DETAILED ACTION
Notice of Pre-AIA or AIA Status
This communication is in response to the Amendments and Arguments filed on 5/25/2026.
Claims 1, 3, 4, 7, and 8 are pending and have been examined.
All previous objections / rejections not mentioned in this Office Action have been withdrawn by the examiner.
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 .
Response to Amendments
Regarding the Applicant’s arguments for the rejections under 35 U.S.C. § 103, applicant has amended claim 1 and 8. Applicant asserts that the prior art references alone or in combination does not teach “generate a first taxonomy graph for the first document based on said first plurality of elements and a second taxonomy graph for the second document based on said second plurality of elements, said first taxonomy graph including first nodes representing elements from said first plurality of elements and first edges representing taxonomical relationships between said first nodes, said second taxonomy graph including second nodes representing elements from said second plurality of elements and second edges representing taxonomical relationships between said second nodes, said first taxonomy graph and said second taxonomy graph being generated by (i) ranking said first plurality of elements and said second plurality of elements using a link analysis function and (ii) forming said first taxonomy graph and second taxonomy graph from elements in said first plurality of elements and said second plurality of elements that exceed a threshold rank; generate a plurality of graph embeddings using a graph embedding model, the plurality of graph embeddings including (i) at least one first graph embedding for at least one element of said first document determined by inputting the first taxonomy graph into the graph embedding model and (ii) at least one second graph embedding for at least one element of said second document determined by inputting the second taxonomy graph into the graph embedding model”. Examiner respectfully disagrees. During patent examination, pending claims must be “given their broadest reasonable interpretation consistent with the specification.” MPEP 2111. Also, claims "must particularly point out and distinctly claim the invention." MPEP 2173. First, the terms “taxonomy graph”, “taxonomical relationship”, “link analysis function”, “threshold rank”, and “graph embedding” are not specifically defined or described in the as-filed disclosure. Therefore, examiner interprets each term broadly according to their plain meaning. MPEP 2111.01. Second, the prior art references in combination teach the disputed claim limitations. Prior art references are analogous art as the claimed invention because the prior art references Atasu, Jacquet, and Nefedov teach the mapping of elements in documents according to string based similarities. Atasu P0009, Jacquet P0013, Nefedov P0024. Prior art Atasu teaches the concept of finding similarities according to features that correspond to embedding features. Atasu P0037. The embedding features can be words that are mapped onto a vector space. Id. The limitations brought forth by applicant are not specifically taught by Atasu. Nefedov, however, teaches the claim limitations. Nefedov, teaches that embedding for finding similarities can be embeddings created from taxonomy graph with nodes representing elements and edges representing relationship between the nodes. Nefedov P0024, P0047. Nefedov, further teaches the ranking of elements using a link analysis function and forming taxonomy graph of elements that exceed a threshold rank. Nefedov P0042, P0046. Lastly, Nefedov teaches generation of graph embedding or a taxon embedding by the taxonomy extraction engine. Nefedov P0047. Therefore, the claims as currently recited does not overcome the 35 U.S.C. § 103 obviousness rejection.
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.
Claims 1, 3, 4, 7, and 8 are rejected under 35 U.S.C. 103 as being unpatentable over Atasu et al. (U.S. PG Pub No. 20210303609), hereinafter Atasu, in view of Jacquet et al. (U.S. PG Pub No. 20150127323), hereinafter Jacquet, in further view of Nefedov et al. (U.S. PG Pub No. 20230074771), hereinafter Nefedov.
Regarding claims 1 and 8, Atasu teaches:
(Claim 1) A control unit to map at least one element present in a plurality of documents, wherein said plurality of documents includes a first document and a second document, said control unit being configured to: (P0087, Computerized systems and devices can be suitably designed for implementing embodiments of the present invention as described herein.; P0009, The proposed solution allows the overall similarity results to be explained (i.e., interpreted), e.g., by identifying specific pairs of words, sentences, paragraphs, or chapters that most contributes to the overall similarity between the compared documents.; P0018, The first features and the second features are mapped onto a vector space and thereby correspond to vectors of this vector space. There, the method may further comprise computing said similarities (i.e., ground similarities) based on vectors corresponding to the first features and vectors corresponding to the second features.; Fig. 1, Illustrates how the similarity between two sentences can be measured based on words mapped onto a vector space.)
