of claims 2-6 on generic computer hardware. Liang discloses the method of obtaining and deriving element representations, and generic hardware for executing the methods is taught by Belligundu, as discussed regarding claim 15. Thus, claims 16-20 are rejected under the same rationales used for claims 2-6, respectively.
Response to Arguments
The following responses address arguments and remarks made in the instant remarks dated 12/29/2025.
Objections
In light of the instant amendments and remarks, previous objections to the drawings have been withdrawn. The Examiner notes that this withdrawal is in light of an argument from the Applicant, suggesting that fine details of certain figures are unneeded for a clear understanding of the claimed invention: “Figs 2A, 2B and 3 have been enlarged for clarity. It is noted that all details (e.g., document page sections) necessary for a clear understanding of the invention are shown clearly. Incidental or background matter (including text appearing in screenshots, GUIs, documents, or images used merely for context) need not be fully legible. See MPEP § 608.02(d).” This argument is present on page 2 of the Request for Reconsideration-After Non-Final Rejection received 12/29/2025.
In light of the instant amendments, previous objections to the claims have been withdrawn.
Claim Interpretation
In light of the instant amendments, claim 11 is no longer considered to have any contingent limitations.
On pages 12-15 of the instant remarks, the Applicant argues that claims interpreted under 35 U.S.C. 112(f) have sufficient structure disclosed:
“The Office asserts in Section IV of the Office Action that certain claim limitations,
including limitations of independent claim 1 and dependent claims 2-6, are being interpreted
under 35 U.S.C. §112(±) (pre-AIA §112, sixth paragraph) on the basis that they purportedly
recite functional language using generic placeholders without sufficient structure. Applicant
respectfully traverses this interpretation. The Office's application of§ 112(±) is legally and
factually improper because the identified claim limitations recite sufficient structure as
understood by one of ordinary skill in the art, particularly in the fields of computer vision,
document layout analysis, and graph-based machine learning.
As an initial matter, none of the cited claim limitations uses the term "means" or "step,"
and therefore a strong rebuttable presumption applies that§ 112(±) does not govern their
interpretation. See Williamson v. Citrix Online, LLC, 792 F.3d 1339, 1348-49 (Fed. Cir. 2015).
That presumption may be overcome only if the claim language, read in light of the specification,
fails to recite sufficient structure for performing the claimed function. That is not the case here.
With respect to independent claim 1, the limitations reciting an "object detection
component," an "evaluation component," an "encoder component," and a "comparison
component" are not mere nonce terms. In the context of the claims and the specification, each of
these components denotes a class of well-understood structures and algorithmic processes
recognized by persons of ordinary skill in the art. The claims do not merely recite results; rather,
they specify concrete operations performed on defined data structures, including page elements,
vectors, knowledge graphs, nodes, edges, and graph topology.
For example, the "object detection component" of claim 1 is expressly defined as
identifying page elements based on page layout analysis and generating a knowledge graph with
nodes associated with identified page elements and their spatial layout relationships. As
described in the specification and illustrated in FIGS. 2A and 2B, this component performs
layout segmentation and object detection on document pages-operations that are well-known,
structurally defined techniques in document image analysis and computer vision. A person of
ordinary skill in the art would readily understand this component to correspond to specific
algorithmic structures, such as OCR-based layout analyzers, document segmentation pipelines,
and page-element classifiers.
Similarly, the "evaluation component" recited in claim 1 does not merely state a desired
outcome. It explicitly determines whether an edge is present between vectors based on a
similarity exceeding a defined threshold. The specification discloses detailed structure for
performing this function, including vector similarity computation, threshold comparison, and
explicit edge formation, as illustrated for example in FIG. 4A (similarity judgment model and
threshold). These disclosures provide algorithmic structure sufficient to perform the recited
function, and therefore preclude interpretation under§ 112(±).
The "encoder component" recited in claim 1 likewise conveys sufficient structure. In
modern machine learning and graph analytics, an "encoder" is a recognized structural class of
components that transform input representations into alternative representations according to
defined architectures. The specification expressly discloses re-encoding of knowledge graphs
using graph-based encoding techniques, including graph attention mechanisms, as shown for
example in FIG. 5. This disclosure provides concrete algorithmic structure, and the term
"encoder component" would be readily understood by a person of ordinary skill in the art to
denote such structure.
The "comparison component" of claim 1 further reinforces that the claims recite
structural, not purely functional, limitations. This component compares the structural topology of
knowledge graphs corresponding to different documents to identify structural divergence
indicative of abnormality. Graph topology, nodes, edges, and relational patterns are concrete data
structures in graph theory and graph neural networks. The specification provides detailed
disclosure of how such comparisons are performed, including anomaly scoring and graph-level
divergence analysis (see, e.g., FIGS. 6A-6D and FIG. 7). Accordingly, this limitation also recites
sufficient structure and is not subject to §112(±).”
Regarding the argument that claim limitations not using “means” or “step” entail a rebuttal presumption, MPEP 2181(I) notes The presumption that 35 U.S.C. 112(f) does not apply to a claim limitation that does not use the term "means" is overcome when "the claim term fails to 'recite sufficiently definite structure' or else recites 'function without reciting sufficient structure for performing that function.'" Williamson, 792 F.3d at 1349, 115 USPQ2d at 1111 (Fed. Cir. 2015) (en banc) (quoting Watts v. XL Systems, Inc., 232 F.3d 877, 880, 56 USPQ2d 1836, 1838 (Fed. Cir. 2000); see also Personalized Media Communications, LLC v. International Trade Commission, 161 F. 3d 696, 704, 48 USPQ2d 1880, 1887 (Fed. Cir. 1998) … Instead of using "means" in such cases, a substitute term acts as a generic placeholder for the term "means" and would not be recognized by one of ordinary skill in the art as being sufficiently definite structure for performing the claimed function.
“an object detection component”, “an evaluation component”, “an encoder component”, and “a comparison component” would not be recognized by one of ordinary skill in the art as having any sort of sufficiently definite structure, and thus these terms act as generic placeholders for “means” and entail the interpretation of the relevant claims under 35 U.S.C. 112(f). For example, an “object detection component” could entail many different families of algorithms (e.g., document scanning, OCR, optical word recognition, intelligent character recognition, intelligent word recognition), data structures (e.g., RNNs, CNNs, other supervised classifiers, unsupervised classifiers) and hardware configurations (e.g., a computer, a phone, a camera, a scanner). All the other generic terms are similarly broad.
