DETAILED ACTION
Claims 1 and 45-63 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 .
Information Disclosure Statement
The information disclosure statements (IDS) submitted on 04/10/2025 are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statements are being considered by the examiner.
Claim Objections
Claim 57 is objected to because of the following informalities:
In claim 57, “as an accepted feature attributes” should be “as an accepted feature attribute”
Appropriate correction is required.
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.
Claim 55 is rejected under 35 U.S.C. 112(b) or pre-AIA 35 U.S.C. 112, 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 pre-AIA the applicant regards as the invention.
Claim 55 recites the limitation “wherein the counts are weighted counts.” There is insufficient antecedent basis for the limitation “the counts” in the claim. For examination purposes examiner has interpreted “the counts” to be “the count.”
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 and 45-63 rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more
Step 1: Claim 1 recites a method. Claims 45-60 recite a method. Claims 61-63 recite one or more non-transitory, computer-readable media. Therefore, claims 1 and 45-60 are directed to a process, and claims 61-63 are directed to a manufacture.
With respect to claim 1:
2A Prong 1: The claim recites a judicial exception.
selecting… one or more inference rules from among a plurality of inference rules; (mental process – evaluation or judgement; selecting inference rules)
inferring… information based on the one or more data records, wherein inferring the information includes (mental process – evaluation or judgement; inferring information based on the records)
generating the information by applying the selected one or more inference rules to at least the one or more feature attributes. (mental process – evaluation or judgement; generating the information by applying the rules to feature attributes)
2A Prong 2: The judicial exception is not integrated into a practical application.
obtaining… the one or more data records; (insignificant extra-solution activity – MPEP 2106.05(g), (3) data gathering and outputting)
by processing hardware comprising one or more processors… by the processing hardware… by the processing hardware and substantially in real time… (mere instructions to apply an exception, (2) Whether the claim invokes computers - MPEP 2106.05(f); generic computer components)
calling a natural language processing (NLP) engine to generate one or more feature attributes of one or more features of unstructured textual data within the one or more data records, and (mere instructions to apply an exception – MPEP 2106.05(f), (3) The particularity or generality of the application of the judicial exception; using a NLP engine to generate feature attributes)
Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is directed to an abstract idea.
2B: The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception.
obtaining… the one or more data records; (insignificant extra-solution activity – MPEP 2106.05(g), (3) data gathering and outputting, and WURC: receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 - MPEP 2106.05(d)(II)(i))
by processing hardware comprising one or more processors… by the processing hardware… by the processing hardware and substantially in real time… (mere instructions to apply an exception, (2) Whether the claim invokes computers - MPEP 2106.05(f); generic computer components)
calling a natural language processing (NLP) engine to generate one or more feature attributes of one or more features of unstructured textual data within the one or more data records, and (mere instructions to apply an exception – MPEP 2106.05(f), (3) The particularity or generality of the application of the judicial exception; using a NLP engine to generate feature attributes)
Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
With respect to claims 45 and 61:
2A Prong 1: The claim recites a judicial exception.
uses a plurality of knowledge maps that collectively map features of the unstructured textual data to candidate feature attributes; and (mental process – evaluation or judgement,--- using knowledge maps to map features to candidate feature attributes)
generating… one or more accepted feature attributes based at least in part on the candidate feature attributes. (mental process – evaluation or judgement,--- generating accepted attributes based on the candidate attributes)
2A Prong 2: The judicial exception is not integrated into a practical application.
obtaining… the unstructured textual data; (insignificant extra-solution activity – MPEP 2106.05(g), (3) data gathering and outputting)
by processing hardware comprising one or more processors… executing, by the processing hardware, a multi-thread mapping process that… by the processing hardware… (mere instructions to apply an exception, (2) Whether the claim invokes computers - MPEP 2106.05(f); generic computer components)
Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is directed to an abstract idea.
2B: The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception.
obtaining… the unstructured textual data; (insignificant extra-solution activity – MPEP 2106.05(g), (3) data gathering and outputting, and WURC: receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 - MPEP 2106.05(d)(II)(i))
by processing hardware comprising one or more processors… executing, by the processing hardware, a multi-thread mapping process that… by the processing hardware… (mere instructions to apply an exception, (2) Whether the claim invokes computers - MPEP 2106.05(f); generic computer components)
Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
With respect to claims 46 and 62:
2A Prong 1: The claim recites a judicial exception.
wherein the multi-thread mapping process concurrently uses two or more of the plurality of knowledge maps to collectively map the features of the unstructured textual data to the candidate feature attributes (mental process – evaluation or judgement,--- using two knowledge maps to map the features to candidate attributes)
With respect to claims 47 and 63:
2A Prong 1: The claim recites a judicial exception.
wherein at least one of the plurality of knowledge maps maps features to candidate feature attributes based on fixed associations between features and feature attributes. (mental process – evaluation or judgement,--- mapping features to attributes based on fixed associations between features and attributes)
With respect to claim 48:
2A Prong 1: The claim recites a judicial exception.
wherein at least one of the plurality of knowledge maps maps features to candidate feature attributes based on logical expressions. (mental process – evaluation or judgement,--- mapping features to attributes based on logical expressions)
With respect to claim 49:
2A Prong 2: The judicial exception is not integrated into a practical application.
further comprising: generating at least one of the plurality of knowledge maps using a machine learning model. (mere instructions to apply an exception – MPEP 2106.05(f), (3) The particularity or generality of the application of the judicial exception; using a ML model to generate knowledge maps)
Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is directed to an abstract idea.
