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 .
Response to Arguments
Applicant’s arguments, see pp. 14, filed 9/24/2025, with respect to the rejection of claim 1 under 35 U.S.C. 103 have been fully considered and are persuasive. Therefore, the rejection has been withdrawn. However, upon further consideration, a new ground of rejection is made in view of Zhang.
101 Rejection
Applicant states (pp. 9) that steps involving a neural network, complex data formats, large datasets, and entailment scores cannot practically be performed in the human mind. Examiner respectfully disagrees, because entailment relationship between terms and phrases can be mentally evaluated and judged (Step 2A Prong One), while neural network and markup language format are high-level recitations of generic computer components and functions that represent mere instructions to apply on a computer (Step 2A Prong Two).
Applicant further states (pp. 10) that human mind cannot be trained to produce a set of entailment scores of locations satisfying a query, and to annotate a location when the matching of premises to hypotheses is above a threshold. Examiner respectfully disagrees. As noted above, entailment relationship between premises and hypotheses can be mentally evaluated and judged, while annotating a location can be mentally performed with the help of pen and paper (Step 2A Prong One).
Applicant further states (pp. 11) that claim 1 recites neural networks that are specialized models for complex tasks, including a first neural network that generates a natural language query from input strings, linked to a second neural network to produce entailment scores indicating locations that satisfy the query. Examiner respectfully disagrees, since both tasks are mentally performable (Step 2A Prong One), and neural networks are simply high-level recitation of generic computer components and functions (Step 2A Prong Two).
In summary, independent claims 1, 6 and 14 are not eligible.
103 Rejection
Applicant states (pp. 14) the cited prior art of record combined does not teach the amended limitations of claim 1. They are taught instead by combining Dang with Zhang.
Dang extracts noun phrases (i.e., locations) from text content, and compares them to terms of an ontology by performing phrase mapping (i.e., satisfying query) using a rule-based algorithm such as WSD [0018]. Dang maps extracted noun phrases to terms of an ontology and to a highest ranked sense of the ontology terms using a WSD algorithm (i.e., first generative neural network) [0019], weights the term significance (i.e., score) [0020] of the mapped ontology concepts from their ontological and statistical features, and generates key-phrases from the weighted concepts as semantic tags [0016]. Dang captures ontological properties hypernym, domain, category and holonym to create lexical chains [0091]. Each sense (i.e., input) is connected to its parent sense to increase WSD scores (i.e., output) of its ancestor senses until reaching a predetermined threshold of ancestor distance (i.e., value) [0093].
Dang does not disclose claim element “entailment scores”; however, Zhang teaches an RTE system for natural language processing (i.e., second generative neural network) that performs lexical, syntactic and semantic analysis of an input text (i.e., premise) in XML format (i.e., markup language) (Zhang: sec. III.A, para. 1), and raises (i.e., matches) a hypothesis that is likely inferred (i.e., with entailment score) from that (Zhang: Abstract).
Therefore, it would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of Zhang to Dang. One having ordinary skill in the art would have found motivation to utilize the RTE system of Zhang to convert textual documents in industry standard XML into feature vectors of content, to be supplied to the WSD algorithm of Dang.
In summary, Dang combined with Zhang teaches the argued limitations of independent claims 1, 6 and 14.
Double Patenting
The nonstatutory double patenting rejection is based on a judicially created doctrine grounded in public policy (a policy reflected in the statute) so as to prevent the unjustified or improper timewise extension of the “right to exclude” granted by a patent and to prevent possible harassment by multiple assignees. A nonstatutory double patenting rejection is appropriate where the conflicting claims are not identical, but at least one examined application claim is not patentably distinct from the reference claim(s) because the examined application claim is either anticipated by, or would have been obvious over, the reference claim(s). See, e.g., In re Berg, 140 F.3d 1428, 46 USPQ2d 1226 (Fed. Cir. 1998); In re Goodman, 11 F.3d 1046, 29 USPQ2d 2010 (Fed. Cir. 1993); In re Longi, 759 F.2d 887, 225 USPQ 645 (Fed. Cir. 1985); In re Van Ornum, 686 F.2d 937, 214 USPQ 761 (CCPA 1982); In re Vogel, 422 F.2d 438, 164 USPQ 619 (CCPA 1970); In re Thorington, 418 F.2d 528, 163 USPQ 644 (CCPA 1969).
