DETAILED ACTION
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 .
This Office action is in response to the RCE, filed on 2/9/2026, and the amendment, arguments and remarks, filed on 1/9/2026, in which claim(s) 1-9, 11-18 and 20 is/are presented for further examination.
Claims(s) 1, 11 and 20 has/have been amended.
Claim(s) 10 and 19 has/have been cancelled.
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant’s submission filed on 1/9/2026 has been entered.
Response to Amendment
Applicant’s amendment(s) to claim(s) 1, 11 and 20 has/have been accepted.
Note: For clarity of the record, claim 1’s status is denoted as “Original”; however, it should be “Currently Amended”.
Response to Arguments
Applicant’s arguments with respect to claim(s) 1-9, 11-18 and 20, filed on 1/9/2026, have been fully considered but they are not persuasive.
Applicant argues that claim(s) 1-20 are patent eligible, see the bottom of page 7 of applicant’s remarks, filed on 1/9/2026.
The examiner respectfully disagrees. Please see the updated 35 U.S.C. 101 rejection(s) in light of applicant’s amended claim(s) below. Additionally, using the BRI, the claim(s) can be interpreted as an abstract idea without significantly more. Applicant argues that the claim(s) improve the functioning of a computer; however, nothing in the claim language recites that. The current claim(s) merely recite(s) creating a taxonomy. If the taxonomy is somehow used by a computer and claimed as such then perhaps the claim(s) could be patent eligible; however, as currently presented the claim(s) is/are not patent eligible.
Applicant’s arguments with respect to the rejection(s) of claim(s) 1, 10, 11, 19 and 20, under 35 U.S.C. 102, and claim(s) 2-9 and 12-18, under 35 U.S.C. 103, see page 8 to page 10 of applicant’s remarks, filed on 1/9/2026, have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
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(s) 1-9, 11-18 and 20 is/are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Claim(s) 1, 11 and 20 recite(s) the limitation(s) of:
“performing cluster analysis on documents associated with a leaf node of the hierarchical taxonomy to determine a plurality of first clusters, each cluster of the plurality of first clusters being associated with one or more documents associated with the leaf node;” (mental process with the aid of pen and paper)
“determining one or more topics associated with the documents in each cluster of the plurality of first clusters;” (mental process)
“performing cluster analysis on the topics to determine a plurality of second clusters, each cluster of the plurality of second clusters being associated with one or more of the topics;” (mental process with the aid of pen and paper)
“determining a name for each cluster of the plurality of second clusters;” (mental process)
“expanding the hierarchical taxonomy by adding a first level below the lowest level comprising a plurality of first nodes, the plurality of first nodes comprising the determined name for each cluster of the plurality of second clusters;” (mental process with the aid of pen and paper)
“further expanding the hierarchical taxonomy by adding a second level below the first level comprising a plurality of second nodes, the plurality of second nodes comprising the determined topics associated with the documents in each cluster of the plurality of first clusters;” (mental process with the aid of pen and paper)
The limitation(s), as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind, see the discussion above. That is, nothing in the claim(s) preclude(s) the steps from practically being performed in the mind. The mere nominal recitation of “a hierarchical taxonomy”, “a corpus of documents” and “a leaf node” in claim 1, as well as “a processing device” and “a non-transitory, processor-readable storage medium” in claim 11 and “a non-transitory, computer-readable storage medium”, “a computer” and “a processing device” in claim 20 do(es) not take the claim limitation(s) out of the mental processes grouping. The mere nominal recitation of a generic memory and a generic processor in claim(s) 11 and 20 do(es) not take the claim limitation(s) out of the mental processes grouping. Thus, the claim(s) recite(s) a mental process.
This judicial exception is not integrated into a practical application. In claim 1, there are no additional elements beyond the identified abstract idea. The recitation(s) of “a processing device” and “a non-transitory, processor-readable storage medium” in claim(s) 11 and 20 is/are no more than mere instructions to apply the exception using generic computer components.
