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 Objections
Claim 6 is objected to because of the following informalities:
Claim 6 states “wherein the structure content includes at least one of page numbers and table content” which is interpreted as “structure content includes at least one of page numbers and at least one of table content”. This should be written as “wherein the structure content includes at least one of page numbers or table content”, in order to describe the two limitations as alternatives.
Appropriate correction is required.
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.
Claim [1-20] rejected under 35 U.S.C. 101 because the claimed invention is directed to a mental process without significantly more.
Regarding claim 1, 9, 17 recite a method, system, and a non-transitory computer
readable medium, for obtaining a plurality of digital documents; dividing the plurality of digital documents into a plurality of document chunks; determining a similarity between each of the plurality of document chunks; and constructing a multi-document graph based on the similarity between each of the plurality of document chunks, wherein the multi-document graph includes a plurality of nodes representing the plurality of document chunks.
[A person could read multiple documents, divide them into sections, determine which sections are similar based on keywords or content, and create a chart or diagram linking related sections].
As described above, these limitations can be carried out as a series of mental steps.
This judicial exception is not integrated into a practical application because the only additional elements recited are a memory, a processor, and a computer program executing instructions, and these additional elements are nothing more than instructions to apply the mental process using a general-purpose software model and general-purpose hardware.
These claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because, as described above, the only additional elements recited are a memory, a processor, and a computer program executing instructions,, and these additional elements are nothing more than instructions to apply the mental process using general-purpose software and hardware.
Regarding claim 2 and 10, recite a method and system wherein determining a similarity between each of the plurality of document chunks, further comprises: determining one or more keywords associated with each document chunk.
[A person could review each document section and identify keywords or topics associated with that section].
As described above, these limitations can be carried out as a series of mental steps.
This judicial exception is not integrated into a practical application because the only additional elements recited are a memory and a processor and these additional elements are nothing more than instructions to apply the mental process using a general-purpose software model and general-purpose hardware.
These claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because, as described above, the only additional elements recited are a memory and a processor, and these additional elements are nothing more than instructions to apply the mental process using general-purpose software and hardware.
Regarding claim 3 and 11 recite a method and system, wherein constructing a multi-document graph based on the similarity between each of the plurality of chunks, further comprises: connecting pairs of nodes representing the document chunks with at least one shared keyword with an edge to form the multi-document graph.
[A person could identify sections sharing common keywords and draw lines between them on a diagram].
As described above, these limitations can be carried out as a series of mental steps.
This judicial exception is not integrated into a practical application because the only additional elements recited are a memory and a processor and these additional elements are nothing more than instructions to apply the mental process using a general-purpose software model and general-purpose hardware.
These claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because, as described above, the only additional elements recited are a memory and a processor, and these additional elements are nothing more than instructions to apply the mental process using general-purpose software and hardware.
Regarding claim 4 and 12, recite a recite a method and system, wherein determining a similarity between each of the plurality of document chunks, further comprises: generating, by a chunk model, a plurality of embeddings, wherein each embedding corresponds to a document chunk; and determining the similarity between each of the plurality of chunks based on a similarity between each of the corresponding plurality of embeddings.
[A person could represent document sections using descriptive summaries or feature lists and compare those representations to determine similarity].
As described above, these limitations can be carried out as a series of mental steps.
This judicial exception is not integrated into a practical application because the only additional elements recited are a memory and a processor and these additional elements are nothing more than instructions to apply the mental process using a general-purpose software model and general-purpose hardware.
These claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because, as described above, the only additional elements recited are a memory and a processor, and these additional elements are nothing more than instructions to apply the mental process using general-purpose software and hardware.
Regarding claim 5 and 13 recite a method and system, wherein constructing a multi-document graph based on the similarity between each of the plurality of chunks, further comprises: connecting pairs of nodes representing the document chunks determined to be similar with an edge to form the multi-document graph.
[A person could compare document sections and connect those deemed sufficiently related].
As described above, these limitations can be carried out as a series of mental steps.
This judicial exception is not integrated into a practical application because the only additional elements recited are a memory and a processor and these additional elements are nothing more than instructions to apply the mental process using a general-purpose software model and general-purpose hardware.
These claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because, as described above, the only additional elements recited are a memory and a processor, and these additional elements are nothing more than instructions to apply the mental process using general-purpose software and hardware.
Regarding claim 6 and 14 recite a method and system, determining structure content of the plurality of digital documents, wherein the structure content includes at least one of page numbers and table content; and adding one or more structure nodes to the multi-document graph corresponding to the structure content.