A timely filed terminal disclaimer in compliance with 37 CFR 1.321(c) or 1.321(d) may be used to overcome an actual or provisional rejection based on nonstatutory double patenting provided the reference application or patent either is shown to be commonly owned with the examined application, or claims an invention made as a result of activities undertaken within the scope of a joint research agreement. See MPEP § 717.02 for applications subject to examination under the first inventor to file provisions of the AIA as explained in MPEP § 2159. See MPEP § 2146 et seq. for applications not subject to examination under the first inventor to file provisions of the AIA . A terminal disclaimer must be signed in compliance with 37 CFR 1.321(b).
The filing of a terminal disclaimer by itself is not a complete reply to a nonstatutory double patenting (NSDP) rejection. A complete reply requires that the terminal disclaimer be accompanied by a reply requesting reconsideration of the prior Office action. Even where the NSDP rejection is provisional the reply must be complete. See MPEP § 804, subsection I.B.1. For a reply to a non-final Office action, see 37 CFR 1.111(a). For a reply to final Office action, see 37 CFR 1.113(c). A request for reconsideration while not provided for in 37 CFR 1.113(c) may be filed after final for consideration. See MPEP §§ 706.07(e) and 714.13.
The USPTO Internet website contains terminal disclaimer forms which may be used. Please visit www.uspto.gov/patent/patents-forms. The actual filing date of the application in which the form is filed determines what form (e.g., PTO/SB/25, PTO/SB/26, PTO/AIA /25, or PTO/AIA /26) should be used. A web-based eTerminal Disclaimer may be filled out completely online using web-screens. An eTerminal Disclaimer that meets all requirements is auto-processed and approved immediately upon submission. For more information about eTerminal Disclaimers, refer to www.uspto.gov/patents/apply/applying-online/eterminal-disclaimer.
Claims 1-15, 17-19 and 21-22 are provisionally rejected on the ground of nonstatutory double patenting as being unpatentable over claims 1-4, 6-10, 12-16, 18-23 of copending Application No. 18/642,744 (reference application) in view of Dang and Zhang. Although the claims at issue are not identical, they are not patentably distinct from each other. In particular, claim 1 of the instant application is mapped to claim 1 of the reference application by the following mapping table.
Instant Application 18/763,950
Reference Application 18/642,744
1.A system, comprising: one or more processors; and memory that stores computer-executable instructions that, as a result of execution by the one or more processors, cause the system to at least:
1.A system, comprising: one or more processors; and memory that stores computer-executable instructions that, as a result of execution by the one or more processors, cause the system to at least:
identify a set of locations in a data object to be annotated as corresponding to metadata of a data object;
identify a plurality of candidate locations in the document as potentially corresponding to the field type and the field value for each candidate location
use a first generative neural network to generate a natural language text query by causing the system to at least: provide as input to the first generative neural network: a first string of metadata in markup language format; and a second string of original data of the data object; and
extract a field type and a field value from the metadata of the document; input the query and the data into a neural network
obtain the natural language text query as output from the first generative neural network, the natural language text query being in a human-readable language;
generate a query derived from the field type and the field value;
cause, using the natural language text query as input, a second generative neural network to produce a set of scores for the set of locations, the set of scores comprising a set of entailment scores that are generated by comparing a set of premises that comprise portions of text around the set of locations in the data object and a set of hypotheses comprising the natural language text query, the set of entailment scores indicating whether locations of the set of locations satisfy a query; and
produce a score indicating a likelihood of the candidate location satisfying the query, wherein the neural network is trained to receive natural language queries derived from metadata about documents;
annotate a location of the set of locations as corresponding to the metadata based on the set of scores, the location annotated as a result of at least one premise of the set of premises matching at least one hypothesis of the set of hypotheses according to a threshold value.
produce an annotation that identifies the candidate location as corresponding to the field type.