2B: The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception.
further comprising: generating at least one of the plurality of knowledge maps using a machine learning model. (mere instructions to apply an exception – MPEP 2106.05(f), (3) The particularity or generality of the application of the judicial exception; using a ML model to generate knowledge maps)
Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
With respect to claim 50:
2A Prong 1: The claim recites a judicial exception.
further comprising: prior to the multi-thread mapping process using the plurality of knowledge maps, selecting… a primary knowledge map; and (mental process – evaluation or judgement,--- selecting a primary knowledge map)
selecting… one or more secondary knowledge maps based on the primary knowledge map, (mental process – evaluation or judgement,--- selecting secondary knowledge maps based on the primary knowledge map)
wherein the multi-thread mapping process uses the primary knowledge map to map the features of the unstructured textual data to the candidate feature attributes, and uses the one or more secondary knowledge maps to determine one or more additional feature attributes. (mental process – evaluation or judgement,--- using the primary knowledge map to map the features to the candidate attributes, and using the secondary knowledge maps to determine additional attributes)
2A Prong 2: The judicial exception is not integrated into a practical application.
by the processing hardware… by processing hardware… (mere instructions to apply an exception, (2) Whether the claim invokes computers - MPEP 2106.05(f); generic computer components)
Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is directed to an abstract idea.
2B: The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception.
by the processing hardware… by processing hardware… (mere instructions to apply an exception, (2) Whether the claim invokes computers - MPEP 2106.05(f); generic computer components)
Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
With respect to claim 51:
2A Prong 1: The claim recites a judicial exception.
wherein at least one of the plurality of knowledge maps maps features to candidate feature attributes based on: semantics of text within the unstructured textual data; and/or positions of text within the unstructured textual data. (mental process – evaluation or judgement,--- mapping features to attributes based on semantics or positions of text)
With respect to claim 52:
2A Prong 1: The claim recites a judicial exception.
wherein at least one of the plurality of knowledge maps determines whether feature attributes are positively or negatively expressed in the unstructured textual data. (mental process – evaluation or judgement,--- determining if attributes are positively or negatively expressed)
With respect to claim 53:
2A Prong 2: The judicial exception is not integrated into a practical application.
wherein the plurality of knowledge maps includes knowledge maps configured to recognize different clinical code formats. (a particular technological environment or field of use – MPEP 2106.05(h); a field of clinical code)
Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is directed to an abstract idea.
2B: The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception.
wherein the plurality of knowledge maps includes knowledge maps configured to recognize different clinical code formats. (a particular technological environment or field of use – MPEP 2106.05(h); a field of clinical code)
Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
With respect to claim 54:
2A Prong 1: The claim recites a judicial exception.
wherein the plurality of knowledge maps includes a set of primary knowledge maps, and wherein generating the one or more accepted feature attributes includes: selectively designating or not designating a particular candidate feature attribute as an accepted feature attribute based at least in part on a count of how many knowledge maps in the set of primary knowledge maps output the particular candidate feature attribute. (mental process – evaluation or judgement,--- designating or not designating a particular attribute as an accepted attribute based on a count of how many knowledge maps output the particular attribute)
With respect to claim 55:
2A Prong 1: The claim recites a judicial exception.
further comprising: assigning… a respective weight to each of one or more of the knowledge maps in the set of primary knowledge maps, wherein the counts are weighted counts determined in accordance with the one or more respective weights (mental process – evaluation or judgement,--- assigning a respective weight to each knowledge maps)
2A Prong 2: The judicial exception is not integrated into a practical application.
by the processing hardware… (mere instructions to apply an exception, (2) Whether the claim invokes computers - MPEP 2106.05(f); generic computer components)
Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is directed to an abstract idea.