[A person could review documents, note structural information such as page numbers or tables, and add corresponding entries to an organization chart].
As described above, these limitations can be carried out as a series of mental steps.
This judicial exception is not integrated into a practical application because the only additional elements recited are a memory and a processor and these additional elements are nothing more than instructions to apply the mental process using a general-purpose software model and general-purpose hardware.
These claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because, as described above, the only additional elements recited are a memory and a processor, and these additional elements are nothing more than instructions to apply the mental process using general-purpose software and hardware.
Regarding claim 7 and 15 recite a method and system, receiving a question to be answered based on the plurality of digital documents; traversing the multi-document graph to obtain a context; and generating an answer to the question based on the context.
[A person could receive a question, review related sections of an index or chart, gather relevant information, and formulate an answer].
As described above, these limitations can be carried out as a series of mental steps.
This judicial exception is not integrated into a practical application because the only additional elements recited are a memory and a processor and these additional elements are nothing more than instructions to apply the mental process using a general-purpose software model and general-purpose hardware.
These claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because, as described above, the only additional elements recited are a memory and a processor, and these additional elements are nothing more than instructions to apply the mental process using general-purpose software and hardware.
Regarding claim 8 and 16 recite a method and system wherein the context includes portions of text corresponding to a plurality of traversed nodes in the multi-document graph.
[A person could collect text excerpts from the sections reviewed while answering a question].
As described above, these limitations can be carried out as a series of mental steps.
This judicial exception is not integrated into a practical application because the only additional elements recited are a memory and a processor and these additional elements are nothing more than instructions to apply the mental process using a general-purpose software model and general-purpose hardware.
These claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because, as described above, the only additional elements recited are a memory and a processor, and these additional elements are nothing more than instructions to apply the mental process using general-purpose software and hardware.
Regarding claim 18, recites a non-transitory computer-readable medium, storing instructions that further cause the processing device to perform operations comprising: determining a starting node of the multi-document graph based on a similarity to the question.
[A person could identify the document section most relevant to a question and begin reviewing information from that section].
As described above, these limitations can be carried out as a series of mental steps.
This judicial exception is not integrated into a practical application because the only additional element recited is a non-transitory computer readable medium and this additional element is nothing more than instructions to apply the mental process using a general-purpose software model and general-purpose hardware.
This claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because, as described above, the only additional element recited is a non-transitory computer readable medium, and this additional element is nothing more than instructions to apply the mental process using general-purpose software and hardware.
Regarding claim 19, recites a non-transitory computer-readable medium, wherein the operation of traversing the multi-document graph to obtain a context, further comprises: identifying a search space including nodes connected to a current node; predicting, using a passage generator, a next text chunk based on the question and a current context including already traversed nodes; and.
[A person could review information already gathered, predict what information may be needed next, and choose the next related section to review based on that prediction].
As described above, these limitations can be carried out as a series of mental steps.
This judicial exception is not integrated into a practical application because the only additional element recited is a non-transitory computer readable medium and this additional element is nothing more than instructions to apply the mental process using a general-purpose software model and general-purpose hardware.
This claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because, as described above, the only additional element recited is a non-transitory computer readable medium, and this additional element is nothing more than instructions to apply the mental process using general-purpose software and hardware.
Regarding claim 20, recites a non-transitory computer-readable medium, wherein the operation of constructing a multi-document graph corresponding to the plurality of digital documents, further comprises: dividing the plurality of digital documents into a plurality of document chunks; and determining a similarity between each of the plurality of document chunks, wherein the similarity is based on keyword similarity or embedding similarity.
[A person could divide documents into sections and compare those sections using keywords, summaries, or descriptive characteristics to determine relatedness].
As described above, these limitations can be carried out as a series of mental steps.
This judicial exception is not integrated into a practical application because the only additional element recited is a non-transitory computer readable medium and this additional element is nothing more than instructions to apply the mental process using a general-purpose software model and general-purpose hardware.
This claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because, as described above, the only additional element recited is a non-transitory computer readable medium, and this additional element is nothing more than instructions to apply the mental process using general-purpose software and hardware.
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.
Claim [ 1, 2, 3, 4, 5, 7, 8, 9, 10, 11, 12, 13, 15, 16, 17, 18, 20 ] are rejected under 35 U.S.C.
103 as being unpatentable over Zhao (US-20260030517-A1) in view of Ghosh (US-
20220277035-A1).