2.
The system of claim 1, wherein the query is a natural language query.
Claim 1 of the reference application does not disclose claim element “generative neural network”; however, Dang maps extracted noun phrases to terms of an ontology and to a highest ranked sense of the ontology terms using a WSD algorithm (i.e., first generative neural network) [0019], weights the term significance (i.e., score) [0020] of the mapped ontology concepts from their ontological and statistical features, and generates key-phrases from the weighted concepts as semantic tags [0016]. Dang captures ontological properties hypernym, domain, category and holonym to create lexical chains [0091]. Each sense (i.e., input) is connected to its parent sense to increase WSD scores (i.e., output) of its ancestor senses until reaching a predetermined threshold of ancestor distance (i.e., value) [0093].
Claim 1 of the reference application does not disclose claim element “entailment scores”; however, Zhang teaches an RTE system for natural language processing (i.e., second generative neural network) that performs lexical, syntactic and semantic analysis of an input text (i.e., premise) in XML format (i.e., markup language) (Zhang: sec. III.A, para. 1), and raises (i.e., matches) a hypothesis that is likely inferred (i.e., with entailment score) from that (Zhang: Abstract).
Therefore, it would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of Dang and Zhang to the reference application. One having ordinary skill in the art would have found motivation to utilize the RTE system of Zhang to convert textual documents in industry standard XML into feature vectors of content, to be supplied to the WSD algorithm of Dang in the reference application.
This is a provisional nonstatutory double patenting rejection.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-15, 17-19 and 21-22 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea as a mental process without significantly more.
Claims 1, 6 and 14 are rejected under 35 U.S.C. 101.
Step 1
Claim 1 recites “A system” comprising processors and memory. Claim 6 recites “A method”. Claim 14 recites “A non-transitory computer-readable storage medium”. Thus they are directed to a statutory category.
Step 2A – Prong One
Claims 1, 6 and 14 recite the following limitations directed to an abstract idea:
“identify a set of locations…to be annotated as corresponding to metadata of a data object” recites an abstract idea as a mental process. One can mentally observe or evaluate the locations of terms in a document object that correspond to recognized key phrases.
“provide as input…a first string of metadata… and second string of original data” and “obtain the natural language text query as output” recite an abstract idea as a mental process. One can mentally observe or evaluate terms in the document to recognize key phrases (persons, places or things) mentioned (spec. [0014]) and to formulate a query accordingly.
“using the natural language text query as input,…produce a set of scores for the set of locations,…indicating whether locations…satisfy a query” recites an abstract idea as a mental process. One can mentally evaluate and judge how close is the match between the identified terms in the document and the recognized key phrases.
“a set of entailment scores…generated by comparing a set of premises…and a set of hypotheses” recites an abstract idea as a mental process. One can mentally compare identified terms in the document with the recognized key phrases to determine how likely one is inferred from another.
“annotate a location…as corresponding to the metadata based on the set of scores…as a result of…premises matching…hypotheses” recites an abstract idea as a mental process. One can mentally annotate, with the help of pen and paper, identified terms in the document with the recognized key phrases that are inferred from them.
Step 2A – Prong Two
This judicial exception is not integrated into a practical application.
Claims 1, 6 and 14 include additional elements “processor”, “memory”, “instruction”, “computer system”, “data object”, “natural language text”, “markup language format”, “human-readable language”, “generative neural network”, “input” and “output”, which are high-level recitations of generic computer components and functions that represent mere instructions to apply on a computer per MPEP §2106.05(f).
Viewing the additional limitations together and the claim as a whole, nothing provides integration into a practical application.