For limitations interpreted under 35 U.S.C. 112(f) using means-plus-function language, the structure of the “means” or the equivalent generic placeholder substitute must be disclosed in the specification itself in a way that one skilled in the art will understand what structure will perform the recited function (MPEP 2181 (II.) A.). As noted in the claim interpretation section of the instant office action, no sufficient structure is found to be provided for any of these generic placeholders in the instant specification.
The Applicant alleges that sufficient structure is provided for each of these placeholders in the instant specification, but fails to actually cite where in the specification these details are provided. The cited figures merely illustrate high-level examples of each function being performed, but do not disclose further details of structure. Figure 4A, for example, shows that the evaluation component can comprise a similarity judgment model that measures similarity on vector representations, but the “similarity judgment model” has no specific structure or algorithm defined either in the specification or the drawings. This figure merely illustrates high-level inputs and outputs of a generic model.
Regarding the Applicant’s assertion that the specification discloses adequate structure for the claims, as discussed in the previous interview, if the Applicant believes the generic placeholders have sufficient structural detail within the specification, the Examiner suggests pointing to the specific language providing it in further remarks.
On page 14 of the instant remarks, the Applicant argues that the claims provide sufficient structural detail for performed functions:
“The Office's application of§ 112(±) to dependent claims 2-6 is particularly unwarranted.
Each of these dependent claims further narrows the corresponding component by reciting
specific algorithmic techniques, such as layout analysis based on object detection, multimodal
embedding, threshold-based similarity determination, and graph attention algorithms. These
limitations add structural detail and cannot reasonably be characterized as lacking sufficient
structure. A dependent claim that introduces additional algorithmic specificity cannot
simultaneously be treated as invoking § 112(±).”
Regarding the Applicant’s arguments above, the Examiner respectfully disagrees. No specific structure is disclosed by any of the claim limitations with generic placeholders. For example, in claim 1, “an object detection component” generically “identifies page elements of the document based on a page layout analysis” with no specific algorithm or structure given. The object detection component further “generates a knowledge graph with vectors corresponding to nodes associated with respective identified page elements and their spatial layout relationships”, with no structure and no specific graph generation algorithm specified beyond generically associating nodes with vectors related to page elements in an unspecified manner. Similar arguments are applicable to all the generic placeholders. If the Applicant believes the generic placeholders have sufficient structural specification by the claim language, the Examiner suggests pointing to the specific claim language providing it in further remarks.
On page 14 of the instant remarks, the Applicant argues that the office action itself implies sufficient structural detail is present in the specification:
“Notably, the Office's own statement that the cited limitations are "interpreted to cover the
corresponding structure described in the specification" implicitly acknowledges that the
specification discloses identifiable structure for performing the claimed functions. This
acknowledgment is inconsistent with a finding under prong (C) of the §112(±) analysis, which
requires an absence of sufficient structure. Where the specification provides algorithmic and
architectural disclosure sufficient for a person of ordinary skill in the art to implement the
claimed functions, §112(±) does not apply.”
Regarding the argument that the office action implicitly acknowledges the specification identifying structure for performing the claimed functions due to its recitation of “Because this/these claim limitation(s) is/are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, it/they is/are being interpreted to cover the corresponding structure described in the specification as performing the claimed function, and equivalents thereof”, the Examiner notes that this statement merely conveys that limitations interpreted under 35 U.S.C. 112(f) are considered to encompass the corresponding structure(s) of means, steps, or generic placeholders as described in the specification. It does not allege that the instant specification actually contains this information. For limitations interpreted under 35 U.S.C. 112(f) using means-plus-function language, the structure of the “means” or the equivalent generic placeholder substitute must be disclosed in the specification itself in a way that one skilled in the art will understand what structure will perform the recited function (MPEP 2181 (II.) A.). Additionally, for a computer-implemented means-plus-function limitation interpreted under 35 U.S.C. 112(f), the specification must disclose an algorithm for performing the claimed specific computer function (MPEP 2181 (II.) A.). Failure to adequately disclose either the structure or algorithm in sufficient detail in the specification for a computer-implemented means-plus-function limitation renders the claim indefinite under 35 U.S.C. 112(b).
On pages 14-15 of the instant remarks, the Applicant argues that Belligundu’s similar functionalities imply definite structure for components of the instant application:
“The assertion that certain limitations such as "object detection component," "evaluation
component," or "encoder component" lack sufficient structure and thus fall under 35 U.S.C.
§ 112(±) is not supported when viewed in light of the specification and the incorporated teachings
of Belligundu. That reference, like the present disclosure, describes components such as
encoders, attention layers, and anomaly evaluators in the context of neural networks and graph-
based models. These components are not merely black boxes; they are commonly recognized by
skilled artisans as algorithmically defined structures. In the current application, the specification
and Figures 3 and 5 explicitly disclose architectures including multimodal vector encoders (see
FIG. 3) and graph attention mechanisms (see FIG. 5), which align with the types of structures
utilized in Belligundu. This correspondence reinforces that the claimed components are not
generic placeholders but rather denote classes of well-understood algorithmic modules-akin to
those acknowledged in the art. Consequently, the use of the term "component" in the claims is
not indicative of§ 112(±) invocation but instead identifies definite structure, supported by both
the intrinsic specification and the state of the art exemplified by Belligundu.”
Regarding the argument that Belligundu is analogous to the instant application, the Examiner notes that Belligundu does not recite means, steps, or equivalent generic placeholders that would entail the interpretation of any claim limitations under 35 U.S.C. 112(f), and thus its subject matter and prosecution are completely irrelevant to interpretation of the instant application’s claim language under 112(f).
112 Rejections
In light of the instant amendments, claims 1-20 are newly rejected under 35 U.S.C. 112(a) for containing new matter not supported by the instant specification.
On pages 15-17 of the instant remarks, the Applicant argues that the claims are fully supported by the specification and thus rejections under 35 U.S.C. 112(a) are improper:
“V. Reiection of Claims 1-7 under 35 U.S.C. § 112(a) or 35 U.S.C. § 112 (pre-AIA ) first
paragraph
Claims 1-7 stand rejected under 35 U.S.C. § l 12(a) or 35 U.S.C. § 112 (pre-AIA ) first
paragraph, as allegedly failing to comply with the written description requirement.
The Office has rejected claims 1-7 under 35 U.S.C. §l 12(a) (pre-AIA §112, first
paragraph), alleging that the claims fail to satisfy the written description requirement. Assignee's
Representative respectfully traverses this rejection. When the claims are properly construed in
light of the specification as filed, the application clearly demonstrates that the inventors were in
possession of the subject matter of claims 1-7 at the time of filing.