The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements when considered both individually and as an ordered combination do not amount to significantly more than the abstract idea. The recitation(s) of “a hierarchical taxonomy”, “a corpus of documents”, “a leaf node”, “a processing device” “a non-transitory, processor-readable storage medium”, “a non-transitory, computer-readable storage medium”, “a computer” and “a processing device” do(es) nothing more than apply the exception with generic, off-the-shelf computer component(s).
Regarding claim(s) 2, 3, 4, 7, 12, 13 and 16, the claim(s) recite(s) the limitation(s) of “using natural language processing to determine vectorizations associated with the documents; and performing cluster analysis of the vectorizations associated with the documents” (mental process - NLP processing is so broadly recited to be a mentally performable process. One can mentally evaluate natural language and determine some vector valuation) in claims 2 and 12, “using natural language processing to determine vectorizations of portions of each of the documents; and performing cluster analysis of the vectorizations” (mental process - NLP processing is so broadly recited to be a mentally performable process. One can mentally evaluate natural language and determine some vector valuation) in claim 3, “determining the plurality of first clusters based on a cosine similarity between the vectorizations associated with the documents” (mental process with the aid of pen and paper) in claims 4 and 13, “using natural language processing to determine vectorizations associated with the topics; and performing cluster analysis on the vectorizations” (mental process - NLP processing is so broadly recited to be a mentally performable process. One can mentally evaluate natural language and determine some vector valuation) in claims 7 and 16, the limitation(s), as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind. The claim(s) do(es) not include any further additional elements, and merely adding a further abstract idea to an already ineligible parent claim directed to an abstract idea without significantly more, cannot confer eligibility to the claim(s).
Regarding claim(s) 5, 6, 8, 9, 14, 15, 17 and 18, the claim(s) recite(s) the limitation(s) of “inputting the documents and a prompt into a large language model; and determining the one or more topics based on an output of the large language model” in claims 5 and 14, “wherein the prompt asks the large language model to generate the one or more topics based on the documents” in claims 6 and 15, “inputting the topics for each cluster of the plurality of second clusters and a prompt into a large language model; and determining the name for each cluster of the plurality of second clusters based on an output of the large language model” in claims 8 and 17, “wherein the prompt asks the large language model to generate the name for each cluster of the plurality of second clusters based on the topics for each cluster” in claims 9 and 18. The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception because the recitation(s) do(es) nothing more than apply the exception with generic, off-the-shelf computer component(s). Note: The large language model (LLM) is explicitly a commercially available product as per [0032] of the specification.
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(s) 1, 11 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Nefedov et al., US 2023/0074771 A1 (hereinafter “Nefedov”) in view of Beller et al., US 2021/0264113 A1 (hereinafter “Beller”).
Claims 1, 11 and 20
Nefedov discloses a method for expanding a hierarchical taxonomy associated with a corpus of documents (Nefedov, [0005], see extracting taxonomies based on hierarchical clustering on graphs related to a corpus of documents and using said taxonomies for classifying documents), the hierarchical taxonomy comprising a plurality of levels with each level comprising one or more nodes (Nefedov, [0005], see the methods comprise applying a community detection algorithm to the graph to detect a plurality of clusters of nodes of the graph at each of hierarchical levels such that, the graph-modularity of plurality of clusters exceeds a minimum modularity threshold at each hierarchical level), the method comprising:
performing cluster analysis on documents associated with a leaf node of the hierarchical taxonomy to determine a plurality of first clusters, each cluster of the plurality of first clusters being associated with one or more documents associated with the leaf node (Nefedov, [0025], see extracting taxonomies for the documents from the graph. For instance, the graph hierarchically clustered [i.e., “cluster analysis”] to form clusters of nodes at different hierarchical levels, for example, by adaptively pruning the graph at each hierarchy until the graph becomes a network of recognizable or identifiable clusters [i.e., “a plurality of first clusters”], and repeating the same procedure for each of the clusters until pruning is no longer possible or acceptable (e.g., causes high level of information loss); Note: That the clusters of nodes are formed from the inputted documents, which means that the cluster(s) is/are associated with the inputted documents);
determining one or more topics associated with the documents in each cluster of the plurality of first clusters (Nefedov, [0024], see co-occurrence matrices for a corpus of documents may be constructed based on similarity among features of the documents that are extracted from the corpus of documents, such as the co-occurrence of words taken from the whole vocabulary over a corpus of documents. … These co-occurrent elements may be seen as dominated “phrases” over all documents. … That is, the nodes represent the words in the corpus of vocabulary and the edges connecting nodes may represent the co-occurrences between the words represented by the connected nodes. Then clustering of the co-occurrence matrix would reveal dominated (and possibly hierarchical) clusters or “topics” over all documents. These “topics” may be used as “labels” for hierarchical clusters of documents described);
performing cluster analysis on the topics to determine a plurality of second clusters, each cluster of the plurality of second clusters being associated with one or more of the topics (Nefedov, [0025], see extracting taxonomies for the documents from the graph. For instance, the graph hierarchically clustered [i.e., “cluster analysis”] to form clusters of nodes at different hierarchical levels, for example, by adaptively pruning the graph at each hierarchy until the graph becomes a network of recognizable or identifiable clusters [i.e., “a plurality of first clusters”], and repeating the same procedure for each of the clusters [i.e., “determine a plurality of second clusters”] until pruning is no longer possible or acceptable (e.g., causes high level of information loss); and Nefedov, [0024], see co-occurrence matrices for a corpus of documents may be constructed based on similarity among features of the documents that are extracted from the corpus of documents, such as the co-occurrence of words taken from the whole vocabulary over a corpus of documents. … These co-occurrent elements may be seen as dominated “phrases” over all documents. … That is, the nodes represent the words in the corpus of vocabulary and the edges connecting nodes may represent the co-occurrences between the words represented by the connected nodes. Then clustering of the co-occurrence matrix would reveal dominated (and possibly hierarchical) clusters or “topics” over all documents. These “topics” may be used as “labels” for hierarchical clusters of documents described);
determining a name for each cluster of the plurality of second clusters (Nefedov, [0024], see co-occurrence matrices for a corpus of documents may be constructed based on similarity among features of the documents that are extracted from the corpus of documents, such as the co-occurrence of words taken from the whole vocabulary over a corpus of documents. … These co-occurrent elements may be seen as dominated “phrases” over all documents. … That is, the nodes represent the words in the corpus of vocabulary and the edges connecting nodes may represent the co-occurrences between the words represented by the connected nodes. Then clustering of the co-occurrence matrix would reveal dominated (and possibly hierarchical) clusters or “topics” over all documents. These “topics” may be used as “labels” [i.e., “name for each cluster”] for hierarchical clusters of documents described);
expanding the hierarchical taxonomy (Nefedov, [0066], see a user may set a threshold indicating that once a number of new features (or new topics) exceed a user-defined threshold, the system automatically updates the taxonomy [i.e., once the threshold is exceeded the topic is added to the taxonomy, e.g., “expanding the hierarchical taxonomy”])
Nefedov does not appear to explicitly disclose a lowest level of the hierarchical taxonomy comprising a plurality of leaf nodes,
by adding a first level below the lowest level comprising a plurality of first nodes, the plurality of first nodes comprising the determined name for each cluster of the plurality of second clusters; and
further expanding the hierarchical taxonomy by adding a second level below the first level comprising a plurality of second nodes, the plurality of second nodes comprising the determined topics associated with the documents in each cluster of the plurality of first clusters.