Step 2B
Claim 1, 6 and 14 include additional elements on providing “as input” and “using the natural language text query as input”, which qualify as “i. Receiving or transmitting data over a network”, and is recognized by the courts as well-understood, routine, and conventional per MPEP §2106.05(d)(II).
The conclusions on the mere instructions to apply the abstract idea using generic computer components and functions carry over and do not add significantly more or provide any "inventive concept".
In summary, independent claims 1, 6 and 14 are not eligible. Claims 2-5, 7-13, 15, 17-19 and 21-22 depend on claims 1, 6 and 14 respectively and recite the same abstract idea.
Step 2A Prong One
The following claims recite additional elements that are mentally performable. Claims 2, 7 and 19 recite generating natural language text query or human-readable language query from metadata. Claim 5 recites computing/inferring scores. Claims 8 and 17-18 recite documents as text, image or audio recording. Claim 9 recites computing the confidence level of matching identified terms in documents to recognized key phrases. Claim 10 recites using the query and identified terms in document objects as input to a generative neural network to identify matching terms. Claim 22 recites “contextual information”.
Step 2A Prong Two
The following claims recite additional elements that are high-level recitations of generic computer components and functions. Claims 3 and 11 recite “Retrieval-Augmented Generation”. Claim 4 recites “large language model”. Claim 13 recites “generative pre-trained transformer”. Claim 15 recites “knowledge graph”. Claim 21 recites “knowledge base”, “synonym” and “acronym”. Claim 22 recites “external database”.
Step 2B
Claim 12 recites storing annotated document objects, which qualifies as “iv. Storing and retrieving information in memory”, and is recognized by the courts as well-understood, routine, and conventional per MPEP §2106.05(d)(II).
Claim 12 recites feeding output of one generative neural network as input to another, which qualifies as “i. Receiving or transmitting data over a network”, and is recognized by the courts as well-understood, routine, and conventional per MPEP §2106.05(d)(II).
In summary, these dependent claims do not add any additional elements sufficient to make the claims non-abstract. Therefore, they are not eligible accordingly.
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 1-2, 5-10, 12, 14-15, 19 and 21 are rejected under 35 U.S.C. 103 as being unpatentable over Dang et al. US patent application 2014/0040275 [herein “Dang”], and further in view of Zhang et al. Recognizing Textual Entailment with Synthetic Analysis Based on SVM and Feature Value Control. 2012 IEEE International Conference on Computer Science and Automation Engineering, pp. 714-717 [herein “Zhang”].
Claim 1 recites “A system, comprising: one or more processors; and memory that stores computer-executable instructions that, as a result of execution by the one or more processors, cause the system to at least: identify a set of locations in a data object to be annotated as corresponding to metadata of a data object;”
Dang maps text documents to ontological entities/concepts and generates semantic keywords or key-phrases as document metadata. Documents (i.e., data objects) are indexed with metadata, enabling search (i.e., query) based on semantic and conventional keywords (i.e., natural language text) [0165].
Claim 1 further recites “use a first generative neural network to generate a natural language text query by causing the system to at least: provide as input to the first generative neural network: a first string of metadata in markup language format; and a second string of original data of the data object; and obtain the natural language text query as output from the first generative neural network, the natural language text query being in a human-readable language;”
Dang extracts noun phrases (i.e., second string) from text content (i.e., original data), maps them to terms (i.e., first string) of an ontology (i.e., metadata) and to a highest ranked sense of the ontology terms using a WSD algorithm (i.e., first generative neural network) [0019], and outputs (i.e., annotates) key-phrases (i.e., query) from the weighted concepts as semantic tags [0016].
Claim 1 further recites “cause, using the natural language text query as input, a second generative neural network to produce a set of scores for the set of locations, the set of scores comprising a set of entailment scores that are generated by comparing a set of premises that comprise portions of text around the set of locations in the data object and a set of hypotheses comprising the natural language text query, the set of entailment scores indicating whether locations of the set of locations satisfy a query; and”.