Regarding claim 1, ZHAO teaches
A method comprising: obtaining a plurality of digital documents;
[0016, line 5 "Documents chunks may be generated from a group of documents"].
dividing the plurality of digital documents into a plurality of document chunks;
[0032 "The computing device 100 may include a chunk generator 110. The chunk generator 110 may be any suitable combination of hardware and software on the computing device 100 that may generate document chunks from documents by dividing documents, such as documents 171, into document chunks, such as document chunks 172"].
determining a similarity between each of the plurality of document chunks;
[0016, line 24 "A number of document chunks based on frequency counts of the links from the entity property triplets and entity relation triplets to the document chunks linked to the entity property triplets and entity relation triplets most similar to the keywords of the search query may be determined”].
However, Zhao does not teach constructing a multi-document graph based on the similarity between each of the plurality of document chunks, wherein the multi-document graph includes a plurality of nodes representing the plurality of document chunks.
But Ghosh teaches
constructing a multi-document graph based on the similarity between each of the plurality of document chunks, wherein the multi-document graph includes a plurality of nodes representing the plurality of document chunks.
[0012 “The systems and methods disclosed herein describe generating a connected network graph based on multiple portions of a text such that each portion is a node of the
network graph. A similarity score of the multiple nodes can be determined and a centrality of each node can be measured using graph centrality. The nodes can be ranked based on the
measured centrality and a summary of the text can be generated by using the top ranked nodes”].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine teachings of Zhao into the teachings of Ghosh because Ghosh teaches constructing a graph in which text portions are represented as nodes and connected according to similarity scores. Combining this would facilitate efficient identification and retrieval of related document chunks across a collection of documents.
Regarding claim 2, Zhao teaches The method of claim 1, wherein determining a similarity between each of the plurality of document chunks, further comprises:
determining one or more keywords associated with each document chunk.
[0016, line 20-38 ”A search query including keywords may be received number of nodes and
edges of the knowledge graph that include entities, entity property triplets, and entity relation triplets most similar to the keywords of the search query may be determined.
A number of document chunks based on frequency counts of the links from the entity property triplets and entity relation triplets to the document chunks linked to the entity property triplets and entity relation triplets most similar to the keywords
of the search query may be determined….The document chunks of the determined number of document chunks and the relevant entity property triplets and entity relation triplets may be
sent as a response to the received search query” where the keywords are associated with document chunks by identifying entities and triplets most similar to keywords and determining the document chunks linked to those entities and triplets which represent semantic keywords concepts extracted from document chunks and serve as semantic keywords for identifying relevant chunks.
Regarding claim 3, Zhao does not teach the method of claim 2, wherein constructing a multi-document graph based on the similarity between each of the plurality of chunks, further comprises:
connecting pairs of nodes representing the document chunks with at least one shared keyword with an edge to form the multi-document graph.
However, Ghosh teaches the method of claim 2, wherein constructing a multi-document graph based on the similarity between each of the plurality of chunks, further comprises:
connecting pairs of nodes representing the document chunks with at least one shared keyword with an edge to form the multi-document graph.
[0012 “The systems and methods disclosed herein describe generating a connected network graph based on multiple portions of a text such that each portion is a node of the
network graph" where each portion of text is the document chunk.];
[0017, line 5 "For n number of nodes, there would be total (n(n-1)/2) number of edges. Edges can have many properties based on connectivity structure. For example, strength of an edge can be based on the similarity ( or similarity score) between the two nodes it is connecting." where nodes are connecting with edges based on similarity score.”];
[0019 “For example, a bag of words (BOW) model can be used in step 120 to vectorize the portion of the text and a Jaccard similarity coefficient can be used to determine the
similarity score. In this model, the text portion can be represented as the bag (multiset) of its words, disregarding grammar and word order but keeping multiplicity" where the words are used to determine similarity score.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine teachings of Zhao into the teachings of Ghosh because combining them would thereby be improving identification of related document chunks and enabling more efficient graph traversal and retrieval of relevant information across multiple documents. This combination improves connectivity between semantically related document chunks, allowing more accurate retrieval and navigation through the multi-document graph.
Regarding claim 4, Zhao does not teach the method of claim 1, wherein determining a similarity between each of the plurality of document chunks, further comprises:
generating, by a chunk model, a plurality of embeddings, wherein each embedding corresponds to a document chunk; and determining the similarity between each of the plurality of chunks based on a similarity between each of the corresponding plurality of embeddings.