Beller discloses a lowest level of the hierarchical taxonomy comprising a plurality of leaf nodes (Beller, [0051], see taxonomy entries include a top level categorical term, followed by levels of subcategories in a group, separated by an “>” indicator; Beller, [0053], see , in response to identifying “head(gloss1)” of “electrical component” present in taxonomy1 304, mapper 230 adds “term1” of “circuit breaker”, which is paired with “gloss 1” in collection of pairs 306 to taxonomy1 304 to generate an updated taxonomy 320 of taxonomy1′ 322 with an entry “basic discrete device>electrical component>circuit breaker”, indicating “circuit breaker” is a subcategory of “electrical component” based on the glossary definition of “circuit breaker” [i.e., corresponding to a 3 level hierarchical taxonomy]; and Beller, Claims 2 and 12, see initial taxonomy comprising the one or more existing terms for the domain identified in the hierarchical structure comprising a parent node and one or more levels of child nodes [i.e., where the farthest child node level from the parent node corresponds to the “lowest level of the hierarchical taxonomy comprising a plurality of leaf nodes”]),
by adding a first level below the lowest level comprising a plurality of first nodes, the plurality of first nodes comprising the determined name for each cluster of the plurality of second clusters (Beller, [0057], see head word extractor 260 searches “gloss1 ‘an electrical component used to protect circuits’” in collection of pairs 406 and extracts a noun phrase of “electrical component”, to identify a “head(gloss1)” of “electrical component”. As illustrated at reference numeral 414, head word extractor 260 identifies that the “head(gloss1)” of “electrical component” is not present in taxonomy2 404, so taxonomy2 404 remains unchanged. In response to identifying that head(gloss1) of “electrical component” is not present in taxonomy2 404, interference controller 262 initiates tiny function 268 to look up the noun phrase of ‘electrical component’ in glossary2 408 to determine if the phrase is in the current glossary. In response to tiny function 268 identifying the noun phrase of ‘electrical component’ in entry <term2,gloss2> of glossary2 408, tiny function 268 builds a tiny taxonomy with the matching entry as a child node and the noun phrase as a parent node, as illustrated by “build T_tiny: [electrical component>circuit breaker]” at reference numeral 416; Beller, [0058], see since “basic discrete device” is present in taxonomy2 404, mapper 230 maps “term2” of “electrical component” to taxonomy2 404, resulting in an updated taxonomy iteration of taxonomy2′ within an entry “basic discrete device>electrical component” [i.e., ‘electric component’ corresponds to the “determined name” in the “first level below the lowest level” and see above for disclosing the nodes], as illustrated at reference numeral 418. In the example, as illustrated at reference numeral 420, since mapped term “electrical component” also matches the root of the tiny taxonomy, mapper 230 puts the entirety of the tiny taxonomy in an additional updated taxonomy iteration of taxonomy2″, with the entry updated to “basic discrete device>electrical component>circuit breaker”. In the example, the final updated taxonomy iteration is stored in updated taxonomy 430 is taxonomy2″ 432 with entry updated to “basic discrete device>electrical component>circuit breaker”); and
further expanding the hierarchical taxonomy by adding a second level below the first level comprising a plurality of second nodes, the plurality of second nodes comprising the determined topics associated with the documents in each cluster of the plurality of first clusters (Beller, [0058], see since “basic discrete device” is present in taxonomy2 404, mapper 230 maps “term2” of “electrical component” to taxonomy2 404, resulting in an updated taxonomy iteration of taxonomy2′ within an entry “basic discrete device>electrical component”, as illustrated at reference numeral 418. In the example, as illustrated at reference numeral 420, since mapped term “electrical component” also matches the root of the tiny taxonomy, mapper 230 puts the entirety of the tiny taxonomy in an additional updated taxonomy iteration of taxonomy2″, with the entry updated to “basic discrete device>electrical component>circuit breaker”. In the example, the final updated taxonomy iteration is stored in updated taxonomy 430 is taxonomy2″ 432 with entry updated to “basic discrete device>electrical component>circuit breaker”; and Beller, [0059], see by automatically generating a tiny taxonomy identifying that that “circuit breaker” [i.e., ‘circuit breaker’ corresponds to the “determined topic” in the “second level below the first level” and see above for disclosing the nodes] is a subcategory of a noun phrase of “electrical component” and mapping the tiny taxonomy to the current taxonomy, domain extension controller 130 efficiently extends initial taxonomy 402 based on a glossary definition into updated taxonomy 420, without requiring an evaluation of each word within a glossary definition with each existing term in glossary2 408).
Nefedov and Beller are analogous art because they are from the same field of endeavor such as natural language processing.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention, having the teachings of Nefedov and Beller before him/her, to modify the cluster analysis of Nefedov to include the taxonomy of Beller because it allow efficient expanding of a knowledge graph.
The suggestion/motivation for doing so would have been to generate an updated taxonomy for a domain, see Beller, [0003].
Therefore, it would have been obvious to combine Beller with Nefedov to obtain the invention as specified in the instant claim(s).