Dang extracts noun phrases (i.e., locations) from text content, and compares them to terms of an ontology by performing phrase mapping (i.e., satisfying query) using a rule-based algorithm such as WSD [0018].
Claim 1 further recites “annotate a location of the set of locations as corresponding to the metadata based on the set of scores, the location annotated as a result of at least one premise of the set of premises matching at least one hypothesis of the set of hypotheses according to a threshold value.”
Dang maps extracted noun phrases to terms of an ontology and to a highest ranked sense of the ontology terms using a WSD algorithm [0019], weights the term significance (i.e., score) [0020] of the mapped ontology concepts from their ontological and statistical features, and generates key-phrases from the weighted concepts as semantic tags [0016]. Dang captures ontological properties hypernym, domain, category and holonym to create lexical chains [0091]. Each sense (i.e., input) is connected to its parent sense to increase WSD scores (i.e., output) of its ancestor senses until reaching a predetermined threshold of ancestor distance (i.e., value) [0093].
Dang does not disclose claim element “entailment scores”; however, Zhang teaches an RTE system for natural language processing (i.e., second generative neural network) that performs lexical, syntactic and semantic analysis of an input text (i.e., premise) in XML format (i.e., markup language) (Zhang: sec. III.A, para. 1), and raises (i.e., matches) a hypothesis that is likely inferred (i.e., with entailment score) from that (Zhang: Abstract).
Therefore, it would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of Zhang to Dang. One having ordinary skill in the art would have found motivation to utilize the RTE system of Zhang to convert textual documents in industry standard XML into feature vectors of content, to be supplied to the WSD algorithm of Dang.
Claims 6 and 14 are analogous to claim 1, and are similarly rejected.
Claim 2 recites “The system of claim 1, wherein the system generates the natural language text query by at least using the metadata of the data object.”
Dang maps text documents to ontological entities/concepts and generates semantic keywords or key-phrases as document metadata. Documents (i.e., data objects) are indexed with metadata, enabling search (i.e., query) based on semantic and conventional keywords (i.e., natural language text) [0165].
Claim 5 recites “The system of claim 1, wherein the set of entailment scores indicate a likelihood that the location of the set of locations corresponds to the metadata.”
Dang extracts noun phrases (i.e., locations) from text content, maps them to terms of an ontology (i.e., metadata) and to a highest ranked sense of the ontology terms using a WSD algorithm [0019], weights the term significance [0020] of the mapped ontology concepts from their ontological and statistical features, and extracts key-phrases from the weighted concepts as semantic tags [0016].
Dang and Zhang teach claim 1, where Zhang teaches an RTE system for natural language processing that performs lexical, syntactic and semantic analysis of an input text in XML format (Zhang: sec. III.A, para. 1), and raises a hypothesis that is likely inferred (i.e., with entailment score) from that (Zhang: Abstract).
Claim 7 recites “The computer-implemented method of claim 6, wherein generating the natural language text query comprises deriving a human-readable language query from the metadata using the first generative neural network or an additional generative neural network.”
Dang maps text documents to ontological entities/concepts and generates (i.e., derives) semantic keywords (i.e., human-readable language) as document metadata. Documents are indexed with metadata, enabling search (i.e., query) based on semantic and conventional keywords (i.e., natural language text) [0165]. Dang extracts noun phrases from text content, and compares them to terms of an ontology by performing phrase mapping using a rule-based algorithm such as WSD (i.e., first generative neural network) [0018].
Dang and Zhang teach claim 6, where Zhang teaches an RTE system for natural language processing (i.e., additional generative neural network) that performs lexical, syntactic and semantic analysis of an input text in XML format (Zhang: sec. III.A, para. 1), and raises a hypothesis inferred from that (Zhang: Abstract).
Claim 8 recites “The computer-implemented method of claim 6, wherein the data object is one of a text file, an image, or an audio recording.”
Dang maps text documents (i.e., data objects) to ontological entities/concepts and generate semantic keywords or key-phrases as document metadata [0165].