(Claim 8) A method of mapping at least one element present in a plurality of documents by a control unit, wherein said plurality of documents includes a first document and a second document, said method comprising: (P0008, present invention is embodied as a computer-implemented method.; P0009, The proposed solution allows the overall similarity results to be explained (i.e., interpreted), e.g., by identifying specific pairs of words, sentences, paragraphs, or chapters that most contributes to the overall similarity between the compared documents.; P0018, The first features and the second features are mapped onto a vector space and thereby correspond to vectors of this vector space. There, the method may further comprise computing said similarities (i.e., ground similarities) based on vectors corresponding to the first features and vectors corresponding to the second features.; Fig. 1, Illustrates how the similarity between two sentences can be measured based on words mapped onto a vector space.)
identify a first plurality of elements in said first document and a second plurality of elements in said second document by (i) creating a corresponding dependency parse tree structure for each of a plurality of considered phrases in said first document and said second document, (ii) identifying said first plurality of elements and said plurality of elements as nouns in the plurality of considered phrases; (P0037, “Features” means any type of components (e.g., vector components) of the datasets. Such components may possibly be extracted features or be features identified using any extraction or parsing method (e.g., to extract pixel values of an image or other types of components, such as words or sentences of a text document). Such features may also correspond to embedding features (e.g., words mapped onto a vector space).)
identify at least one semantic bridge between said first document and said second document using string based similarities; (P0009, The proposed solution allows the overall similarity results to be explained (i.e., interpreted), e.g., by identifying specific pairs of words, sentences, paragraphs, or chapters that most contributes to the overall similarity between the compared documents.)
correlate said plurality of graph embeddings of said at least one element of said first document and said at least one element of said second document; and (P0041, Distance is also a measure of similarity, or on angles between vectors onto which the features are mapped. Eventually, though, similarities need be obtained, to solve the similarity-flow problem.; P0042, Pair contributions to the overall similarity are computed, in order to obtain contributive elements. The computation of the pair contributions makes use of the ground similarities between the first features and the second features. )
map said at least one element of said first document to said at least one element of said second document based on the correlated plurality of graph embeddings using a vector function. (P0041, Several methods are known, which allow distances (or angles between vectors) to be simply transformed in similarities (e.g., using the Euclidean distance or the cosine of the angles formed between such vectors).)
Atasu does not specifically teach:
identify a first plurality of elements in said first document and a second plurality of elements in said second document by (i) creating a corresponding dependency parse tree structure for each of a plurality of considered phrases in said first document and said second document, (ii) identifying said first plurality of elements and said plurality of elements as nouns in the plurality of considered phrases;
generate a first taxonomy graph for the first document based on said first plurality of elements and a second taxonomy graph for the second document based on said second plurality of elements, said first taxonomy graph including first nodes representing elements from said first plurality of elements and first edges representing taxonomical relationships between said first nodes, said second taxonomy graph including second nodes representing elements from said second plurality of elements and second edges representing taxonomical relationships between said second nodes, said first taxonomy graph and said second taxonomy graph being generated by (i) ranking said first plurality of elements and said second plurality of elements using a link analysis function and (ii) forming said first taxonomy graph and second taxonomy graph from elements in said first plurality of elements and said second plurality of elements that exceed a threshold rank;
generate a plurality of graph embeddings using a graph embedding model, the plurality of graph embeddings including (i) at least one first graph embedding for at least one element of said first document determined by inputting the first taxonomy graph into the graph embedding model and (ii) at least one second graph embedding for at least one element of said second document determined by inputting the second taxonomy graph onto the graph embedding model;
Jacquet, however, teaches:
identify a first plurality of elements in said first document and a second plurality of elements in said second document by (i) creating a corresponding dependency parse tree structure for each of a plurality of considered phrases in said first document and said second document, (ii) identifying said first plurality of elements and said plurality of elements as nouns in the plurality of considered phrases; (P0063, The parser may comprise any suitable syntactic dependency parser which is configured for generating a parse tree. During parsing of the document, the parser annotates the text strings of the document with tags (labels) which correspond to grammar rules, such as lexical rules and syntactic and/or semantic dependency rules. The lexical rules define features of terms such as words and multi-word expressions. The lexical rules may include assigning parts of speech to terms in the text, such as noun, verb, etc., from a predefined set of parts of speech to be recognized.)