However, Ghosh teaches wherein determining a similarity between each of the plurality of document chunks, further comprises:
generating, by a chunk model, a plurality of embeddings, wherein each embedding corresponds to a document chunk;
[0023 “In an example embodiment, Bidirectional Encoder Representations from Transformers (BERT) model can be used in step 120. Portions of the text can be vectorized using
the BERT model to represent each portion of the text with a fix length vectors ( or numerical representation) to generate contextual meaning of each sentence" where portion of texts are document chunks];
[0033 “Following are the BERT embeddings determined based by the process….];
[0038] [ 0.6159851 , 0.76033145, 0.345692 ,0.12330166, 0.32752758, -0.08700331 ]], dtype=float32) Matrix Shape =(7,768). That is, 768-dimensional vector
for each sentence" where 7 means separate vectors/embeddings indicating plurality of embeddings];
and determining the similarity between each of the plurality of chunks based on a similarity between each of the corresponding plurality of embeddings.
[0039 “The similarity score matrix that indicates similarity between any two pair of sentences using cosine similarity" where the similarity score matrix is between sentences derived from vectors].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine teachings of Zhao into the teachings of Ghosh because similarities among document chunks could be determined using their corresponding embeddings rather than solely keywords or entity matching, thereby providing a more robust semantic measure of relatedness for graph construction and retrieval. This combination enables semantically related document chunks to be identified, resulting in improved graph connectivity, retrieval accuracy and discovery of relevant information across multiple documents.
Regarding claim 5, Zhao does not teach the method of claim 4, wherein constructing a multi-document graph based on the similarity between each of the plurality of chunks, further comprises: connecting pairs of nodes representing the document chunks determined to be similar with an edge to form the multi-document graph.
But Ghosh teaches the method of claim 4, wherein constructing a multi-document graph based on the similarity between each of the plurality of chunks, further comprises:
connecting pairs of nodes representing the document chunks determined to be similar with an edge to form the multi-document graph.
[0012 “The systems and methods disclosed herein describe generating a connected network graph based on multiple portions of a text such that each portion is a node of the
network graph”];
[0017 “In an example embodiment, each of the multiple portions of the text ( e.g. sentence 1, sentence 2, sentence 3 .......... sentence n) can be a node of the connected network
graph. So, for a text with n portions, there can be a total of n number of nodes. For n number of nodes, there would be total (n(n-1)/2) number of edges. Edges can have many
properties based on connectivity structure. For example, strength of an edge can be based on the similarity ( or similarity score) between the two nodes it is connecting”];
[0018 “The method 100 may include a step 120 of determining a similarity score of the multiple nodes of the network graph. The similarity score of a node can be based
on its similarity with other nodes of the network graph.”].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine teachings of Zhao into the teachings of Ghosh because combining them would enable the graph structure to explicitly capture relationships between semantically related chunks and facilitate more efficient retrieval and navigation. This combination allows related documents chunks to be directly connected within graph, improving graph traversal efficiency and increasing the likelihood of retrieving relevant information distributed across multiple documents.
Regarding claim 7, Zhao teaches the method of claim 6, further comprising:
receiving a question to be answered based on the plurality of digital documents;
[0039, line 5 "The search query handler 160 may receive a search query in any suitable form, such as, for example, text input by a user"].
traversing the multi-document graph to obtain a context;
[0026, line 6 "The knowledge graph may be traversed to a suitable maximum depth that may be less than a depth that would result in traversing the entirety of the knowledge graph. During the traversal, all of the entity property triplets and entity relation triplets corresponding to traversed to traversed edges may be retrieved as candidate triplets, as they may be entity property triplets and entity relation triplets that have relevance to the search query due to their proximity in the knowledge graph to nodes that correspond to any of the entities in the top-N
entities and triplets. An LLM may be used to select relevant entity property triplets and entity relation triplets from among the candidate entity property triplets and entity
relation triplets. For example, a prompt that includes the candidate triplets and the search query may be input to the LLM, which may then select the candidate triplets considered
most relevant to the search query"].
and generating an answer to the question based on the context.
[0027 “The document chunks of the determined number of document chunks and the relevant entity property triplets and entity relation triplets may be sent as a response to the
received search query. The top-K document chunks and the relevant entity property triplets and entity relation triplets may be returned as the results of the search query and sent
to any suitable computing device or system”]; [0034 “The computing device 100 may include a large language model (LLM) 130.The LLM 130 may be any suitable combination of hardware and software on the computing device 100 for implementing a large language model
that may be trained in any suitable manner to process natural language prompts and generate appropriate text output based on the prompts”].
Regarding claim 8. Zhao teaches
The method of claim 1, wherein the context includes portions of text corresponding to a plurality of traversed nodes in the multi-document graph.