2B: The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception.
by the processing hardware… (mere instructions to apply an exception, (2) Whether the claim invokes computers - MPEP 2106.05(f); generic computer components)
Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
With respect to claim 56:
2A Prong 1: The claim recites a judicial exception.
wherein selectively designating or not designating the particular candidate feature attribute as an accepted feature attribute includes: designating or not designating the particular candidate feature attribute as an accepted feature attribute according to a voting scheme. (mental process – evaluation or judgement,--- designating or not designating a particular attribute as an accepted attribute by voting)
With respect to claim 57:
2A Prong 1: The claim recites a judicial exception.
wherein selectively designating or not designating the particular candidate feature attribute as an accepted feature attributes includes: designating or not designating the particular candidate feature attribute as an accepted feature attribute based on whether a threshold number of knowledge maps in the set of primary knowledge maps output the particular candidate feature attribute. (mental process – evaluation or judgement,--- designating or not designating a particular attribute based on if a threshold)
With respect to claim 58:
2A Prong 1: The claim recites a judicial exception.
wherein the plurality of knowledge maps includes a set of primary knowledge maps, and wherein generating the one or more accepted feature attributes includes: selectively designating or not designating a particular candidate feature attribute as an accepted feature attribute based on which knowledge map in the set of primary knowledge maps is weighted most heavily. (mental process – evaluation or judgement,--- designating or not designating a particular attribute based on the knowledge map with the most weight)
With respect to claim 59:
2A Prong 2: The judicial exception is not integrated into a practical application.
wherein obtaining the unstructured textual data, executing the multi-thread mapping process, and generating the one or more accepted feature attributes occur substantially in real time. (a particular technological environment or field of use – MPEP 2106.05(h), a use of real-time application; in light of specification [0015] “… a real-time clinical decision support (CDS) application that uses the inferencing and analytics capabilities of the system of FIG. 1.”; [0020] “… a wide-range of near-real-time clinical rule evaluation processes (e.g., computable phenotyping, clinical decision support operations, implementing risk algorithms, etc.).)
Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is directed to an abstract idea.
2B: The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception.
wherein obtaining the unstructured textual data, executing the multi-thread mapping process, and generating the one or more accepted feature attributes occur substantially in real time. (a particular technological environment or field of use – MPEP 2106.05(h), a use of real-time application; in light of specification [0015] “… a real-time clinical decision support (CDS) application that uses the inferencing and analytics capabilities of the system of FIG. 1.”; [0020] “… a wide-range of near-real-time clinical rule evaluation processes (e.g., computable phenotyping, clinical decision support operations, implementing risk algorithms, etc.).)
Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
With respect to claim 60:
2A Prong 2: The judicial exception is not integrated into a practical application.
by the processing hardware… (mere instructions to apply an exception, (2) Whether the claim invokes computers - MPEP 2106.05(f); generic computer components)
further comprising: providing… the one or more accepted feature attributes, and/or one or more other feature attributes derived from the one or more feature attributes, as inputs to an inference engine that applies one or more inference rules to the one or more accepted feature attributes. (insignificant extra-solution activity – MPEP 2106.05(g), (3) data gathering and outputting)
Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is directed to an abstract idea.
2B: The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception.
by the processing hardware… (mere instructions to apply an exception, (2) Whether the claim invokes computers - MPEP 2106.05(f); generic computer components)
further comprising: providing… the one or more accepted feature attributes, and/or one or more other feature attributes derived from the one or more feature attributes, as inputs to an inference engine that applies one or more inference rules to the one or more accepted feature attributes. (insignificant extra-solution activity – MPEP 2106.05(g), (3) data gathering and outputting, and WURC: receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 - MPEP 2106.05(d)(II)(i))
Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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.
Claim 1 rejected under 35 U.S.C. 103 as being unpatentable over Liu ("Tensor Graph Convolutional Networks for Text Classification" 20200403) in view of Monnin ("Matching and mining in knowledge graphs of the Web of data - Applications in pharmacogenomics" 20210126) in further view of Zheng ("Diagnostic Regions Attention Network (DRA-Net) for Histopathology WSI Recommendation and Retrieval" 20201222)
In regard to claim 1, Liu teaches: A method for efficiently inferring information from one or more data records, the method comprising: (Liu, p. 8409, Abstract "A new framework TensorGCN (tensor graph convolutional networks), is presented for this task... The first is intra-graph propagation used for aggregating information from neighborhood nodes in a single graph. The second isinter-graph propagation used for harmonizing heterogeneous information between graphs. [(using GCN model to) infer information]"; p. 8409, Introduction "In practice, numerous real applications can be cast into a text classification problem, such as document organization, news filtering, spam detection, EHR based disease diagnoses [EHR data records]")
obtaining… the one or more data records; (Liu, p. 8409, Introduction "Text classification is one of the most fundamental tasks in the natural language processing community. It can be simply formulated as X→y, where X is a piece of text (such as a sentence/document) , and y ∈ [0, 1]n is the corresponding label vector... In practice, numerous real applications can be cast D13into a text classification problem, such as document organization, news filtering, spam detection, EHR based disease diagnoses [EHR data records]")
… calling a natural language processing (NLP) engine to generate one or more feature attributes of one or more features of unstructured textual data within the one or more data records, and (Liu, p. 8409, Introduction "Text classification is one of the most fundamental tasks in the natural language processing [NLP engine] community... numerous real applications can be cast into a text classification problem, such as document organization, news filtering, spam detection, EHR based disease diagnoses [unstructured textual data within the one or more data records]"; p. 8410, Graph convolutional networks (GCN) "In GCN learning, hidden layer representations are obtained by encoding both graph structure and features of nodes with a kind of propagation rule H (l+1) = ... (1) [generating
PNG
media_image1.png
454
1454
media_image1.png
Greyscale
feature attributes]")
Liu does not teach, but Monnie teaches: selecting… one or more inference rules from among a plurality of inference rules; (Monnin, p. 92, 5.3 Evaluating the influence of applying inference rules associated with domain knowledge "Semantic Web knowledge graphs are represented within formalims such as Description Logics [9] that are equipped with inference rules... when considering such inference rules, independently or combined. Here, we only consider the following logic axioms: class and predicate assertions, equivalence axioms between entities or classes, subsumption axioms between classes or predicates, and axioms defining predicate inverses. Accordingly, we generate six different graphs by running over K the inference rules associated with these different axioms [selected inference rules] until saturation.")