Claim(s) 11 and 20 recite(s) similar limitations to claim 1 and is/are rejected under the same rationale.
With respect to claim 11, Nefedov discloses a system for expanding a hierarchical taxonomy associated with a corpus of documents (Nefedov, [0005], see extracting taxonomies based on hierarchical clustering on graphs related to a corpus of documents and using said taxonomies for classifying documents), the hierarchical taxonomy comprising a plurality of levels with each level comprising one or more nodes (Nefedov, [0005], see the methods comprise applying a community detection algorithm to the graph to detect a plurality of clusters of nodes of the graph at each of hierarchical levels such that, the graph-modularity of plurality of clusters exceeds a minimum modularity threshold at each hierarchical level), the system comprising:
a processing device (Nefedov, [0006], see processor); and
a non-transitory, processor-readable storage medium comprising one or more programming instructions stored thereon (Nefedov, [0006], see memory device).
With respect to claim 20, Nefedov discloses a non-transitory, computer-readable storage medium that is operable by a computer to expand a hierarchical taxonomy associated with a corpus of documents (Nefedov, [0005], see extracting taxonomies based on hierarchical clustering on graphs related to a corpus of documents and using said taxonomies for classifying documents), the hierarchical taxonomy comprising a plurality of levels with each level comprising one or more nodes (Nefedov, [0005], see the methods comprise applying a community detection algorithm to the graph to detect a plurality of clusters of nodes of the graph at each of hierarchical levels such that, the graph-modularity of plurality of clusters exceeds a minimum modularity threshold at each hierarchical level), the non-transitory, computer-readable storage medium comprising one or more programming instructions stored thereon (Nefedov, [0006], see memory device).
9. Claim(s) 2-8 and 12-18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Nefedov in view of Beller in further view of Burton, US 12,210,839 B1 (hereinafter “Burton”).
Claims 2 and 12
Claims 2 and 12 incorporate all of the limitations above.
The combination of Nefedov and Beller does not appear to explicitly disclose wherein performing the cluster analysis on the documents associated with the leaf node of the hierarchical taxonomy comprises:
using natural language processing to determine vectorizations associated with the documents; and
performing cluster analysis of the vectorizations associated with the documents.
Burton discloses wherein performing the cluster analysis on the documents associated with the leaf node of the hierarchical taxonomy comprises:
using natural language processing to determine vectorizations associated with the documents (See below); and
performing cluster analysis of the vectorizations associated with the documents (Burton, Col. 35, line 55-Col. 36, line 7, see the topic clusters tool focuses on the topic clusters, keywords, and keyword incidences that can be discovered in the document by means of topic analysis from the field of natural language processing. The interactive portion of the system may be largely agnostic to the computational method used for topic analysis, however, the system includes at a base usage tier for the user the relatively slow method of Latent Dirichlet Allocation for topic modeling which uses a Markov Chain Monte Carlo process to derive topic clusters bottom-up from text in the document, which yields a hierarchical data structure that includes keywords potentially multiply in small clusters of other keywords grouped by common mention and which keeps incidence data subordinate to cluster data; and Burton, Col. 32, lines 1-9, see the vector space tool, focusing on viewing constituent elements (e.g. sentences, sequence blocks, paragraphs, human or machine annotations) of the written artifact through the learned low-dimensional embeddings typical to the field of unsupervised learning [i.e., “natural language processing”], allows the user to conduct a spatial analysis of the writing in the document that removes the chronologically-specified narrative progression in favor of position computed, depending upon the similarity of the units in a low-dimensional word-embedding space).
Nefedov, Beller and Burton are analogous art because they are from the same field of endeavor such as natural language processing.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention, having the teachings of Nefedov, Beller and Burton before him/her, to modify the taxonomy cluster analysis of the combination of Nefedov and Beller to include the vectoring of Burton because it is an efficient way to determine how similar clusters are.
The suggestion/motivation for doing so would have been because vector space analyses allows for examining the kinship of sentences and performing fuzzy matching more powerfully than by using edit distances or bag of words models, see Burton, Col. 4, line 48-Col. 5, line 12.