The written description requirement does not demand ipsissimis verbis support or a
verbatim recitation of claim language in the specification. Rather, the test is whether the
disclosure, viewed as a whole, reasonably conveys to a person of ordinary skill in the art that the
inventor had possession of the claimed subject matter. Ariad Pharm., Inc. v. Eli Lilly & Co., 598
F.3d 1336, 1351 (Fed. Cir. 2010) (en bane). Functional claim language is permissible so long as
the specification provides sufficient description of the structure, process, or algorithms that
15
18/051,570 P202201555US01/IBMP1015US
perform the recited functions. Id; Capon v. Eshhar, 418 F.3d 1349, 1358 (Fed. Cir. 2005).
Here, claims 1-7 are fully supported by the specification's detailed disclosure of a system for
abnormal document self-discovery based on document layout analysis, multimodal embedding,
knowledge graph construction, edge determination, graph re-encoding, and structural comparison
across documents. The specification repeatedly and consistently describes each of these
elements, both individually and in combination, and further illustrates them through detailed
figures and flow diagrams.
With respect to claim 1, the specification expressly discloses an object detection
component that identifies page elements based on page layout analysis and generates a
knowledge graph in which nodes correspond to page elements and their spatial relationships.
This disclosure is described in detail in the discussion of layout analysis and object detection,
and is illustrated, for example, in FIGS. 2A and 2B, which show segmentation of a document
page into page elements. The specification further explains how these page elements are encoded
and represented as nodes within a knowledge graph, thereby providing clear written description
support for the claimed object detection component and graph generation functionality.
The specification also provides explicit support for the evaluation component recited in claim 1.
In particular, the disclosure describes determining whether an edge is present between nodes by
computing similarity between corresponding vectors and comparing the similarity to a threshold.
This process is explained in connection with the similarity judgment model and thresholding
mechanism, and is illustrated, for example, in FIG. 4A. These disclosures demonstrate
possession of the claimed concept of explicit edge determination based on similarity exceeding a
defined threshold.
Further, the encoder component recited in claim 1 is fully supported by the
specification's disclosure of re-encoding the knowledge graph to produce a transformed
representation while preserving relational information derived from page layout relationships.”
Regarding the Applicant’s arguments above, the Examiner respectfully disagrees. As argued previously and noted in the claim interpretations section of the office action, claims 1-7 are found to contain generic placeholders substituting “means”, and are interpreted under 35 U.S.C. 112(f) as using means-plus-function language.
For limitations interpreted under 35 U.S.C. 112(f) using means-plus-function language, the written description under 35 U.S.C. 112(a) must adequately link or associate particular structure, material, or acts to perform the function or it must be clear based on the facts of the application that one skilled in the art would have known what structure, material, or acts disclosed in the specification perform the recited function (MPEP 2163(II) A. (3)). It’s insufficient to merely describe the function in the instant specification.
As argued previously for claim interpretation, the specification fails to provide sufficient support for the structure of generic placeholders interpreted under 35 U.S.C. 112(f), and thus these claims are rejected under 35 U.S.C. 112(a). Additionally, upon further consideration of the instant amendments, the amended independent claims are found to contain new matter not disclosed by the instant specification.
The Applicant alleges that sufficient support is provided for all claims in the instant specification, but fails to cite where in the specification these details are provided. Thus, with no evidence of additional support for these claims, all rejections under 35 U.S.C. 112(a) are maintained.
On pages 18-19 of the instant remarks, the Applicant argues that the claims are clear and definite, and thus satisfy requirements of 35 U.S.C. 112(b):
“VI. Reiection of Claims 1-7 under 35 U.S.C. § 112(b) or 35 U.S.C. § 112(pre-AIA second
paragraph)
Claims 1-7 stand rejected under 35 U.S.C. § l 12(b) or 35 U.S.C. § l 12(pre-AIA second
paragraph), as allegedly being indefinite for failing to particularly point out and distinctly claim
the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA
35 U.S.C. 112, the Assignee), regards as the invention.
The Office has rejected claims 1-7 under 35 U.S.C. §l 12(b) (pre-AIA §112, second
paragraph), alleging that the claims fail to particularly point out and distinctly claim the subject
matter regarded as the invention. Assignee's Representative respectfully traverses this rejection.
When the claims are read under their broadest reasonable interpretation in light of the
specification and from the perspective of a person of ordinary skill in the art, claims 1-7 are
clear, definite, and satisfy the requirements of§ l 12(b ).
The standard for definiteness is whether the claims, read in light of the specification and
the prosecution history, inform those skilled in the art about the scope of the invention with
reasonable certainty. Nautilus, Inc. v. Biosig Instruments, Inc., 572 U.S. 898, 910 (2014).
Absolute precision is not required, nor is the elimination of all potential ambiguity. Rather,
claims are indefinite only if they are "insolubly ambiguous" or fail to provide reasonable notice
of their scope. See Interval Licensing LLC v. AOL, Inc., 766 F.3d 1364, 1370-71 (Fed. Cir.
2014).
Here, the Office has not identified any specific claim term or limitation that is
ambiguous, unclear, or incapable of being understood by a person of ordinary skill in the art.
Instead, the rejection appears to be based on the general presence of functional language in
claims 1-7. Functional claiming, however, is expressly permitted under U.S. patent law and does
not render a claim indefinite so long as the claim language, in view of the specification, provides
clear boundaries of scope. See Halliburton Energy Servs., Inc. v. M-I LLC, 514 F.3d 1244, 1255
(Fed. Cir. 2008).
Each limitation of claim 1 recites concrete elements and operations grounded in wellunderstood
technical concepts in document analysis, computer vision, and graph-based machine
learning. For example, the recitation of "page elements," "page layout analysis," "knowledge
graph," "nodes," "vectors," "edges," and "structural topology" refers to established data
structures and processing constructs that have recognized meaning to skilled artisans. The
18
18/051,570 P202201555US01/IBMP1015US
specification provides detailed explanations of these concepts and illustrates them through
multiple figures ( e.g., FIGS. 2-7), thereby providing clear context for the claim language. When
read in light of this disclosure, the scope of each limitation is reasonably certain.
Terms such as "similarity," "defined threshold," "structural topology," and "structural
divergence indicative of abnormality" likewise do not render the claims indefinite. The Federal
Circuit has repeatedly held that the use of comparative or functional terms does not violate
§ l 12(b) where the specification provides guidance as to how such terms are determined. See
Sonix Tech. Co. v. Publications Int'l, Ltd, 844 F.3d 1370, 1378-79 (Fed. Cir. 2017). In the
present application, the specification explains how similarity between vectors is computed, how
thresholds are applied, and how abnormality is determined through graph-level comparisons and
anomaly scoring. These disclosures provide objective boundaries for the claimed subject matter
and enable a person of ordinary skill in the art to understand the scope of the claims with
reasonable certainty.