inferring… information based on the one or more data records, wherein inferring the information includes (Monnin, p. 39, Using data mining techniques to mine Semantic Web knowledge graphs "... knowledge graphs themselves are also interesting targets to mine. [inferring information on KG] Hence, numerous approaches have been proposed for such a Semantic Web mining… Mining knowledge graphs can also be useful in the biomedical domain. Hence, Odgers and Dumontier [128] propose to transform Electronic Health Records [data records, based on EHR data] of the STRIDE Clinical Data Warehouse into a knowledge graph. STRIDE is a database that contains EHR data from the Lucile Packard Children’s Hospital and Stanford Hospital and Clinics. The resulting knowledge graph... We also consider such a combined use in our work, particularly in Chapter 5 and Chapter 6.")
PNG
media_image2.png
341
1150
media_image2.png
Greyscale
… generating the information by applying the selected one or more inference rules to at least the one or more feature attributes. (Monnin, p. 89, Figure 5.1 "Outline of our approach. Gold clusters are computed from existing similarity links in the knowledge graph (e.g., owl:sameAs, skos:broadMatch, skos:related, etc.). These similarity links are then removed and various inferences rules associated with domain knowledge are applied on the knowledge graph. [by applying the selected one or more inference rules]"; p. 95, Table 5.3 "Visual summary of the transformations of K to evaluate the influence of the application of inference rules associated with domain knowledge on node matching. Go is the baseline that corresponds to no inference rules being run and the systematic addition of abstract inverses."; generating the transformed knowledge graph(KG) by applying inference rules to nodes [feature attributes] in the KG)
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 Liu to incorporate the teachings of Monnin by applying inference rules associated with domain knowledge to the KG. Doing so would allow us to think that the model is able to 'rediscover' the alignment relations. (Monnin, p. 90, Chapter 5, Rediscovering alignment relations with Graph Convolutional Networks "First, we measure the improvement in clustering results when considering different inference rules associated with domain knowledge, e.g., hierarchies of classes and predicates, symmetry of predicates, etc.... Such results allow us to think that the model is able to 'rediscover' these alignment relations. To the best of our knowledge, our approach is the first one to investigate these aspects in a matching task using GCNs and clustering.")
Liu and Monnie do not teach, but Zheng teaches: … by processing hardware comprising one or more processors… by the processing hardware… by the processing hardware and substantially in real time… (Zheng, p. 1098, F. Efficiency of the Recommendation "The proposed SHIR application is desired to provide real-time assistant to pathologists during the diagnosis... ROI feature extraction based on the GCN... As a result, the average time was 475 ms [substantially in real time] by using one GPU in our experiment environment. The speed is promising to develop real-time AI assistant for cancer diagnosis.")
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 Liu and Monnie to incorporate the teachings of Zheng by including hardware implementation with real-time application. Doing so would provide real-time assistant to pathologists during the diagnosis. (Zheng, p. 1098, F. Efficiency of the Recommendation "The proposed SHIR application is desired to provide real-time assistant to pathologists during the diagnosis.")