[0025 line 6, “A number of document chunks based on frequency counts of the number of times the document chunks are linked to the entities, entity property triplets and entity
relation triplets most similar to the keywords of the search query may be determined.
The document chunks that are linked to the top-N entities, entity property triplets and entity
relation triplets corresponding to nodes and edges of the knowledge graph may have the frequency with which the document chunks are linked to any of the top-N entities,
entity property triplets and entity relation triplets counted, resulting in a frequency count for each document chunk" where document chunks are portions of text.].
[0026 line 9 "The knowledge graph may be traversed starting from each node that represent one of the entities in the top-N entities, entity property triplets, and entity relation triplets. The knowledge graph may be traversed to a suitable maximum depth that may be less than a depth that would result in traversing the entirety of the knowledge graph. During the traversal, all of the entity property triplets and entity relation triplets corresponding to traversed to traversed edges may be retrieved as candidate triplets, as they may be entity property
triplets and entity relation triplets that have relevance to the search query due to their proximity in the knowledge graph to nodes that correspond to any of the entities in the top-N
entities and triplets"].
Regarding claim 9, ZHAO teaches
a memory component; and
[0066” The computer (e.g., user computer, enterprise computer, etc.) 20 includes a bus 21 which interconnects major components of the computer 20, such as a central processor 24, a memory 27 (typically RAM, but which may also include ROM, flash RAM, or the like), …”]
a processing device coupled to the memory component, the processing device to perform operations comprising:
[0067 “The bus 21 enable data communication between the central processor 24 and the memory 27, which may include read-only memory (ROM) or flash memory (neither shown), and random access memory (RAM) (not shown), as previously noted”].
obtaining a plurality of digital documents;
[0016, line 5 "Documents chunks may be generated from a group of documents"];
dividing the plurality of digital documents into a plurality of document chunks;
[0032 "The computing device 100 may include a chunk generator 110. The chunk generator 110 may be any suitable combination of hardware and software on the computing device 100
that may generate document chunks from documents by dividing documents, such as documents 171, into document chunks, such as document chunks 172"].
determining a similarity between each of the plurality of document chunks;
[0016 "A number of document chunks based on frequency counts of the links from the entity property triplets and entity relation triplets to the document chunks linked to the entity property triplets and entity relation triplets most similar to the keywords of the search query may be determined”].
However, Zhao does not teach constructing a multi-document graph based on the similarity between each of the plurality of document chunks, wherein the multi-document graph includes a plurality of nodes representing the plurality of document chunks.
But Ghosh teaches
constructing a multi-document graph based on the similarity between each of the plurality of document chunks, wherein the multi-document graph includes a plurality of nodes representing the plurality of document chunks.
[0012 “The systems and methods disclosed herein describe generating a connected network graph based on multiple portions of a text such that each portion is a node of the
network graph. A similarity score of the multiple nodes can be determined and a centrality of each node can be measured using graph centrality. The nodes can be ranked based on the
measured centrality and a summary of the text can be generated by using the top ranked nodes”].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine teachings of Zhao into the teachings of Ghosh because Ghosh teaches constructing a graph in which text portions are represented as nodes and connected according to similarity scores. Combining this would facilitate efficient identification and retrieval of related document chunks across a collection of documents.
Regarding claim 10, Zhao teaches the system of claim 9, wherein the operation of determining a similarity between each of the plurality of document chunks, further comprises: determining one or more keywords associated with each document chunk.
Claim 10 is rejected for the same reasons as claim 2.
Regarding claim 11, Zhao in view of Ghosh does teach the system of claim 10, wherein the operation of constructing a multi-document graph based on the similarity between each of the plurality of chunks, further comprises: connecting pairs of nodes representing the document chunks with at least one shared keyword with an edge to form the multi-document graph.
Claim 11 is rejected for the same reasons as claim 3.
Regarding claim 12, Zhao in view of Ghosh does teach the system of claim 10, wherein the operation of determining a similarity between each of the plurality of document chunks, further comprises: generating, by a chunk model, a plurality of embeddings, wherein each embedding corresponds to a document chunk; and determining the similarity between each of the plurality of chunks based on a similarity between each of the corresponding plurality of embeddings.
Claim 12 is rejected for the same reasons as claim 4.
Regarding claim 13, Zhao in view of Ghosh does teach The system of claim 12, wherein the operation of constructing a multi-document graph based on the similarity between each of the plurality of chunks, further comprises: connecting pairs of nodes representing the document chunks determined to be similar with an edge to form the multi-document graph.