Dependent claims 2-7 further narrow and clarify the scope of claim 1 by reciting specific
techniques, including object detection-based layout analysis, multimodal embedding, thresholdbased
edge determination, multimodal comparison, graph attention algorithms, and abnormality
determination based on anomaly scores. Far from introducing ambiguity, these dependent claims
reinforce the definiteness of the claimed invention by providing additional structural and
algorithmic detail.
To the extent the Office's indefiniteness rejection is premised on concerns more properly
associated with written description or means-plus-function interpretation, Assignee's
Representative respectfully submits that such concerns are misplaced in the context of§ l 12(b ).
Definiteness is assessed based on whether the claim language provides clear notice of scope, not
on whether the Office believes the claims could have been drafted more narrowly or with
additional detail. See Nautilus, 572 U.S. at 909-10.
Accordingly, Assignee's Representative respectfully submits that claims 1-7, when read
in light of the specification, particularly point out and distinctly claim the subject matter regarded
as the invention with reasonable certainty. The rejection of claims 1-7 under 35 U.S.C. § l 12(b)
should therefore be withdrawn.”
Regarding the Applicant’s arguments above, the Examiner respectfully disagrees. As argued previously and noted in the claim interpretations section of the office action, claims 1-7 are found to contain generic placeholders substituting “means”, and are interpreted under 35 U.S.C. 112(f) as using means-plus-function language.
For limitations interpreted under 35 U.S.C. 112(f) using means-plus-function language, the structure of the “means” or the equivalent generic placeholder substitute must be disclosed in the specification itself in a way that one skilled in the art will understand what structure will perform the recited function (MPEP 2181 (II.) A.). Additionally, for a computer-implemented means-plus-function limitation interpreted under 35 U.S.C. 112(f), the specification must disclose an algorithm for performing the claimed specific computer function (MPEP 2181 (II.) A.). Failure to adequately disclose either the structure or algorithm in sufficient detail in the specification for a computer-implemented means-plus-function limitation renders the claim indefinite under 35 U.S.C. 112(b).
As argued previously for claim interpretation, the specification fails to provide sufficient support for the structure of generic placeholders interpreted under 35 U.S.C. 112(f), and thus these claims are rejected under 35 U.S.C. 112(b). Thus, rejections under 35 U.S.C. 112(b) are maintained on these grounds.
101 Rejections
On page 20 of the instant remarks, the Applicant argues that the claims do not recite abstract ideas:
“Alice Step One - The Claims Are Not Directed to an Abstract Idea
At Alice step one, the proper inquiry is whether the claims are "directed to" an abstract
idea, not whether they merely involve data, information, or analysis. See Enfish, LLC v.
Microsoft Corp., 822 F.3d 1327, 1335-36 (Fed. Cir. 2016). Claims 1-20 are not directed to the
abstract idea of "comparing documents" or "analyzing information." Rather, they are directed to
a specific technological solution for detecting abnormal documents through page-layoutanchored
knowledge graph construction, explicit edge determination, graph re-encoding, and
inter-document structural comparison.
The claims recite concrete technical elements, including page elements derived from page
layout analysis, vectors, nodes, edges, knowledge graphs, graph topology, and re-encoded graph
representations. These elements are not mental constructs or results-only abstractions; they are
computer-specific data structures that do not exist outside a computing environment. As in
Enfish, the claims focus on an improvement to the way computers represent and process
document layout information, rather than on an abstract idea performed on a generic computer.”
In regards to the Applicant’s arguments above, the Examiner respectfully disagrees that the claimed invention, as amended, recites no mental processes. As stated in MPEP 2106.04(a)(2)(III), The courts do not distinguish between mental processes that are performed entirely in the human mind and mental processes that require a human to use a physical aid (e.g., pen and paper or a slide rule) to perform the claim limitation. See, e.g., Benson, 409 U.S. at 67, 65, 175 USPQ at 674-75, 674 … Nor do the courts distinguish between claims that recite mental processes performed by humans and claims that recite mental processes performed on a computer. As the Federal Circuit has explained, "[c]ourts have examined claims that required the use of a computer and still found that the underlying, patent-ineligible invention could be performed via pen and paper or in a person’s mind." Versata Dev. Group v. SAP Am., Inc., 793 F.3d 1306, 1335, 115 USPQ2d 1681, 1702 (Fed. Cir. 2015). See also Intellectual Ventures I LLC v. Symantec Corp., 838 F.3d 1307, 1318, 120 USPQ2d 1353, 1360 (Fed. Cir. 2016) (‘‘[W]ith the exception of generic computer-implemented steps, there is nothing in the claims themselves that foreclose them from being performed by a human, mentally or with pen and paper.’’); Mortgage Grader, Inc. v. First Choice Loan Servs. Inc., 811 F.3d 1314, 1324, 117 USPQ2d 1693, 1699 (Fed. Cir. 2016) (holding that computer- implemented method for "anonymous loan shopping" was an abstract idea because it could be "performed by humans without a computer").
Claim 1 recites limitations amounting to mental processes executed by generic computer components that are insufficient to render a mentally performable task non-abstract. For example, claim 1 recites the limitation “an encoder component that re-encodes the knowledge graph while preserving relational information derived from the page layout relationships between the nodes”, reciting a mental process of “re-encodes the knowledge graph while preserving relational information derived from the page layout relationships between the nodes” performed by an encoder component, a generic computer component insufficient to render the limitation non-abstract. Similar reasoning applies to substantially similar independent claims 8 and 15.
The Examiner asserts that the claimed invention, as amended, recites mental processes, and maintains its rejection on the basis of the Alice/Mayo tests performed (See 101 rejections).
On page 20 of the instant remarks, the Applicant argues that the claimed invention addresses a technical problem and is thus eligible:
“Further, the claims address a technical problem rooted in computer technology-namely,
the inability of conventional similarity algorithms to reliably detect abnormal documents that
differ only subtly in layout or structure-and provide a technical solution grounded in graph-based
representations derived from document layout. This places the claims squarely within the
category of eligible inventions recognized in DDR Holdings, LLC v. Hotels.com, L.P., 773 F.3d
1245, 1257 (Fed. Cir. 2014).”
In response to the Applicant’s argument that the claimed invention represents an improvement to existing technology or technical field, the Examiner notes that improvements cannot be made through a recited judicial exception. As noted by MPEP 2106.05(a), It is important to note, the judicial exception alone cannot provide the improvement. The improvement can be provided by one or more additional elements. See the discussion of Diamond v. Diehr, 450 U.S. 175, 187 and 191-92, 209 USPQ 1, 10 (1981)) in subsection II, below. In addition, the improvement can be provided by the additional element(s) in combination with the recited judicial exception. See MPEP § 2106.04(d) (discussing Finjan, Inc. v. Blue Coat Sys., Inc., 879 F.3d 1299, 1303-04, 125 USPQ2d 1282, 1285-87 (Fed. Cir. 2018)).