Claims 45-46, 49, 51, 54-56, 59 and 61-62 rejected under 35 U.S.C. 103 as being unpatentable over Liu in further view of Zheng
In regard to claims 45 and 61, Liu teaches: A method for efficient natural language processing of unstructured textual data, the method comprising: (Liu, p. 8409, Introduction "Text classification is one of the most fundamental tasks in the natural language processing [NLP engine] community... numerous real applications can be cast into a text classification problem, such as document organization, news filtering, spam detection, EHR based disease diagnoses [unstructured textual data]") obtaining... the unstructured textual data; (Liu, p. 8409, Introduction "numerous real applications can be cast into a text classification problem, such as document organization, news filtering, spam detection, EHR based disease diagnoses [unstructured textual data]")
executing… a multi-thread mapping process that uses a plurality of knowledge maps that collectively map features of the unstructured textual data to candidate feature attributes; and (Liu, p. 8409, Abstract "A new framework TensorGCN (tensor graph convolutional networks), is presented for this task. A text graph tensor is firstly constructed to describe semantic, syntactic, and sequential contextual information. [multi-thread] Then, two kinds of propagation learning perform on the text graph tensor. The first is intra-graph propagation used for aggregating information from neighborhood nodes in a single graph. [multi-thread] The second is inter-graph propagation used for harmonizing heterogeneous information between graphs. [collectively]"; p. 8410, Graph convolutional networks (GCN) "In GCN learning, hidden layer representations are obtained by encoding both graph structure and features of nodes with a kind of propagation rule H (l+1) = ... (1) [candidate feature attributes]"; see Fig. 1 and Fig. 3, three graphs (semantic, syntactic, and sequential graphs) are provided to a GCN model, going through intra-graph and inter-graph propagation)
generating... one or more accepted feature attributes based at least in part on the candidate feature attributes. (Liu, p. 8410, Graph convolutional networks (GCN) "In GCN learning, hidden layer representations are obtained by encoding both graph structure and features of nodes with a kind of propagation rule H (l+1) = ... (1)"; p. 8412, Graph tensor learning "we employ a simple edge-wise attention strategy to harmonize edge weights from different graphs. The adjacency matrix of the merged graph is A_merge = pooling(A) = ... For each layer of TensorGCN, we perform two kinds of propagation learning: first intra-graph propagation and then inter-graph propagation (see figure 3). We take the lth layer of TensorGCN for example. H(l)→H(l)→H(l+1) (6)"; through propagation, the original representation is refined into the updated version, i.e. generating updated representation [accepted
PNG
media_image3.png
436
1472
media_image3.png
Greyscale
feature attributes] based on the original representation [the candidate feature attributes])
Liu does not teach, but Zheng teaches: by processing hardware comprising one or more processors… by the processing hardware… by the processing hardware… (Zheng, p. 1098, F. Efficiency of the Recommendation "The proposed SHIR application is desired to provide real-time assistant to pathologists during the diagnosis... ROI feature extraction based on the GCN... As a result, the average time was 475 ms by using one GPU in our experiment environment. The speed is promising to develop real-time AI assistant for cancer diagnosis.")
Claim 61 recite substantially the same limitation as claim 1, therefore the rejection applied to claim 45 also apply to claim 61. In addition, Zheng teaches: One or more non-transitory, computer-readable media storing instructions that, when executed by a computing system, cause the computing system to: (Zheng, p. 1098, F. Efficiency of the Recommendation "the average time was 475 ms by using one GPU in our experiment environment.")
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 Liu to incorporate the teachings of Zheng by including hardware implementation with real-time application. Doing so would provide real-time assistant to pathologists during the diagnosis. (Zheng, p. 1098, F. Efficiency of the Recommendation "The proposed SHIR application is desired to provide real-time assistant to pathologists during the diagnosis.")
In regard to claims 46 and 62, Liu teaches: wherein the multi-thread mapping process concurrently uses two or more of the plurality of knowledge maps to collectively map the features of the unstructured textual data to the candidate feature attributes. (Liu, p. 8410, Introduction "A text graph tensor is constructed to describe contextual information with semantic, syntactic, and sequential constraints, respectively"; see Fig. 1 and Fig. 3, three graphs (semantic, syntactic, and sequential graphs) [two or more of the knowledge maps] are provided to a GCN model)
In regard to claim 49, Liu teaches: further comprising: generating at least one of the plurality of knowledge maps using a machine learning model. (Liu, p. 8409, Abstract "A new framework TensorGCN (tensor graph convolutional networks), [a machine learning model] is presented for this task."; see Fig. 1 and Fig. 3, GCN model generating the updated graph structures [knowledge maps] by using intra-graph and inter-graph propagation)
In regard to claim 51, Liu teaches: wherein at least one of the plurality of knowledge maps maps features to candidate feature attributes based on: semantics of text within the unstructured textual data; and/or positions of text within the unstructured textual data. (Liu, p. 8411, Semantic-based graph "we propose a LSTM-based method to construct a semantic-based graph from text documents [semantics of text]")
In regard to claim 54, Liu teaches: wherein the plurality of knowledge maps includes a set of primary knowledge maps, and wherein generating the one or more accepted feature attributes includes: selectively designating or not designating a particular candidate feature attribute as an accepted feature attribute based at least in part on a count of how many knowledge maps in the set of primary knowledge maps output the particular candidate feature attribute. (Liu, p. 8410, Graph convolutional networks (GCN) "In GCN learning, hidden layer representations are obtained by encoding both graph structure and features of nodes with a kind of propagation rule H (l+1) = ... (1)... D is the diagonal node degree matrix... [a count of connected edges (to other nodes or KGs)]"; through propagation, the original representation is refined [selectively designating or not designating ] into the updated version)