Claim 9 recites “The computer-implemented method of claim 6, wherein a score of the set of scores satisfies the natural language text query based, at least in part, on: determining a score of the set of scores that reaches a value relative to a confidence interval; and determining, as a result of inputting the score to the second generative neural network, that the score satisfies the natural language text query based on output from the second generative neural network.”
Dang compares extracted noun phrases (i.e., natural language text) to terms of an ontology (i.e., metadata) by performing phrase mapping (i.e., satisfying query) when there is not an exact match [0018]. Dang captures ontological properties hypernym, domain, category and holonym to create lexical chains [0091]. Each sense (i.e., input) is connected to its parent sense to increase WSD scores (i.e., output) of its ancestor senses until reaching a predetermined threshold (i.e., confidence interval) of ancestor distance (i.e., value) [0093].
Dang and Zhang teach claim 6, where Zhang teaches an RTE system for natural language processing (i.e., second generative neural network) that performs lexical, syntactic and semantic analysis of an input text in XML format (Zhang: sec. III.A, para. 1), and raises a hypothesis inferred from that (Zhang: Abstract).
Claim 10 recites “The computer-implemented method of claim 6, wherein identifying the set of locations includes using the natural language text query as input to the second generative neural network to identify the set of locations.”
Dang maps text documents to ontological entities/concepts and generates semantic keywords or key-phrases as document metadata. Documents (i.e., data objects) are indexed with metadata, enabling search (i.e., query) based on semantic and conventional keywords (i.e., natural language text) [0165]. Dang extracts noun phrases from text content (i.e., locations), and compares them to terms of an ontology by performing phrase mapping using a rule-based algorithm such as WSD [0018].
Dang and Zhang teach claim 6, where Zhang teaches an RTE system for natural language processing (i.e., second generative neural network) that performs lexical, syntactic and semantic analysis of an input text in XML format (Zhang: sec. III.A, para. 1), and raises a hypothesis inferred from that (Zhang: Abstract).
Claim 12 recites “The computer-implemented method of claim 6, further comprising: storing a second data object comprising the annotated candidate location; providing the second data object to an additional neural network; and causing the additional neural network to perform at least one of training or an inference.”
Dang extracts noun phrases from text content (i.e., candidate locations), maps them to terms of an ontology and to a highest ranked sense of the ontology terms using a WSD algorithm [0019], weights the term significance (i.e., score) [0020] of the mapped ontology concepts from their ontological and statistical features, generates (i.e., annotates) and outputs (i.e., stores) key-phrases (i.e., second data object) from the weighted concepts as semantic tags [0016].
Dang and Zhang teach claim 6, where Zhang teaches an RTE system for natural language processing (i.e., additional neural network) that performs lexical, syntactic and semantic analysis of an input text in XML format (Zhang: sec. III.A, para. 1), and raises a hypothesis inferred from that (Zhang: Abstract).
Claim 15 recites “The non-transitory computer-readable storage medium of claim 14, wherein the metadata comprises a knowledge graph.”
Semantic tags of Dang transfer a document into a graph (i.e., knowledge graph) where nodes are named entities from content and links are semantic relations among the entities [0199].
Claim 19 recites “The non-transitory computer-readable storage medium of claim 14, wherein generating the query comprises: using the metadata of the data object to produce human-readable language; and generating the natural language text query from the human-readable language.”
Dang extracts (i.e., produces) noun phrases from text content (i.e., human-readable language), maps them to terms of an ontology (i.e., metadata) and to a highest ranked sense of the ontology terms using a WSD algorithm [0019], weights the term significance [0020] of the mapped ontology concepts from their ontological and statistical features, and extracts key-phrases (i.e., natural language text) from the weighted concepts as semantic tags (i.e., query) [0016].
Claim 21 recites “The system of claim 1, wherein the memory further stores computer-executable instructions that cause the system to utilize a knowledge base comprising synonyms, acronyms, or alternate names for metadata terms.”