The Applicant is arguing improvement through graph-based representations of document layouts in abnormal document detection, recited in claim 1 as, in part, “identifies page elements of the document based on a page layout analysis and generates a knowledge graph with vectors corresponding with nodes associated with respective identified page elements and their spatial layout relationships representative of a page layout of a document”, which can be performed as a mental process. Mere instruction to execute this mental process with a generic additional element (“an object detection component”) is not representative of an improvement to technology or a technical field. As detailed further in the 101 rejections section, none of the additional elements of the claimed invention are sufficiently representative of the alleged improvement to technology argued by the Applicant.
Thus, no rejections are withdrawn on these grounds. See the 101 rejections section for more detail.
On page 21 of the instant remarks, the Applicant argues that the claimed invention reflects a non-conventional and non-generic inventive concept and thus amounts to significantly more:
“Alice Step Two - The Claims Recite an Inventive Concept
Even assuming, arguendo, that the claims are characterized as involving an abstract idea,
the claims nonetheless satisfy Alice step two because they recite an inventive concept that
amounts to significantly more than any alleged abstract idea.
The claimed invention does not merely apply conventional document analysis or anomaly
detection techniques. Instead, it requires, inter alia:
• identifying page elements through page layout analysis and representing those elements
as nodes in a knowledge graph;
• explicitly determining edge existence between nodes based on similarity thresholds tied
to relationships between page elements;
• re-encoding the knowledge graph while preserving layout-derived relational information;
and
• comparing graph-level structural topology or relational patterns across multiple
documents to detect abnormality.
These steps reflect a non-conventional and non-generic combination of operations that
transform document layout data into a graph-based representation and then use that
representation in a manner not taught or suggested by generic data analysis techniques. The
specification explains these steps in detail and illustrates them in FIGS. 2-7, demonstrating that
the claims are not mere result-oriented statements, but rather recite a specific architecture and
processing pipeline.
The Federal Circuit has consistently held that such specific, technology-based solutions
constitute an inventive concept. SeeMcRO, Inc. v. Bandai Namco Games Am. Inc., 837 F.3d
1299, 1315-16 (Fed. Cir. 2016) (claims that use specific rules to achieve a technological
improvement are patent-eligible); SRI Int'l, Inc. v. Cisco Sys., Inc., 918 F.3d 1368, 1375-76
(Fed. Cir. 2019) (claims directed to a specific technique for improving computer network
security are eligible).
Importantly, the claims do not merely recite "analyze," "compare," or "classify" at a high
level. They require particular data representations (knowledge graphs with layout-derived nodes
and edges) and particular operations on those representations (explicit edge determination, graph
re-encoding, and structural comparison). This is precisely the type of "significantly more" that
distinguishes eligible claims from those found ineligible in cases such as Electric Power Group,
LLC v. Alstom S.A., 830 F.3d 1350 (Fed. Cir. 2016)
No Preemption or Mental Process Concerns
The claims do not risk preempting an abstract idea. They do not cover all ways of
detecting abnormal documents, nor all ways of comparing graphs or documents. Instead, they are
limited to a specific technical approach that relies on page-layout-derived graph structures and
explicit relational processing. Moreover, the claimed steps cannot be practically performed in the
human mind and require computer-implemented data structures and processing, further
supporting eligibility.”
In response to the Applicant’s argument that the claimed invention amounts to significantly more by virtue of providing a specific, technology-based system, the Examiner notes that for the sake of Alice / Mayo analysis, an inventive concept cannot be made through a recited judicial exception. As noted by MPEP 2106.05(I), An inventive concept "cannot be furnished by the unpatentable law of nature (or natural phenomenon or abstract idea) itself." Genetic Techs. Ltd. v. Merial LLC, 818 F.3d 1369, 1376, 118 USPQ2d 1541, 1546 (Fed. Cir. 2016). See also Alice Corp., 573 U.S. at 21-18, 110 USPQ2d at 1981 (citing Mayo, 566 U.S. at 78, 101 USPQ2d at 1968 (after determining that a claim is directed to a judicial exception, "we then ask, ‘[w]hat else is there in the claims before us?") (emphasis added)); RecogniCorp, LLC v. Nintendo Co., 855 F.3d 1322, 1327, 122 USPQ2d 1377 (Fed. Cir. 2017) ("Adding one abstract idea (math) to another abstract idea (encoding and decoding) does not render the claim non-abstract"). Instead, an "inventive concept" is furnished by an element or combination of elements that is recited in the claim in addition to (beyond) the judicial exception, and is sufficient to ensure that the claim as a whole amounts to significantly more than the judicial exception itself. Alice Corp., 573 U.S. at 27-18, 110 USPQ2d at 1981 (citing Mayo, 566 U.S. at 72-73, 101 USPQ2d at 1966).
The Applicant is arguing inventiveness through abstract mental processes. For example, as recited in claim 1, “compares a structural topology structure of the knowledge graph with one or more other knowledge graphs corresponding to one or more other documents to identify structural divergence indicative of abnormality to determine if the document is abnormal” can be performed as a mental process. A “comparison component”, a generic additional element that executes this mental process, does not amount to a non-conventional, non-generic technical element.
While the claimed invention contains several additional elements, they are insufficient to furnish an inventive concept and make the claim as a whole amount to significantly more than its recited judicial exceptions.
Thus, no rejections are withdrawn on these grounds. See the 101 rejections section for more detail.
103 Rejections
On pages 23-24 of the instant remarks, the Applicant argues that Liang doesn’t disclose the claimed invention:
“Liang et al. Does Not Teach or Suggest the Claimed Invention
Liang et al. is directed to logical labeling of document images using layout graph
matching for classification of document components. Liang' s graphs are used to match a
document against predefined templates or classes to assign labels ( e.g., title, text, table). Liang
does not teach or suggest detecting abnormal documents within a population of documents, nor
does Liang disclose graph-level anomaly detection based on structural divergence across
documents.
23
18/051,570 P202201555US01/IBMP1015US
Critically, Liang does not disclose:
(1) explicit determination of edge existence based on similarity thresholds
between multimodal vectors;
(2) re-encoding a knowledge graph while preserving layout-derived relational
information; or
(3) comparing graph-level structure across multiple documents to identify
abnormality.