In regard to claim 55, Liu teaches: further comprising: assigning, by the processing hardware, a respective weight to each of one or more of the knowledge maps in the set of primary knowledge maps, wherein the counts are weighted counts determined in accordance with the one or more respective weights. (Liu, p. 8410, Graph convolutional networks (GCN) "In GCN learning, hidden layer representations are obtained by encoding both graph structure and features of nodes with a kind of propagation rule H (l+1) = ... (1)... where A^ = DAD is a symmetric normalization of the self-connections added adjacency matrix... D is the diagonal node degree matrix... [also as a respective weight (to other nodes or KGs), D scales the edge weight by the degrees of both the sending and receiving nodes]"; through propagation, the original representation is refined [selectively designating or not designating ] into the updated version)
In regard to claim 56, Liu teaches: wherein selectively designating or not designating the particular candidate feature attribute as an accepted feature attribute includes: designating or not designating the particular candidate feature attribute as an accepted feature attribute according to a voting scheme. (Liu, p. 8410, Graph convolutional networks (GCN) "In GCN learning, hidden layer representations are obtained by encoding both graph structure and features of nodes with a kind of propagation rule H (l+1) = ... (1)"; p. 8412, Graph tensor learning "we employ a simple edge-wise attention strategy to harmonize edge weights from different graphs. The adjacency matrix of the merged graph is A_merge = pooling(A) = Σ [a voting scheme] W_edge A... For each layer of TensorGCN, we perform two kinds of propagation learning: first intra-graph propagation and then inter-graph propagation (see figure 3). We take the lth layer of TensorGCN for example. H(l)→H(l)→H(l+1) (6)"; through propagation, the original representation is refined [selectively designating or not designating ] into the updated version)
In regard to claim 59, Liu teaches: Liu does not teach, but Zheng teaches: wherein obtaining the unstructured textual data, executing the multi-thread mapping process, and generating the one or more accepted feature attributes occur substantially in real time. (Zheng, p. 1098, F. Efficiency of the Recommendation "The proposed SHIR application is desired to provide real-time assistant to pathologists during the diagnosis... ROI feature extraction based on the GCN... As a result, the average time was 475 ms [substantially in real time] by using one GPU in our experiment environment. The speed is promising to develop real-time AI assistant for cancer diagnosis."; the whole process can be done in real time)
The rationale for combining the teachings of Liu and Zheng is the same as set forth in the rejection of claim 45.
Claims 47, 52-53 and 63 rejected under 35 U.S.C. 103 as being unpatentable over Liu and Zheng, as applied to claim 45, and in further view of Luo ("Automatic lymphoma classification with sentence subgraph mining from pathology reports" 20140731)
In regard to claims 47 and 63, Liu and Zheng do not teach, but Luo teaches: wherein at least one of the plurality of knowledge maps maps features to candidate feature attributes based on fixed associations between features and feature attributes. (Luo, p. 2 Methods "We then performed a two-phase sentence-parsing step, grouping token subsequences that match concept unique identifiers (CUIs) in the Unified Medical Language System (UMLS) Metathesaurus [mapping using a fixed association] as parsing units to Stanford Parser instead of individual tokens"; the UMLS Metathesaurus is a biomedical vocabulary database. By using the database, medical terms can be found as nodes; in light of specification [0099] "The knowledge maps may be configured to map features to candidate feature attributes based on fixed associations between features and feature attributes (e.g., in a relational database)")
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 Liu and Zheng to incorporate the teachings of Luo by including the Unified Medical Language System (UMLS) Metathesaurus. Doing so would ensure meaningful interpretations over the sentence graphs. (Luo, p. 1, Abstract "We design a novel framework that translates sentences into graph representations, automatically mines sentence subgraphs, reduces redundancy in mined subgraphs, and automatically generates subgraph features for subsequent classification tasks. To ensure meaningful interpretations over the sentence graphs, we use the Unified Medical Language System Metathesaurus to map token subsequences to concepts...")
In regard to claim 52, Liu and Zheng do not teach, but Luo teaches: wherein at least one of the plurality of knowledge maps determines whether feature attributes are positively or negatively expressed in the unstructured textual data. (Luo, p. 2 Method "For the UMLS CUI matching, we experimented with the entire set or subsets of CUIs and chose the following approach, which balances the coverage and accuracy on our data. If the token subsequence had only one CUI match, [positively expressed] this CUI was used. If the token subsequence had multiple CUI matches, we selected the one supported by the most sources. If there was a tie, we preferred the CUI supported by Systematized Nomenclature of Medicine—Clinical Terms (SNOMED CT) if there was one, or flipped a coin otherwise."; in light of specification [0039] "Feature attributes may include any attributes that are explicitly or implicitly expressed by or otherwise associated with the features, such as specific codes (e.g., ICD9, ICD10, SNOMED, etc.), dates, ethnicity, gender, age, whether the features positively or negatively express other feature attributes, and so on.")
The rationale for combining the teachings of Liu, Zheng and Luo is the same as set forth in the rejection of claim 47.