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to create dependency parse tree to identify elements. It would have been obvious to combine the references because creating a dependency parse tree is a known method to yield a predictable result of identifying nouns and their link to verbs and other words. (Jacquet P0063)
Atasu in view of Jacquet does not specifically teach:
generate a first taxonomy graph for the first document based on said first plurality of elements and a second taxonomy graph for the second document based on said second plurality of elements, said first taxonomy graph including first nodes representing elements from said first plurality of elements and first edges representing taxonomical relationships between said first nodes, said second taxonomy graph including second nodes representing elements from said second plurality of elements and second edges representing taxonomical relationships between said second nodes, said first taxonomy graph and said second taxonomy graph being generated by (i) ranking said first plurality of elements and said second plurality of elements using a link analysis function and (ii) forming said first taxonomy graph and second taxonomy graph from elements in said first plurality of elements and said second plurality of elements that exceed a threshold rank;
generate a plurality of graph embeddings using a graph embedding model, the plurality of graph embeddings including (i) at least one first graph embedding for at least one element of said first document determined by inputting the first taxonomy graph into the graph embedding model and (ii) at least one second graph embedding for at least one element of said second document determined by inputting the second taxonomy graph onto the graph embedding model;
Nefedov, however, teaches:
generate a first taxonomy graph for the first document based on said first plurality of elements and a second taxonomy graph for the second document based on said second plurality of elements, said first taxonomy graph including first nodes representing elements from said first plurality of elements and first edges representing taxonomical relationships between said first nodes, said second taxonomy graph including second nodes representing elements from said second plurality of elements and second edges representing taxonomical relationships between said second nodes, said first taxonomy graph and said second taxonomy graph being generated by (i) ranking said first plurality of elements and said second plurality of elements using a link analysis function and (ii) forming said first taxonomy graph and second taxonomy graph from elements in said first plurality of elements and said second plurality of elements that exceed a threshold rank; (P0039, NLP engine may also be configured to generate a features-domain co-occurrence matrix for the processed corpus of documents. Co-occurrence refers to the occurrence of two or more features. Examples of such features include terms or words, titles, sections, labels, topics, sentences, paragraphs, concepts, and/or the like, as well as embedding vectors thereof in the same context, for example, in the same document of a corpus of documents.; P0040, The clustering engine may apply the clustering algorithm to the features-domain co-occurrence matrix to generate a graph of nodes representing the features and the edges representing the context of the co-occurrence matrix (the context being occurring or appearing in the same document of the corpus of documents, for example).; P0042, The taxonomy extraction engine is configured to extract taxonomies from the graphs by identifying groups of nodes of the graphs that may be viewed as clusters, i.e., groups of nodes that are more densely connected within the cluster than outside the cluster. To accomplish this goal, in various embodiments, the taxonomy extraction engine can apply a community detection algorithm to the graph to identify or detect a community structure of nodes, e.g., groups of nodes that may be viewed or determined to be clusters.; P0046, the taxonomy extraction engine 124 can adaptively prune the clusters hierarchically to form clusters at each hierarchical level having modularity exceeding the minimum modularity threshold.; P0048, The taxonomy extraction engine, to extract a hierarchical taxonomy composed of taxons, where each taxon of the taxon graph represents a cluster of nodes of the graph.; P0048, to extract a hierarchical taxonomy composed of taxons, where each taxon of the taxon graph represents a cluster of nodes of the graph.)