Liang' s graph matching is deterministic and template-driven, not adaptive, and is used
for labeling, not anomaly detection. Thus, Liang fails to teach the core objectives and
mechanisms of the rejected claims.
…
Liang et al. is expressly limited to graph matching against a predefined model graph for
purposes of logical labeling. Liang neither discloses nor suggests dynamically inferring edge
connectivity based on similarity thresholds, constructing a knowledge graph absent a known
template, or comparing graph-level structural divergence across documents. To the contrary,
Liang presumes that the document conforms to one of several known layout classes and seeks
only to classify page elements within that class. See Liang, Section 3 .1 ("Layout Graph"), which
describes a matching operation against a model graph, not discovery or abnormality detection.”
In response to the Applicant's argument that Liang fails to disclose limitations of the amended claims, the Examiner notes that one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986).
Regarding amended claim 1, Liang discloses generating a knowledge graph with node vectors associated with page elements and spatial layout information based on a document (Liang, page 225, paragraph 1 & page 227, paragraph 1). Liang additionally discloses determining edges between nodes based on page element relationships (Liang, page 227, paragraph 1 & page 228, paragraph 1) and encoding a knowledge graph in a manner that preserves page layout relationships between nodes (Liang, page 228, paragraph 5 & page 229, paragraphs 2-7).
While Liang fails to disclose determining edges between nodes based on a similarity score exceeding a threshold, this deficiency is remedied by Peng, which discloses connecting nodes in a graph if and only if they meet or exceed a similarity threshold (Peng, page 3514, right column, paragraph 2 & page 3515, left column, paragraph 1).
While Liang fails to disclose comparing knowledge graphs to find anomalies, this deficiency is remedied by Fan, which teaches identification of an abnormal graph through structural abnormalities (Fan, page 2, left column, paragraph 2; page 3, right column, paragraph 3; page 4, left column, paragraph 1), which can be applied to the one or more other knowledge graphs in the combination for comparison.
It would have been obvious for one of ordinary skill in the art to determine edges in Liang’s knowledge graph based on similarity metrics, as disclosed by Peng, motivated by Peng’s admission that the vast majority of graph anomaly detection is dependent on homophily between similar nodes (Peng, page 3513, right column, paragraph 2). It would also have been obvious to one of ordinary skill in the art to compare graphs for anomalies, as disclosed by Fan, motivated by Fan’s method significantly outperforming many contemporary anomaly detection methods and effectively capturing interactions often missed by similar anomaly detection methods (Fan, page 1, Abstract & page 4, left column, paragraph 3).
Combined with Belligundu’s generic hardware, amended claim 1 is found to be obvious over the prior art in view of Liang, Peng, Fan, and Belligundu. Claims 19-20 are also found to be obvious over the prior art in view of these rejections. See the 103 rejections section for more detail. Thus, no rejections are withdrawn on these grounds.
On pages 24-26 of the instant remarks, the Applicant argues that Fan does not disclose the claimed invention:
“Fan et al. Does Not Cure the Deficiencies of Liang et al.
Fan et al. (ANOMAL YDAE) is directed to anomaly detection on attributed networks
using a dual autoencoder. Fan operates on pre-existing abstract graphs (e.g., social or network
graphs) and focuses on learning latent representations of nodes and edges to identify anomalous
nodes.
Fan does not disclose or suggest:
(1) constructing a graph from document page layout;
(2) nodes corresponding to page elements derived from layout analysis;
(3) explicit edge formation based on similarity thresholds tied to page-element
relationships; or
(4) comparing graph-level document structures across multiple documents.
Fan presumes the existence of a network and applies anomaly detection within that
network. Fan does not address document images, page layout, document-level abnormality, or
structural divergence among documents. Importantly, Fan does not re-encode graphs in a manner
that preserves page-layout semantics, because Fan's graphs are not layout-derived to begin with.
For example, see section 3 of Fan reproduced below:
3. METHOD
In this section, we introduce the proposed AnomalyDAE
in detail. As shown in Fig. 1, AnomalyDAE is an end-toend
joint representation learning framework that consists of
a structure autoencoder for network structure reconstruction,
and an attribute autoencoder for node attributes reconstruction.
Take the learned node embedding from the structure encoder
and the learned attribute embedding from the attribute
encoder as inputs, the interactions between the network structure
and the node attribute are jointly captured by both structure
decoder and attribute decoder during the training. Finally,
anomalies in the network can be measured by the reconstruction
errors of network structure and node attribute.
Fan et al. presumes a pre-existing attributed network in which the adjacency matrix is
supplied as an input and remains fixed throughout processing. Fan does not disclose generating a
knowledge graph from document layout, inferring edge connectivity based on similarity
thresholds, or comparing graph-level structural topology across documents. Instead, Fan
reconstructs a given adjacency matrix to compute node-level anomaly scores. Thus, Fan operates
on an already-defined network and cannot reasonably be combined to teach Assignee's claimed
layout-derived knowledge graph construction and structural divergence analysis.”
In response to the Applicant's argument that Fan fails to disclose limitations of the amended claims, the Examiner notes that one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986). As noted in previous responses above, the claimed invention is found to be obvious over the prior art in view of Liang, Peng, Fan, and Belligundu, not merely Fan on its own. Thus, no rejections are withdrawn on these grounds.
On pages 26-27 of the instant remarks, the Applicant argues that Belligundu does not disclose the claimed invention:
“Belligundu Does Not Render the Claims Obvious
Belligundu discloses anomaly detection using multimodal knowledge graphs, generally in
the context of enterprise or operational data. While Belligundu may describe combining
multimodal data into a knowledge graph and performing anomaly detection, it does not disclose
or suggest the document-specific, layout-anchored architecture recited in the claims.
In particular, Belligundu does not teach:
(1) deriving nodes from page elements identified through document page layout
analysis;
(2) explicitly determining edges between page-element nodes based on similarity
thresholds;
(3) re-encoding a layout-derived graph while preserving layout-based relational
information; or
(4) comparing graph-level structural topology of documents to identify abnormal
document structures.
At most, Belligundu shows that anomaly detection can be applied to multimodal
knowledge graphs in general. It does not provide the missing teachings required to transform
Liang' s document labeling graphs into the claimed abnormal document self-discovery system.
Belligundu et al. relies on a pre-constructed multimodal knowledge graph in which
entities and relationships are stored prior to analysis. Belligundu does not disclose constructing a
knowledge graph from document layout, inferring edges based on similarity evaluation, or
dynamically generating graph topology from input documents. Instead, relationships are
retrieved from an existing graph and used as contextual background information. Accordingly,
Belligundu cannot supply the claimed limitations directed to layout-derived knowledge graph
construction or edge inference.