In regard to claim 53, Liu and Zheng do not teach, but Luo teaches: wherein the plurality of knowledge maps includes knowledge maps configured to recognize different clinical code formats. (Luo, p. 2, Methods "We then performed a two-phase sentence-parsing step, grouping token subsequences that match concept unique identifiers (CUIs) in the Unified Medical Language System (UMLS) Metathesaurus [recognize different clinical code formats.] as parsing units to Stanford Parser instead of individual tokens"; the UMLS Metathesaurus is a biomedical vocabulary database, which contains ICD-9 and ICD-10. CPT, etc.)
The rationale for combining the teachings of Liu, Zheng and Luo is the same as set forth in the rejection of claim 47.
Claims 48 and 60 rejected under 35 U.S.C. 103 as being unpatentable over Liu and Zheng, as applied to claim 45, and in further view of Monnie
In regard to claim 48, Liu and Zheng do not teach, but Monnie teaches: wherein at least one of the plurality of knowledge maps maps features to candidate feature attributes based on logical expressions. (Monnin, p. 89, Figure 5.1 "Outline of our approach. Gold clusters are computed from existing similarity links in the knowledge graph (e.g., owl:sameAs, skos:broadMatch, skos:related, etc.). These similarity links are then removed and various inferences rules associated with domain knowledge are applied on the knowledge graph. [mapping based on inference rules, which are based on logical expressions]"; p. 92, 5.3 Evaluating the influence of applying inference rules associated with domain knowledge "Semantic Web knowledge graphs are represented within formalims such as Description Logics [9] that are equipped with inference rules... when considering such inference rules, independently or combined. Here, we only consider the following logic axioms: [logical expressions] class and predicate assertions, equivalence axioms between entities or classes, subsumption axioms between classes or predicates, and axioms defining predicate inverses. Accordingly, we generate six different graphs by running over K the inference rules associated with these different axioms until saturation.")
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 Liu and Zheng to incorporate the teachings of Monnie by applying inference rules associated with domain knowledge to the KG. Doing so would allow us to think that the model is able to 'rediscover' the alignment relations. (Monnin, p. 90, Chapter 5, Rediscovering alignment relations with Graph Convolutional Networks "First, we measure the improvement in clustering results when considering different inference rules associated with domain knowledge, e.g., hierarchies of classes and predicates, symmetry of predicates, etc.... Such results allow us to think that the model is able to 'rediscover' these alignment relations. To the best of our knowledge, our approach is the first one to investigate these aspects in a matching task using GCNs and clustering.")
In regard to claim 60, Liu and Zheng do not teach, but Monnie teaches: further comprising: providing, by the processing hardware, the one or more accepted feature attributes, and/or one or more other feature attributes derived from the one or more feature attributes, as inputs to an inference engine that applies one or more inference rules to the one or more accepted feature attributes. (Monnin, p. 89, Figure 5.1 "Outline of our approach. Gold clusters are computed from existing similarity links in the knowledge graph (e.g., owl:sameAs, skos:broadMatch, skos:related, etc.). These similarity links are then removed and various inferences rules associated with domain knowledge are applied on the knowledge graph. [an inference engine that applies one or more inference rules]"; p. 95, Table 5.3 "Visual summary of the transformations of K to evaluate the influence of the application of inference rules associated with domain knowledge on node matching. Go is the baseline that corresponds to no inference rules being run and the systematic addition of abstract inverses."; applying inference rules to nodes [feature attributes] in the KG)
The rationale for combining the teachings of Liu, Zheng and Monnie is the same as set forth in the rejection of claim 48.
Claim 50 rejected under 35 U.S.C. 103 as being unpatentable over Liu and Zheng, as applied to claim 45, and in further view of Ruan ("Relation Extraction for Chinese Clinical Records Using Multi-View Graph Learning" 20201111)
In regard to claim 50, Liu and Zheng do not teach, but Ruan teaches: further comprising: prior to the multi-thread mapping process using the plurality of knowledge maps, selecting, by the processing hardware, a primary knowledge map; and selecting, by processing hardware, one or more secondary knowledge maps based on the primary knowledge map, (Ruan, p. 215616, Obtaining View-Oriented Features "Aspect-based (also known as aspect-level) sentiment classification aims at identifying the sentiment polarities of aspects explicitly given in sentences [20]. Learning from this idea of mining text from multiple view, the features of graph should be able to reflect prior knowledge of the labels of edges. we aim to obtain view-oriented features by applying multi-view graph neural networks over the context of a clinical record, and imposing an view-specific graph according to different features. Inspired by work [4], we employed three view that categories of features to construct our heterogeneous medical graph: Co-occurring view... Lexical view... Semantic view... [secondary knowledge maps] "; before heterogeneous medical graph are provided to GCN model [prior to the mapping process] , multi-view graphs are generated by extracting different features or relations from a single primary graph [selecting secondary knowledge maps based on the primary knowledge map]; see Fig. 3 Multi-view, all views share identical node structure [primary knowledge map] but use different edge connections [secondary knowledge maps])
wherein the multi-thread mapping process uses the primary knowledge map to map the features of the unstructured textual data to the candidate feature attributes, and uses the one or more secondary knowledge maps to determine one or more additional feature attributes. (Ruan, p. 215617, C. Heterogeneous Graph Embedding "The convolution computation for node i at the l-th layer of GCN, which takes the input feature representation h(l-1) as input and outputs the induced representation h(il) , can be defined as: h(il) = … (3)... We draw on the idea of the work [29] that considers the difference of various types of information and projects them into an implicit common space with a dual-level attention mechanism... We use the heterogeneous graph convolution to learn the representation of heterogeneous entities [determine additional feature attributes, determining the node representation] formulated as follows: H (l+1) = ...(4)... The transformation matrix W... considers the difference of different feature spaces [using the primary knowledge map and secondary knowledge
PNG
media_image4.png
441
1027
media_image4.png
Greyscale
maps] and projects them into an implicit common space...")