Dang analyzes a user search query, and suggests semantic search phrases generated from a stored vocabulary (i.e., knowledge base) of an ontology [0022], including synonyms (i.e., metadata terms) of the searched term [0189].
Claims 3-4, 11, 13 and 22 are rejected under 35 U.S.C. 103 as being unpatentable over Dang as applied to claims 1 and 6 above respectively, in view of Zhang, and further in view of Lewis et al. Retrieval-Augmented Generation for Knowledge-intensive NLP Tasks. NeurIPS 2020, pp. 1-16 [herein “Lewis”].
Claim 3 recites “The system of claim 1, wherein the computer-executable instructions that cause the system to identify the set of locations include instructions that cause the system to use Retrieval-Augmented Generation to identify the set of candidate locations.”
Dang extracts noun phrases from text content (i.e., candidate locations) [0018] using Natural Language Processing (NLP) to enhance tagging quality [0079].
Dang teaches claim 1, but does not disclose this claim; however, Lewis builds Retrieval-Augmented-Generation models where the parametric memory is a pre-trained seq2seq transformer, and the non-parametric memory is a dense vector index of Wikipedia, accessed with a pre-trained neural retriever (Lewis: sec. 1, para. 2), to improve performance of knowledge-intensive NLP tasks.
Therefore, it would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of Lewis to Dang. One having ordinary skill in the art would have found motivation to incorporate the RAG models of Lewis to improve semantic tagging performance in Dang.
Claim 4 recites “The system of claim 1, wherein the first generative neural network is a large language model.”
Dang extracts noun phrases from text content, and compares them to terms of an ontology by performing phrase mapping using a rule-based algorithm such as WSD (i.e., first generative neural network) [0018].
Dang teaches claim 1, but does not disclose this claim; however, Lewis builds Retrieval-Augmented-Generation models where the parametric memory is a pre-trained seq2seq transformer, and the non-parametric memory is a dense vector index of Wikipedia, accessed with a pre-trained neural retriever (Lewis: sec. 1, para. 2), to improve performance of large pre-trained language models storing knowledge in parameters (Lewis: Abstract).
Therefore, it would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of Lewis to Dang. One having ordinary skill in the art would have found motivation to incorporate the large language models of Lewis to improve semantic tagging performance in Dang.
Claim 11 recites “The computer-implemented method of claim 6, wherein the set of scores is obtained based at least in part on using Retrieval-Augmented Generation.”
Dang maps extracted noun phrases to terms of an ontology and to a highest ranked sense of the ontology terms using a WSD algorithm [0019], weights the term significance (i.e., score) [0020] of the mapped ontology concepts from their ontological and statistical features, and extracts key-phrases from the weighted concepts as semantic tags [0016].
Dang teaches claim 6, but does not disclose this claim; however, Lewis builds Retrieval-Augmented-Generation models where the parametric memory is a pre-trained seq2seq transformer, and the non-parametric memory is a dense vector index of Wikipedia, accessed with a pre-trained neural retriever (Lewis: sec. 1, para. 2), to improve performance of knowledge-intensive NLP tasks.
Therefore, it would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of Lewis to Dang. One having ordinary skill in the art would have found motivation to incorporate the RAG models of Lewis to improve semantic tagging performance in Dang.
Claim 13 recites “The computer-implemented method of claim 6, wherein at least one of the first and second generative neural networks is a generative pre-trained transformer.”
Dang extracts noun phrases from text content, and compares them to terms of an ontology by performing phrase mapping using a rule-based algorithm such as WSD (i.e., first generative neural network) [0018].
Dang teaches claim 6, but does not disclose this claim; however, Lewis builds Retrieval-Augmented-Generation models where the parametric memory is a pre-trained seq2seq transformer (i.e., generative transformer), and the non-parametric memory is a dense vector index of Wikipedia, accessed with a pre-trained neural retriever (Lewis: sec. 1, para. 2), to improve performance of knowledge-intensive NLP tasks.