The anomalies addressed by Belligundu are detected at the attribute or rule-violation
level, based on deviations from expected values stored in the knowledge graph. Belligundu does
not detect anomalies by comparing graph-level structure, topology, or relational divergence
between multiple documents. In contrast, Applicant's claims expressly require detecting
abnormal documents based on structural differences in inferred knowledge graphs, a concept
wholly absent from Belligundu.
Even if combined with Liang or Fan, Belligundu would not remedy the deficiencies of
those references. Liang presumes predefined layout models, Fan assumes a fixed adjacency
matrix, and Belligundu relies on a static knowledge graph. None of the references teaches or
suggests dynamically constructing a knowledge graph from document layout, explicitly inferring
relational edges via similarity thresholds, or comparing graph-level topology across documents
to identify abnormal document structure. The Office's combination therefore requires
impermissible hindsight reconstruction.”
In response to the Applicant's argument that Belligundu fails to disclose limitations of the amended claims, the Examiner notes that one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986). As noted in previous responses above, the claimed invention is found to be obvious over the prior art in view of Liang, Peng, Fan, and Belligundu, not merely Belligundu on its own. Thus, no rejections are withdrawn on these grounds.
On page 27 of the instant remarks, the Applicant argues that combinations of the rejection are based on improper hindsight:
“The Proposed Combination Is Improper and Based on Hindsight
The Office's rejection relies on combining:
• Liang' s template-based layout graph matching for labeling,
• Fan's dual-autoencoder anomaly detection on abstract networks, and
• Belligundu' s general multimodal knowledge graph anomaly concepts.
There is no teaching, suggestion, or motivation in the cited art that would lead a person of
ordinary skill in the art to modify Liang' s document labeling system to perform document-level
anomaly detection by importing Fan's network-centric anomaly autoencoders and further
modifying the system using Belligundu' s multimodal knowledge graph concepts. The references
address different problems in different technical domains, and the proposed combination would
require wholesale redesign of Liang's system, not routine optimization.
Such reconstruction can only be achieved with knowledge of Applicant's invention,
which is precisely the type of hindsight reasoning prohibited under § 103. See KSR Int 'l Co. v.
Teleflex Inc., 550 U.S. 398, 421 (2007) (rejecting hindsight reconstruction).”
In response to applicant's argument that the Examiner's conclusion of obviousness is based upon improper hindsight reasoning, it must be recognized that any judgment on obviousness is in a sense necessarily a reconstruction based upon hindsight reasoning. But so long as it takes into account only knowledge which was within the level of ordinary skill at the time the claimed invention was made, and does not include knowledge gleaned only from the applicant's disclosure, such a reconstruction is proper. See In re McLaughlin, 443 F.2d 1392, 170 USPQ 209 (CCPA 1971).
Prior to the filing of the instant application, Liang disclosed a method of constructing knowledge graphs from documents, and Fan disclosed a method of analyzing knowledge graphs for structural anomalies. One of ordinary skill in the art in possession of Liang would have clearly seen the benefit of applying an anomaly detection method to the constructed graphs, as this would automatically determine errors and / or inconsistencies in the graphs and consequently their corresponding documents. Fan’s method would have been a beneficial choice, as it outperformed many contemporary similar methods and more effectively captured cross-modality interactions between graph structures and node attributes often ignored in similar art, as made evident by Fan, page 1, Abstract & page 4, left column, paragraph 3. Additionally, using the output of Liang’s system (a knowledge graph) as input for Fan’s system would certainly not require wholesale design of Liang’s system. Belligundu merely discloses generic computer hardware capable of running a similar system, which one of ordinary skill would have known to use to actually perform the methods of Liang and Fan.
Regarding the assertion that Liang, Fan, and Belligundu address different problems in different technical domains, the Examiner notes that references in a rejection under 35 U.S.C. 103 need not be analogous to each other, as noted by MPEP 2141.01(a): In order for a reference to be proper for use in an obviousness rejection under 35 U.S.C. 103 , the reference must be analogous art to the claimed invention. In re Bigio, 381 F.3d 1320, 1325, 72 USPQ2d 1209, 1212 (Fed. Cir. 2004). A reference is analogous art to the claimed invention if: (1) the reference is from the same field of endeavor as the claimed invention (even if it addresses a different problem); or (2) the reference is reasonably pertinent to the problem faced by the inventor (even if it is not in the same field of endeavor as the claimed invention) … When more than one prior art reference is used as the basis of an obviousness rejection, it is not required that the references be analogous art to each other. See Sanofi-Aventis Deutschland GMbH v. Mylan Pharms. Inc., 66 F.4th 1373, 1380, 2023 USPQ2d 552 (Fed. Cir. 2023) and Corephotonics, Ltd. v. Apple Inc., 84 F.4th 990, 1007, 2023 USPQ2d 1202 (Fed. Cir. 2023).
Regardless of the technical domains of Liang, Fan, and Belligundu individually, each one is directly analogous to the claimed invention and lies in its same field of endeavor. Fan and Belligundun are directed toward anomaly detection in knowledge graphs, while Liang is directed to constructing knowledge graphs representative of document properties, both core functions of the claimed invention. No rejections are withdrawn on these grounds. See the 103 rejections section for more detail.
On pages 29-30 of the instant remarks, the Applicant argues that Liang, Fan, and Belligundu fail to disclose claims 4, 11, and 18:
“Claims 4, 11, and 18 recite determining whether an edge is present between vectors when
a similarity between the vectors exceeds a predetermined edge threshold. Liang does not disclose
threshold-based edge formation driven by similarity between vector representations of page
elements; instead, Liang relies on graph matching techniques for labeling document components
relative to predefined logical structures. Fan discloses reconstruction error-based anomaly
detection using autoencoders on attributed networks, but Fan presumes a pre-existing network
and does not teach explicit formation or suppression of edges based on similarity thresholds tied
to document layout relationships. Belligundu likewise does not disclose explicit, threshold-based
edge determination between page-element nodes derived from document layout. Accordingly,
the cited art does not teach or suggest the threshold-based edge determination recited in these
claims.”
In response to the Applicant's argument that Liang, Fan, and Belligundu fail to disclose limitations of claims 4, 11, and 18, the Examiner notes that one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986). Peng discloses a method of determining edges between nodes in a graph based on similarities of the connected nodes relative to a threshold (Peng, page 3514, right column, paragraph 2 & page 3515, left column, paragraph 1), commensurate in scope with the claim language. While Peng fails to disclose performing operations with a processor, this deficiency is remedied by Belligundu, which discloses generic computer hardware (Belligundu, [0017]). Thus, no rejections are withdrawn on these grounds.