Claim 13 is rejected for the same reasons as claim 5.
Regarding claim 15, Zhao teaches the system of claim 14, further comprising: receiving a question to be answered based on the plurality of digital documents; traversing the multi-document graph to obtain a context; and generating an answer to the question based on the context.
Claim 15 is rejected for the same reasons as claim 7.
Regarding claim 16, Zhao teaches The system of claim 10, wherein the context includes portions of text corresponding to a plurality of traversed nodes in the multi-document graph.
Claim 16 is rejected for the same reasons as claim 8.
Regarding claim 17, Zhao teaches
A non-transitory computer-readable medium storing executable instructions, which when executed by a processing device, cause the processing device to perform operations comprising:
[0071 line 4 “Implementations also may be implemented in the form of a computer program product having computer program code containing instructions implemented in non-transitory and/or tangible media, such as floppy diskettes, CD-ROMs, hard drives, USB (universal serial bus) drives, or any other machine readable storage medium, wherein, when the computer program code is loaded into and executed by a computer,..”].
receiving a question to be answered based on the plurality of digital documents;
[0039, line 5 "The search query handler 160 may receive a search query in any suitable form, such as, for example, text input by a user"].
traversing the multi-document graph to obtain a context;
[0026, line 6 " The knowledge graph may be traversed to a suitable maximum depth that may be less than a depth that would result in traversing the entirety of the knowledge graph. During the traversal, all of the entity property triplets and entity relation triplets corresponding to traversed to traversed edges may be retrieved as candidate triplets, as they may be entity property triplets and entity relation triplets that have relevance to the search query due to their proximity in the knowledge graph to nodes that correspond to any of the entities in the top-N
entities and triplets. An LLM may be used to select relevant entity property triplets and entity relation triplets from among the candidate entity property triplets and entity
relation triplets. For example, a prompt that includes the candidate triplets and the search query may be input to the LLM, which may then select the candidate triplets considered
most relevant to the search query."].
and generating an answer to the question based on the context.
[0027 “The document chunks of the determined number of document chunks and the relevant entity property triplets and entity relation triplets may be sent as a response to the
received search query. The top-K document chunks and the relevant entity property triplets and entity relation triplets may be returned as the results of the search query and sent
to any suitable computing device or system”]; [0034 “The computing device 100 may include a large language model (LLM) 130.The LLM 130 may be any suitable combination of hardware and software on the computing device 100 for implementing a large language model
that may be trained in any suitable manner to process natural language prompts and generate appropriate text output based on the prompts”].
However, Zhao does not teach constructing a multi-document graph corresponding to a plurality of digital documents, wherein each node of the multi-document graph represents a document chunk of one of the digital documents and wherein each edge represents a similarity between nodes;
But Ghosh teaches constructing a multi-document graph corresponding to a plurality of digital documents, wherein each node of the multi-document graph represents a document chunk of one of the digital documents and wherein each edge represents a similarity between nodes;
[0012 “The systems and methods disclosed herein describe generating a connected network graph based on multiple portions of a text such that each portion is a node of the
network graph. A similarity score of the multiple nodes can be determined and a centrality of each node can be measured using graph centrality. The nodes can be ranked based on the
measured centrality and a summary of the text can be generated by using the top ranked nodes”]; [0017 “In an example embodiment, each of the multiple portions of the text ( e.g. sentence 1, sentence 2, sentence 3 .......... sentence n) can be a node of the connected network
graph. So, for a text with n portions, there can be a total of n number of nodes. For n number of nodes, there would be total (n(n-1)/2) number of edges. Edges can have many
properties based on connectivity structure. For example, strength of an edge can be based on the similarity ( or similarity score) between the two nodes it is connecting”];
[0018 “The method 100 may include a step 120 of determining a similarity score of the multiple nodes of the network graph. The similarity score of a node can be based
on its similarity with other nodes of the network graph.”].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine teachings of Zhao into the teachings of Ghosh because combining them would enable the graph structure to explicitly capture relationships between semantically related chunks and facilitate more efficient retrieval and navigation. This combination allows related documents chunks to be directly connected within graph, improving graph traversal efficiency and increasing the likelihood of retrieving relevant information distributed across multiple documents.
Regarding claim 18, Zhao teaches the non-transitory computer-readable medium of claim 17, storing instructions that further cause the processing device to perform operations comprising: determining a starting node of the multi-document graph based on a similarity to the question.