Therefore, it would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of Lewis to Dang. One having ordinary skill in the art would have found motivation to incorporate the transformers of Lewis to improve semantic tagging performance in Dang.
Claim 22 recites “The system of claim 1, wherein the computer-executable instructions that cause the system to obtain the natural language text query further include executable instructions that further cause a third generative neural network to refine the natural language text query by incorporating contextual information from an external database.”
Dang extracts noun phrases from text content, and compares them to terms of an ontology by performing phrase mapping using a rule-based algorithm such as WSD (i.e., first generative neural network) [0018].
Dang and Zhang teach claim 6, where Zhang teaches an RTE system for natural language processing (i.e., second generative neural network) that performs lexical, syntactic and semantic analysis of an input text in XML format (Zhang: sec. III.A, para. 1), and raises a hypothesis inferred from that (Zhang: Abstract).
Dang and Zhang do not disclose this claim; however, Lewis builds Retrieval-Augmented-Generation models (i.e., third generative neural network) where the parametric memory is a pre-trained seq2seq transformer, and the non-parametric memory is a dense vector index of Wikipedia, accessed as an external knowledge source (i.e., external database) (Lewis: sec. 1, para. 4) with a pre-trained neural retriever (Lewis: sec. 1, para. 2), to improve performance of knowledge-intensive NLP (i.e., contextual information) tasks.
Therefore, it would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of Lewis to Dang and Zhang. One having ordinary skill in the art would have found motivation to incorporate the RAG models of Lewis in the RTE system of Zhang to improve semantic tagging performance in Dang.
Claim 17 is rejected under 35 U.S.C. 103 as being unpatentable over Dang as applied to claim 14 above, in view of Zhang, and further in view of Bloomberg et al. US patent 5,825,919 [herein “Bloomberg”].
Claim 17 recites “The non-transitory computer-readable storage medium of claim 14, wherein the data object is image data and the location corresponds to a representation of an object within the image data.”
Dang maps text documents (i.e., data objects) to ontological entities/concepts and generate semantic keywords or key-phrases as document metadata [0165].
Dang teaches claim 14, but does not disclose this claim; however, Bloomberg constructs bounding boxes (i.e., locations) of words or text lines to recognize user-defined keywords in a bitmap image (i.e., representation of data object) (Bloomberg: 2:42-46).
Therefore, it would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of Bloomberg to Dang. One having ordinary skill in the art would have found motivation to incorporate keyword recognition in images as Bloomberg such that they can be tagged and searchable in Dang.
Claim 18 is rejected under 35 U.S.C. 103 as being unpatentable over Dang as applied to claim 14 above, in view of Zhang, and further in view of Toub et al. US patent application 2007/0027844 [herein “Toub”].
Claim 18 recites “The non-transitory computer-readable storage medium of claim 14, wherein the data object is an audio recording and the location corresponds to a position of a sound clip within the audio recording.”
Dang maps text documents (i.e., data objects) to ontological entities/concepts and generate semantic keywords or key-phrases as document metadata [0165].
Dang teaches claim 14, but does not disclose this claim; however, Toub searches for keywords in multimedia content (i.e., audio recording) segmented into chapters, scenes, clips, songs, images and other pre-defined audio/video segments (Toub: [0004]), by indexing metadata transcribed from dialog, monolog, lyrics, or other words within the content files (i.e., data objects) (Toub: [0014]), and then jumping to those positions in the content where keyword match occurs (Toub: [0024]).
Therefore, it would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of Toub to Dang. One having ordinary skill in the art would have found motivation to incorporate keyword recognition in audio recordings as Toub such that they can be tagged and searchable in Dang.
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 SHELLY X. QIAN whose telephone number is (408)918-7599. The examiner can normally be reached Monday - Friday 8-5 PT.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Tony Mahmoudi can be reached at (571)272-4078. 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.
/SHELLY X QIAN/Examiner, Art Unit 2163
/ALEX GOFMAN/Primary Examiner, Art Unit 2163