On page 30 of the instant remarks, the Applicant argues that Liang, Fan, and Belligundu fail to disclose amended claims 5, 12, and 19:
“Claims 5, 12, and 19 further recite determining whether vectors include an edge by
comparing a multimodal embedding manner of page elements. Liang does not disclose
multimodal embedding of page elements; its layout graphs are based on structural and spatial
relationships, not on combined semantic, visual, and positional embeddings. Fan operates on
attributed networks with learned embeddings but does not disclose multimodal embeddings
derived from document page elements or using such embeddings to determine edge existence in
a document layout graph. While Belligundu may generally discuss multimodal knowledge
graphs, it does not teach applying multimodal embeddings specifically to document page
elements for the purpose of determining edge connectivity in a layout-anchored knowledge
graph. Thus, the multimodal embedding-based edge determination recited in these claims is
absent from the cited art.”
In response to the Applicant's argument that Liang, Fan, and Belligundu fail to disclose limitations of claims 5, 12, and 19, the Examiner notes that one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986). Peng discloses a method of determining edge connections in a graph by comparing node vector embeddings (Peng, page 3515, left column, paragraph 1), commensurate in scope with the claim language. While Peng fails to disclose determining edges by comparing multimodal page element embeddings, this deficiency is remedied by Liang, which discloses generating multimodal page element embeddings as nodes in a graph. Thus, no rejections are withdrawn on these grounds.
On page 30 of the instant remarks, the Applicant argues that Liang, Fan, and Belligundu do not disclose claims 6, 13, or 20:
“Claims 6, 13, and 20 recite employing pairwise comparisons based on a graph attention
algorithm for the re-encoded knowledge graph and one or more other knowledge graphs. Liang
predates graph attention mechanisms and does not disclose re-encoding graphs using attention-based
neural architectures. Fan discloses dual autoencoders for anomaly detection, not graph
attention algorithms applied to layout-derived knowledge graphs or pairwise comparison of reencoded
graphs. Belligundu does not disclose or suggest re-encoding document layout graphs
using graph attention algorithms for pairwise comparison across documents. The introduction of
a graph attention-based re-encoding and comparison step represents a further technical
distinction that is neither taught nor suggested by the cited references.”
In response to the Applicant's argument that Liang, Fan, and Belligundu fail to disclose limitations of claims 6, 13, and 20, the Examiner notes that one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986). Fan discloses a system that performs pairwise comparisons between nodes based on an attention mechanism (Fan, page 2, right column, paragraphs 2-3), commensurate in scope with the claim language. While Fan doesn’t disclose performing this procedure on a re-encoded knowledge graph, this deficiency is remedied by Liang, which discloses re-encoding a knowledge graph. Thus, no rejections are withdrawn on these grounds.
On pages 30-31 of the instant remarks, the Applicant argues that Liang, Fan, and Belligundu fail to disclose claims 7 and 14:
“Claims 7 and 14 recite determining abnormality when an abnormal score of a node is
significantly higher than corresponding nodes of other documents. Liang does not compute
anomaly scores at all; its system performs labeling, not anomaly detection. Fan computes
anomaly scores within a single attributed network, not node-level abnormality across multiple
document-derived graphs based on layout structure. Belligundu may discuss anomaly scores in a
general sense, but does not teach node-level abnormality determination within layout-derived
document graphs or comparing such scores across documents to determine abnormal document
structures. Accordingly, the abnormal score-based determinations recited in these claims are not
suggested by the cited art.”
In response to the Applicant's argument that Liang, Fan, and Belligundu fail to disclose limitations of claims 7 and 14, the Examiner notes that one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986). Fan discloses a method of determining node anomaly scores, which are compared to a threshold to detect anomalous graphs (Fan, page 3, right column, paragraph 3), commensurate in scope with the claim language. While Fan doesn’t disclose performing this procedure on graphs representing documents, this deficiency is remedied by Liang, which discloses creating knowledge graphs representing documents. Thus, no rejections are withdrawn on these grounds.
On page 31 of the instant remarks, the Applicant argues that it would not have been obvious to combine the relied upon references:
“The Office's rejection of these dependent claims appears to rely on the same generalized
rationale asserted for the independent claims-namely, that it would have been obvious to
combine Liang's layout graphs with Fan's anomaly detection and Belligundu's multimodal
knowledge graphs. As explained above, this rationale is improper because it requires significant
redesign of the cited systems and relies on Applicant's disclosure as a blueprint. The additional
limitations recited in claims 4-7, 11-14, and 18-20 further increase the technical distance
between the claimed invention and the cited art, and underscore the absence of any teaching,
suggestion, or motivation to arrive at the claimed subject matter.”
Regarding the Applicant’s arguments above, the Examiner respectfully disagrees. Regarding claim 1, Liang discloses a system for deriving representative knowledge graphs from documents. Peng discloses a system of determining edges based on node similarities, which would entail only a slight modification of Liang’s method to generate edges in a different way. Peng’s method ensures homophily between nodes, a key property to maintain if one were trying to perform graph anomaly detection on Liang’s graphs, as disclosed by Peng, page 3513, right column, paragraph 2. Fan, as discussed in previous arguments, can provide significant benefits in terms of error detection ability for Liang’s graphs and greater efficiency than comparable graph anomaly detection methods, and wouldn’t require any reworking of Liang or Peng’s systems. Belligundu discloses generic computer hardware capable of executing these systems, which one have been obvious to integrate for one of ordinary skill in the art if they were to execute these systems with real data. Similar arguments can be made for the rest of the claims, and are in greater detail in the 103 rejections section.
No rejections are withdrawn on these grounds.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Ma et al. (A Comprehensive Survey on Graph Anomaly Detection with Deep Learning, published April 2022, arXiv:2106.07178v5) discloses a multitude of state-of-the-art methods for graph anomaly detection from near the effective filing date of the instant application
Ni et al. (Semantic Documents Relatedness using Concept Graph Representation, published 2016, WSDM '16: Proceedings of the Ninth ACM International Conference on Web Search and Data Mining Pages 635 – 644) discloses a method of constructing a concept graph with similarity edges from a document
Hu et al. (An Embedding Approach to Anomaly Detection, published 2016, ICDE 2016) discloses a method of embedding and connecting nodes in a graph such that each cluster of nodes is related to some particular community.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Aaron P Gormley whose telephone number is (571)272-1372. The examiner can normally be reached Monday - Friday 12:00 PM - 8:00 PM EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Michelle T Bechtold can be reached at (571) 431-0762. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/AG/Examiner, Art Unit 2148 /MICHELLE T BECHTOLD/ Supervisory Patent Examiner, Art Unit 2148