[0026 “Relevant entity property triplets and entity relation triplets may be determined by traversing the knowledge graph to a specified depth starting at the nodes in the number
of nodes and edges of the knowledge graph that include entities most similar to the keywords of the search query. The knowledge graph may be traversed starting from each
node that represent one of the entities in the top-N entities, entity property triplets, and entity relation triplets.”].
Regarding claim 20, Zhao teach the non-transitory computer-readable medium of claim 17, wherein the operation of constructing a multi-document graph corresponding to the plurality of digital documents, further comprises:
dividing the plurality of digital documents into a plurality of document chunks; and
[0032 "The computing device 100 may include a chunk generator 110. The chunk generator 110 may be any suitable combination of hardware and software on the computing device 100
that may generate document chunks from documents by dividing documents, such as documents 171, into document chunks, such as document chunks 172"].
determining a similarity between each of the plurality of document chunks, wherein the similarity is based on keyword similarity or embedding similarity.
[0016” A search query including keywords may be received number of nodes and
edges of the knowledge graph that include entities, entity property triplets, and entity relation triplets most similar to the keywords of the search query may be determined.
A number of document chunks based on frequency counts of the links from the entity property triplets and entity relation triplets to the document chunks linked to the entity property triplets and entity relation triplets most similar to the keywords
of the search query may be determined….The document chunks of the determined number of document chunks and the relevant entity property triplets and entity relation triplets may be
sent as a response to the received search query” where the keywords are associated with document chunks by identifying entities and triplets most similar to keywords and determining the document chunks linked to those entities and triplets which represent semantic keywords concepts extracted from document chunks and serve as semantic keywords for identifying relevant chunks].
Claim [ 6, 14 ] are rejected under 35 U.S.C. 103 as being
unpatentable over Zhao (US-20260030517-A1) in view of Ghosh (US-20220277035-A1) and in
further view of Sundaresan(US-6606620-B1).
Regarding claim 6, Zhao in view of Ghosh do not teach the method of claim 1, further comprising:
determining structure content of the plurality of digital documents, wherein the structure content includes at least one of page numbers and table content; and
adding one or more structure nodes to the multi-document graph corresponding to the structure content.
However, Sundaresan teaches determining structure content of the plurality of digital documents, wherein the structure content includes at least one of page numbers and table content; and
[Column 10, lines 3- 9 “ With reference to FIG. 7, the training process begins by introducing a large number of documents 352 with known class labels into the classifier 10.
For each document 352, the classifier 10 creates a structured vector, in particular, extracting textual and structural information from the document 352 and constructing a directed graph containing the information. This process is known as vectorization" where vectorization means converting the document's text into a directed graph of information nodes].
adding one or more structure nodes to the multi-document graph corresponding to the structure content.
[Column 10, lines 54-56 " The vectorization sub-module 405 begins its function by parsing the document 352 to create a parse tree of the document 352" where creating a parse tree of the document implies adding nodes since tree is made up by nodes];
[Column 6, lines 32-36 " Tree: A hierarchical structure which is made up by nodes. Nodes are connected by edges from one node (parent) to another (child). A single node at apex of the tree is known as the root node, while the terminus of a path in the opposite direction is a leaf”].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine teachings of Zhao in view of Ghosh with the teachings of Sundaresan because it would provide a graph that captures both semantic relationships among document chunks and structural relationships within the documents, thereby improving retrieval, navigation, and contextual understanding of information distributed across multiple documents.
Regarding claim 14, Zhao in view of Ghosh and in further view of Sundaresan do teach
the system of claim 10, wherein the processing device performs further operations comprising: determining structure content of the plurality of digital documents; and adding one or more structure nodes to the multi-document graph corresponding to the structure content
Claim 14 is rejected for the same reasons as claim 6.
Claim [ 19 ] are rejected under 35 U.S.C. 103 as being
unpatentable over ZHAO (US-20260030517-A1) in view of GHOSH (US-20220277035-A1) and in
view of Trivedi (Trivedi, Harsh, et al. "Interleaving retrieval with chain-of-thought reasoning for knowledge-intensive multi-step questions." Proceedings of the 61st annual meeting of the association for computational linguistics (volume 1: long papers). 2023.).