Therefore, it would have been obvious to combine Burton with the combination of Nefedov and Beller to obtain the invention as specified in the instant claim(s).
Claim 3
Claim 3 incorporates all of the limitations above.
The combination of Nefedov and Beller does not appear to explicitly disclose wherein performing the cluster analysis on the documents associated with the leaf node of the hierarchical taxonomy comprises:
using natural language processing to determine vectorizations of portions of each of the documents; and
performing cluster analysis of the vectorizations.
Burton discloses wherein performing the cluster analysis on the documents associated with the leaf node of the hierarchical taxonomy comprises:
using natural language processing to determine vectorizations of portions of each of the documents (See below); and
performing cluster analysis of the vectorizations (Burton, Col. 35, line 55-Col. 36, line 7, see the topic clusters tool focuses on the topic clusters, keywords, and keyword incidences that can be discovered in the document by means of topic analysis from the field of natural language processing. The interactive portion of the system may be largely agnostic to the computational method used for topic analysis, however, the system includes at a base usage tier for the user the relatively slow method of Latent Dirichlet Allocation for topic modeling which uses a Markov Chain Monte Carlo process to derive topic clusters bottom-up from text in the document, which yields a hierarchical data structure that includes keywords potentially multiply in small clusters of other keywords grouped by common mention and which keeps incidence data subordinate to cluster data; and Burton, Col. 32, lines 1-9, see the vector space tool, focusing on viewing constituent elements (e.g. sentences, sequence blocks, paragraphs, human or machine annotations) of the written artifact through the learned low-dimensional embeddings typical to the field of unsupervised learning [i.e., “natural language processing”], allows the user to conduct a spatial analysis of the writing in the document that removes the chronologically-specified narrative progression in favor of position computed, depending upon the similarity of the units in a low-dimensional word-embedding space).
See claims 2 and 12 above for the motivation to combine.
Claims 4 and 13
With respect to claims 4 and 13, the combination of Nefedov, Beller and Burton discloses further comprising:
determining the plurality of first clusters based on a cosine similarity between the vectorizations associated with the documents (Burton, Col. 35, line 55-Col. 36, line 7, see the topic clusters tool focuses on the topic clusters, keywords, and keyword incidences that can be discovered in the document by means of topic analysis from the field of natural language processing. The interactive portion of the system may be largely agnostic to the computational method used for topic analysis, however, the system includes at a base usage tier for the user the relatively slow method of Latent Dirichlet Allocation for topic modeling which uses a Markov Chain Monte Carlo process to derive topic clusters bottom-up from text in the document, which yields a hierarchical data structure that includes keywords potentially multiply in small clusters of other keywords grouped by common mention and which keeps incidence data subordinate to cluster data; Burton, Col. 32, lines 1-9, see the vector space tool, focusing on viewing constituent elements (e.g. sentences, sequence blocks, paragraphs, human or machine annotations) of the written artifact through the learned low-dimensional embeddings typical to the field of unsupervised learning [i.e., “natural language processing”], allows the user to conduct a spatial analysis of the writing in the document that removes the chronologically-specified narrative progression in favor of position computed, depending upon the similarity of the units in a low-dimensional word-embedding space; and Burton, Col. 50, lines 4-45, see the cosine distance between the bags of words vector representations of named entities).
Claims 5 and 14
Claims 5 and 14 incorporate all of the limitations above.
The combination of Nefedov and Beller does not appear to explicitly disclose wherein determining the one or more topics associated with the documents in each cluster of the plurality of first clusters comprises: inputting the documents and a prompt into a large language model; and determining the one or more topics based on an output of the large language model.
Burton discloses wherein determining the one or more topics associated with the documents in each cluster of the plurality of first clusters comprises:
inputting the documents and a prompt into a large language model (See below); and
determining the one or more topics based on an output of the large language model (Burton, Col. 9, lines 13-39, see producing a prompt that can be dispatched to an LLM [i.e., “large language model”] to produce a discursive synthetic analysis of the anomaly, point or region of interest, or data cluster; and Burton, Col. 35, line 55-Col. 36, line 7, see the topic clusters tool focuses on the topic clusters, keywords, and keyword incidences that can be discovered in the document by means of topic analysis from the field of natural language processing. The interactive portion of the system may be largely agnostic to the computational method used for topic analysis, however, the system includes at a base usage tier for the user the relatively slow method of Latent Dirichlet Allocation for topic modeling which uses a Markov Chain Monte Carlo process to derive topic clusters bottom-up from text in the document, which yields a hierarchical data structure that includes keywords potentially multiply in small clusters of other keywords grouped by common mention and which keeps incidence data subordinate to cluster data).