generate a plurality of graph embeddings using a graph embedding model, the plurality of graph embeddings including (i) at least one first graph embedding for at least one element of said first document determined by inputting the first taxonomy graph into the graph embedding model and (ii) at least one second graph embedding for at least one element of said second document determined by inputting the second taxonomy graph onto the graph embedding model; (P0047, The taxonomy extraction engine can then analyze the extracted taxonomy to generate a features presentation of the taxons, which for each taxon may include but is not limited to an identification of the taxon, features of the taxon (e.g., feature name, weight (e.g., the number of documents in the taxon), feature embedding, etc.), a taxon embedding in vector space, a human-readable name of the taxon, and/or the like. A taxon embedding can be a graph embedding.)
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to generate a taxonomy graph representing taxonomical relationships between nodes representing elements. It would have been obvious to combine the references because the use of taxonomy graph has the benefit of rapid labelling and classification of documents according to respective taxonomies. (Nefedov P0022)
Regarding claim 3 Atasu in view of Jacquet and further view of Nefedov teach claim 1:
Atasu does not specifically teach:
wherein said control unit is further configured to identify tags in each of said plurality of considered phrases.
Jacquet, however, teaches:
wherein said control unit is further configured to identify tags in each of said plurality of considered phrases. (P0063, The parser may comprise any suitable syntactic dependency parser which is configured for generating a parse tree. During parsing of the document, the parser annotates the text strings of the document with tags (labels) which correspond to grammar rules, such as lexical rules and syntactic and/or semantic dependency rules. The lexical rules define features of terms such as words and multi-word expressions. The lexical rules may include assigning parts of speech to terms in the text, such as noun, verb, etc., from a predefined set of parts of speech to be recognized.; P0066, Extraction of triples. For every path, all the occurrences of nouns that instantiate each of its two slots are logged, as well as the frequency of these instantiations.)
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to identify tags and detect elements. It would have been obvious to combine the references because identifying a triple or path from a sentence in the parse tree makes it easy to calculate similarity between the sentences. (Jacquet P0032)
Regarding claim 4 Atasu in view of Jacquet and further view of Nefedov teach claim 3.
Atasu does not specifically teach:
wherein said control unit is further configured to create respective graphs between said first plurality of elements and said second plurality of elements using undirected and unidentified edges, to which the link analysis function is applied.
Jacquet, however, teaches:
wherein said control unit is further configured to create respective graphs between said first plurality of elements and said second plurality of elements using undirected and unidentified edges, to which the link analysis function is applied. (P0063, The parser may comprise any suitable syntactic dependency parser which is configured for generating a parse tree. … The lexical rules may include assigning parts of speech to terms in the text, such as noun, verb, etc., from a predefined set of parts of speech to be recognized. The dependency rules include rules for identifying dependency relations between terms, … The parser outputs for each text string, such as a sentence, a parse tree in which nouns are linked to the verbs and other words where a dependency has been identified.)
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to create a graph between node or elements. It would have been obvious to combine the references because identifying a triple or path from a sentence in the parse tree makes it easy to calculate similarity between the sentences. (Jacquet P0032)
Regarding claim 7 Atasu in view of Jacquet and further view of Nefedov teach claim 1.
Atasu further teaches:
wherein said plurality of documents are chosen from a group of documents including manuals, legal documents, contracts. (P0068, Relevance of a query document with respect to a reference document. It may for instance be useful to identify news, blogs, or analyst reports. [A person of ordinary skill in the art can identify analyst report to encompass reports on legal compliance or legal risk assessment.])
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DANIEL WONSUK CHUNG whose telephone number is (571)272-1345. The examiner can normally be reached Monday - Friday (7am-4pm)[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, PIERRE-LOUIS DESIR can be reached at (571)272-7799. 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.
/DANIEL W CHUNG/Examiner, Art Unit 2659
/PIERRE LOUIS DESIR/Supervisory Patent Examiner, Art Unit 2659
Prosecution Insights