Regarding claim 19, Zhao in view of Ghosh do teach
The non-transitory computer-readable medium of claim 17, wherein the operation of traversing the multi-document graph to obtain a context, further comprises: identifying a search space including nodes connected to a current node;
Zhao teaches [0071 “Implementations also may be implemented in the form of a computer program product having computer program code containing instructions implemented in non-transitory and/or tangible media, such as floppy diskettes, CD-ROMs, hard drives, USB (universal serial bus) drives, or any other machine readable storage medium, wherein, when the computer program code is loaded into and executed by a computer”]; [0026"The knowledge graph may be traversed starting from each node that represent one of the entities in the top-N entities, entity property triplets, and entity relation triplets. The knowledge graph may be traversed to a suitable maximum depth that may be less than a depth that would result in traversing the entirety of the knowledge graph. During the traversal, all of the entity property triplets and entity relation triplets corresponding to traversed to traversed edges may be retrieved as candidate triplets, as they may be entity property triplets and entity relation triplets that have relevance to the search query due to their proximity in the knowledge graph to nodes that correspond to any of the entities in the top-N entities and triplets" where nodes and edges connected to the current traversal node form the search space].
and selecting a next node from the search space
[0026 “Relevant entity property triplets and entity relation triplets may be determined by traversing the knowledge graph to a specified depth starting at the nodes in the number
of nodes and edges of the knowledge graph that include entities most similar to the keywords of the search query. The knowledge graph may be traversed starting from each
node that represent one of the entities in the top-N entities, entity property triplets, and entity relation triplets. The knowledge graph may be traversed to a suitable maximum
depth that may be less than a depth that would result in traversing the entirety of the knowledge graph” where entity property triplets and entity relation triplets are part of the node selection and relevant entity property triplets and relation triplets are being determined (next chunk)].
However, Zhao in view of Ghosh do not teach
predicting, using a passage generator, a next text chunk based on the question and a current context including already traversed nodes;
However, Trivedi teaches
predicting, using a passage generator, a next text chunk based on the question and a current context including already traversed nodes;
[Page 2, lines, 27-35 “We begin by retrieving a base set of paragraphs using the question as a query. Subsequently, we alternate between the following two steps: (i) extend CoT: use the question, the paragraphs collected thus far, and the CoT sentences generated thus far to generate the next CoT sentence; (ii) expand retrieved information: use the last CoT sentence as a query to retrieve additional paragraphs to add to the collected set”];
[Page 3., lines 17-28, 3.1 “Our proposed retriever method, IRCoT, can be instantiated from the following three ingredients: (i) a base retriever that can take a query and return a given number of paragraphs from a corpus or knowledge source; (ii) a language model with zero/few-shot Chain-of-Thought (CoT) generation capabilities; and (iii) a small number of annotated questions with reasoning steps explaining how to arrive at the answer in natural language (chain of thoughts) and a set of paragraphs from the knowledge source that collectively support the reasoning chain and the answer” where the LLM acts as an iterative query generator to guide an external paragraph retriever.];
[Page 3, lines 29-36, 3.1 ” The overview of IRCoT is given in Fig. 2. We first gather a base set of paragraphs by retrieving K paragraphs using the question Q as the query. Then, we interleave two steps (reason and retrieve) iteratively until the termination criterion is met. The retrieval-guided reasoning step (“Reason”) generates the next CoT sentence using the question, the paragraphs collected thus far, and the CoT sentences generated thus far” where the question is used to generate CoT sentences( next text chunk) , where the paragraphs collected thus far(current context), and CoT sentences generated thus far(already traversed nodes) are also used to generate the next chunk (CoT sentences)].
based on the predicted next text chunk
[3.1 ” The overview of IRCoT is given in Fig. 2. We first gather a base set of paragraphs by retrieving K paragraphs using the question Q as the query. Then, we interleave two steps (reason and retrieve) iteratively until the termination criterion is met. The retrieval-guided reasoning step (“Reason”) generates the next CoT sentence using the question, the paragraphs collected thus far, and the CoT sentences generated thus far, and the CoT sentences generated thus far to generate the next CoT sentence; (ii) expand retrieved information: use the last CoT sentence as a query to retrieve additional paragraphs to add to the collected set.” where the CoT sentences play the role of the predicted next chunk].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine teachings of Zhao in view of Ghosh with the teachings of Trivedi because generating a subsequent text segment responsive to user’s question and in turn, guide retrieval of additional related information. This improves the relevance of retrieved information and supports more effective multi-hop question answering.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SHEZA ABDUL AZIZ whose telephone number is (571)272-9610. The examiner can normally be reached Monday-Friday 7:30am-5pm Alternate Fridays off.
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, Daniel Washburn can be reached at (571) 272-5551. 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.
/SHEZA ABDUL AZIZ/Examiner, Art Unit 2657
/DANIEL C WASHBURN/Supervisory Patent Examiner, Art Unit 2657
Prosecution Insights