See claims 2 and 12 above for the motivation to combine.
Claims 6 and 15
With respect to claims 6 and 15, the combination of Nefedov, Beller and Burton discloses wherein the prompt asks the large language model to generate the one or more topics based on the documents (Burton, Col. 9, lines 13-39, see producing a prompt that can be dispatched to an LLM [i.e., “large language model”] to produce a discursive synthetic analysis of the anomaly, point or region of interest, or data cluster; and Burton, Col. 35, line 55-Col. 36, line 7, see the topic clusters tool focuses on the topic clusters, keywords, and keyword incidences that can be discovered in the document by means of topic analysis from the field of natural language processing. The interactive portion of the system may be largely agnostic to the computational method used for topic analysis, however, the system includes at a base usage tier for the user the relatively slow method of Latent Dirichlet Allocation for topic modeling which uses a Markov Chain Monte Carlo process to derive topic clusters bottom-up from text in the document, which yields a hierarchical data structure that includes keywords potentially multiply in small clusters of other keywords grouped by common mention and which keeps incidence data subordinate to cluster data).
Claims 7 and 16
Claims 7 and 16 incorporate all of the limitations above.
The combination of Nefedov and Beller does not appear to explicitly disclose wherein performing the cluster analysis on the topics comprises:
using natural language processing to determine vectorizations associated with the topics; and
performing cluster analysis on the vectorizations.
Burton discloses wherein performing the cluster analysis on the topics comprises:
using natural language processing to determine vectorizations associated with the topics (See below); and
performing cluster analysis on the vectorizations (Burton, Col. 35, line 55-Col. 36, line 7, see the topic clusters tool focuses on the topic clusters, keywords, and keyword incidences that can be discovered in the document by means of topic analysis from the field of natural language processing. The interactive portion of the system may be largely agnostic to the computational method used for topic analysis, however, the system includes at a base usage tier for the user the relatively slow method of Latent Dirichlet Allocation for topic modeling which uses a Markov Chain Monte Carlo process to derive topic clusters bottom-up from text in the document, which yields a hierarchical data structure that includes keywords potentially multiply in small clusters of other keywords grouped by common mention and which keeps incidence data subordinate to cluster data; and Burton, Col. 32, lines 1-9, see the vector space tool, focusing on viewing constituent elements (e.g. sentences, sequence blocks, paragraphs, human or machine annotations) of the written artifact through the learned low-dimensional embeddings typical to the field of unsupervised learning [i.e., “natural language processing”], allows the user to conduct a spatial analysis of the writing in the document that removes the chronologically-specified narrative progression in favor of position computed, depending upon the similarity of the units in a low-dimensional word-embedding space).
See claims 2 and 12 above for the motivation to combine.
Claims 8 and 17
Claims 8 and 17 incorporate all of the limitations above.
The combination of Nefedov and Beller does not appear to explicitly disclose wherein determining the name for each cluster of the plurality of second clusters comprises:
inputting the topics for each cluster of the plurality of second clusters and a prompt into a large language model; and
determining the name for each cluster of the plurality of second clusters based on an output of the large language model.