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 Liu and Zheng to incorporate the teachings of Ruan by including multi-view graphs. Doing so would allow the extracted features capture and reflect prior knowledge, and these features can be used in the model. (Ruan, p. 215616, Obtaining View-Oriented Features "Learning from this idea of mining text from multiple view, the features of graph should be able to reflect prior knowledge of the labels of edges.")
Claims 57-58 rejected under 35 U.S.C. 103 as being unpatentable over Liu and Zheng, as applied to claim 54, and in further view of Shen ("NPI-GNN: Predicting ncRNA–protein interactions with deep graph neural networks" 20210405)
In regard to claim 57, Liu teaches: wherein selectively designating or not designating the particular candidate feature attribute as an accepted feature attributes includes: designating or not designating the particular candidate feature attribute as an accepted feature attribute… (Liu, p. 8410, Graph convolutional networks (GCN) "In GCN learning, hidden layer representations are obtained by encoding both graph structure and features of nodes with a kind of propagation rule H (l+1) = ... (1)"; p. 8412, Graph tensor learning "we employ a simple edge-wise attention strategy to harmonize edge weights from different graphs. The adjacency matrix of the merged graph is A_merge = pooling(A) = ... For each layer of TensorGCN, we perform two kinds of propagation learning: first intra-graph propagation and then inter-graph propagation (see figure 3). We take the lth layer of TensorGCN for example. H(l)→H(l)→H(l+1) (6)"; through propagation, the original representation is refined [selectively designating or not designating ] into the updated version)
Liu and Zheng do not teach, but Shen teaches: based on whether a threshold number of knowledge maps in the set of primary knowledge maps output the particular candidate feature attribute. (Shen, p. 4, Top-k pooling layer "Given a graph with n nodes, the top-k pooling layer will reduce the number of nodes to ⌈kn⌉ according to a projection score against a learnable vector p... top-k function selects top-k indices [a threshold number] from a given vector...")
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 Liu and Zheng to incorporate the teachings of Shen by including top-K ppooling. Doing so would reduce the size of the graph as the GNN goes deeper. (Shen, p. 4, Top-k pooling layer "In NPI-GNN, we use top-k pooling to reduce the size of the graph as the GNN goes deeper.")
In regard to claim 58, Liu teaches: wherein the plurality of knowledge maps includes a set of primary knowledge maps, and wherein generating the one or more accepted feature attributes includes: selectively designating or not designating a particular candidate feature attribute as an accepted feature attribute... (Liu, p. 8410, Graph convolutional networks (GCN) "In GCN learning, hidden layer representations are obtained by encoding both graph structure and features of nodes with a kind of propagation rule H (l+1) = ... (1)"; p. 8412, Graph tensor learning "we employ a simple edge-wise attention strategy to harmonize edge weights from different graphs. The adjacency matrix of the merged graph is A_merge = pooling(A) = ... For each layer of TensorGCN, we perform two kinds of propagation learning: first intra-graph propagation and then inter-graph propagation (see figure 3). We take the lth layer of TensorGCN for example. H(l)→H(l)→H(l+1) (6)"; through propagation, the original representation is refined [selectively designating or not designating ] into the updated version)
Liu and Zheng do not teach, but Shen teaches: based on which knowledge map in the set of primary knowledge maps is weighted most heavily. (Shen, p. 4, Top-k pooling layer "Given a graph with n nodes, the top-k pooling layer will reduce the number of nodes to ⌈kn⌉ according to a projection score against a learnable vector p... top-k function selects top-k indices [weighted most heavily] from a given vector...")
The rationale for combining the teachings of Liu, Zheng and Shen is the same as set forth in the rejection of claim 57.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SU-TING CHUANG whose telephone number is (408)918-7519. The examiner can normally be reached Monday - Thursday 8-5 PT.
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, Usmaan Saeed can be reached at (571) 272-4046. 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.
/S.C./Examiner, Art Unit 2146
/USMAAN SAEED/Supervisory Patent Examiner, Art Unit 2146
Prosecution Insights