Burton discloses wherein determining the name for each cluster of the plurality of second clusters comprises:
inputting the topics for each cluster of the plurality of second clusters and a prompt into a large language model (See below); and
determining the name for each cluster of the plurality of second clusters based on an output of the large language model (Burton, Col. 9, lines 13-39, see producing a prompt that can be dispatched to an LLM [i.e., “large language model”] to produce a discursive synthetic analysis of the anomaly, point or region of interest, or data cluster; Burton, Col. 35, line 55-Col. 36, line 7, see the topic clusters tool focuses on the topic clusters, keywords, and keyword incidences that can be discovered in the document by means of topic analysis from the field of natural language processing. The interactive portion of the system may be largely agnostic to the computational method used for topic analysis, however, the system includes at a base usage tier for the user the relatively slow method of Latent Dirichlet Allocation for topic modeling which uses a Markov Chain Monte Carlo process to derive topic clusters bottom-up from text in the document, which yields a hierarchical data structure that includes keywords potentially multiply in small clusters of other keywords grouped by common mention and which keeps incidence data subordinate to cluster data; and Burton, Col. 32, lines 1-9, see the vector space tool, focusing on viewing constituent elements (e.g. sentences, sequence blocks, paragraphs, human or machine annotations) of the written artifact through the learned low-dimensional embeddings typical to the field of unsupervised learning [i.e., “natural language processing”], allows the user to conduct a spatial analysis of the writing in the document that removes the chronologically-specified narrative progression in favor of position computed, depending upon the similarity of the units in a low-dimensional word-embedding space).
See claims 2 and 12 above for the motivation to combine.
Claims 9 and 18
With respect to claims 9 and 18, the combination of Nefedov, Beller and Burton discloses wherein the prompt asks the large language model to generate the name for each cluster of the plurality of second clusters based on the topics for each cluster (Burton, Col. 9, lines 13-39, see producing a prompt that can be dispatched to an LLM [i.e., “large language model”] to produce a discursive synthetic analysis of the anomaly, point or region of interest, or data cluster; Burton, Col. 35, line 55-Col. 36, line 7, see the topic clusters tool focuses on the topic clusters, keywords [i.e., “name”], and keyword incidences that can be discovered in the document by means of topic analysis from the field of natural language processing. The interactive portion of the system may be largely agnostic to the computational method used for topic analysis, however, the system includes at a base usage tier for the user the relatively slow method of Latent Dirichlet Allocation for topic modeling which uses a Markov Chain Monte Carlo process to derive topic clusters bottom-up from text in the document, which yields a hierarchical data structure that includes keywords potentially multiply in small clusters of other keywords grouped by common mention and which keeps incidence data subordinate to cluster data; Nefedov, [0025], see extracting taxonomies for the documents from the graph. For instance, the graph hierarchically clustered [i.e., “cluster analysis”] to form clusters of nodes at different hierarchical levels, for example, by adaptively pruning the graph at each hierarchy until the graph becomes a network of recognizable or identifiable clusters [i.e., “a plurality of first clusters”], and repeating the same procedure for each of the clusters [i.e., “determine a plurality of second clusters”] until pruning is no longer possible or acceptable (e.g., causes high level of information loss); and Nefedov, [0024], see co-occurrence matrices for a corpus of documents may be constructed based on similarity among features of the documents that are extracted from the corpus of documents, such as the co-occurrence of words taken from the whole vocabulary over a corpus of documents. … These co-occurrent elements may be seen as dominated “phrases” over all documents. … That is, the nodes represent the words in the corpus of vocabulary and the edges connecting nodes may represent the co-occurrences between the words represented by the connected nodes. Then clustering of the co-occurrence matrix would reveal dominated (and possibly hierarchical) clusters or “topics” over all documents. These “topics” may be used as “labels” [i.e., “name for each cluster”] for hierarchical clusters of documents described).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
– Feldman et al., 6442545 for term-level text with mining taxonomies;
– Saxena et al., 2021/0099282 for management of software defined network configuration data based on hash trees; and
– Lyon-Smith et al., 2011/0137899 for partitioned list.
Point of Contact
Any inquiry concerning this communication or earlier communications from the examiner should be directed to HUBERT G CHEUNG whose telephone number is (571) 270-1396. The examiner can normally be reached M-R 8:00A-5:00P EST; alt. F 8:00A-4:00P EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Neveen Abel-Jalil can be reached at (571) 270-0474. 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.
HUBERT G. CHEUNG
Assistant Examiner
Art Unit 2152
Examiner: Hubert Cheung
/Hubert Cheung/Assistant Examiner, Art Unit 2152Date: April 1, 2026
/NEVEEN ABEL JALIL/Supervisory Patent Examiner, Art Unit 2152