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 Amendment
The amendments filed 06/15/2026 have been accepted and considered in this office action. Claims 1-3, 5-13, 15-17, 19-20 have been Amended. Claim 14 has been cancelled. Claim 21 has been added. Claims 1-13, 15-21 are pending.
Response to Arguments
Applicant’s arguments regarding the 35 U.S.C. 101 rejection with respect to the 35 U.S.C. 101 rejections have been considered and are persuasive after amendment the 35 U.S.C. 101 rejections have been withdrawn. The amended independent claims now recite specific machine learning operations, including computing an activation pattern for a profile query model and executing the profile query model according to that activation pattern to generate a response. These limitations are not practically performed in the human mind and, when considered as an ordered combination with the claimed verification operations, sufficiently integrate any recited abstract idea into a practical application. Accordingly, the claims are no longer considered directed to a judicial exception without significantly more.
Applicant’s arguments regarding the 35 U.S.C 103 rejections with respect to claim(s) 1-20 have been considered but are moot in view of new grounds of rejection necessitated by the applicant’s amendments to the claims.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1, 2, 4, 5, 7, 9, 12, 15, 17, 19 are rejected under 35 USC 103 as being unpatentable over Gampa et al. (hereinafter Gampa) (US 20240211477 A1) in view of Lee et al. (hereinafter Lee) (US 20230134852 A1) in further view of Gao et al. (hereinafter Gao) (PAL: Program-aided Language Models) and Gou et al. (hereinafter Gou) (CRITIC: LARGE LANGUAGE MODELS CAN SELF CORRECT WITH TOOL-INTERACTIVE CRITIQUING).
Regarding claim 1, Gampa discloses:
A system for interfacing with data profilers using a machine learning model, the system comprising (Gampa, P[0035]-P[0039], teaches AI data processing system connected to internal/external data sources, historical communications, databases, and machine learning models, and an AI processor to process communications, execute queries, degenerate predictions, etc.):
one or more processors (Gampa, P[0036] AI processor 140); and
one or more non-transitory, computer-readable media comprising instructions that, when executed by the one or more processors, cause operations comprising (Gampa, P[0005]):
receiving, via an input/output path of a conversational program, a user query from a user device and retrieving, from memory, one or more data profile attributes of associated with the user query (Gampa, P[0022], P[0026]-P[0028], teaches receiving natural language request for information from a client device through input/output or messaging interface, identifying information stored in unified data types, and using query metadata, user data, and interface information in processing the request, P[0064]-P[0066], teaches retrieving stored unified data types having associated identifiers features, context, semantics, type/classification and historical information responsive to a query);
executing a language interpretation model, to computationally pre-process the user query and the one or more data profile attributes to compute an activation pattern for the a profile query model (Gampa, P[0050]-P[0052], teaches parsing user's question to determine query type, requested information, interface type and contextual features, identifying responsive unified data types using confidence values, and translating the selected information for downstream response generation, mapping under BRI the query/profile-derived operational configuration or "activation pattern"), and wherein the profile query model is trained to generate text-based responses to queries based on training data comprising data profile attributes and previous user queries (Gampa, P[0024]-P[0025], teaches formatting feature vectors from historical communications and a dataset and training a machine learning model with those vectors to generate natural language communications responsive to stored information, P[0071]-P[0073], teaches training response-generation models using historical communications and queries and generating responses from stored unified-data representations);
and
transmitting the verified response in to the user device via the conversational program in response to the user query (Gampa, P[0030], teaches returning/transmitting the generated natural language response through the conversational/interface path to the client device in response to the received request).
Gampa does not explicitly disclose:
wherein the language interpretation model is trained to produce real-valued embeddings associated with the user query
executing the profile query model to process the real-valued embeddings and the one or more data profile attributes according to the activation pattern to computationally generate a preliminary response to the user query
computing, independently of the profile query model, an expected result based on the one or more data profile attributes
executing a verification program to construct a verified response based on the preliminary response, wherein the verification program compares the expected result against a reported result extracted from the preliminary response to trigger a modification of the preliminary response for increasing a factual accuracy of the verified response based on the expected result indicated by an accuracy metric;
However, Lee discloses:
wherein the language interpretation model is trained to produce real-valued embeddings associated with the user query (Lee, P[0035]-P[0037], P[0056]-P[0059], teaches an embedding as a mathematical valued latent vector and a language model based vector encoder trained to convert text having similar meanings into similar embedding vectors, the semantic phrase parsed from the user's query is converted into first embedding vector)),
executing the profile query model to process the real-valued embeddings and the one or more data profile attributes according to the activation pattern to computationally generate a preliminary response to the user query (Lee, P[0059]-P[0065], teaches converting semantic content of the user query into an embedding vector and processing that query embedding against stored data embeddings and associated metadata to identify information responsive to the query, thus, using the real-valued query representation together with stored profile/metadata information to computationally produce the preliminary query responsive result));
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Gampa in view of Lee. Doing so would have combined Lee’s trained query embedding technique (Lee, Abstract) with Gampa’s conversational data-response system (Gampa, Abstract) to predictable improve semantic interpretation and retrieval of query-relevant profiled data.
The combination of Gampa and Lee does not explicitly disclose:
computing, independently of the profile query model, an expected result based on the one or more data profile attributes
executing a verification program to construct a verified response based on the preliminary response, wherein the verification program compares the expected result against a reported result extracted from the preliminary response to trigger a modification of the preliminary response for increasing a factual accuracy of the verified response based on the expected result indicated by an accuracy metric;
However, Gao discloses:
computing, independently of the profile query model, an expected result based on the one or more data profile attributes (Gao, Page 1 upper middle, Abstract, Page 3 lower middle, teaches the language model generating the programmatic solution while delegating actual computation to a separate Python interpreter/solver, which executes the program using the supplied problem data and returns the independently calculated result, mapping the claimed independently computed expected result);
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Gampa in view of Lee and Gao. Doing so would have combined the independent programmatic computation of Gao (Gao, Abstract) with the query-embedding of Lee (Lee, Abstract) with Gampa’s conversational data-response system (Gampa, Abstract) would have predictably improved semantic interpretation of the query while providing an independent computed result from the relevant data.
The combination of Gampa, Lee, and Gao does not explicitly disclose:
executing a verification program to construct a verified response based on the preliminary response, wherein the verification program compares the expected result against a reported result extracted from the preliminary response to trigger a modification of the preliminary response for increasing a factual accuracy of the verified response based on the expected result indicated by an accuracy metric
However, Gou discloses:
executing a verification program to construct a verified response based on the preliminary response, wherein the verification program compares the expected result against a reported result extracted from the preliminary response to trigger a modification of the preliminary response for increasing a factual accuracy of the verified response based on the expected result indicated by an accuracy metric (Gou, P 3, middle, Fig 2, P. 4 upper, P 23, teaches taking an initial generated answer, obtaining independent tool-supported evidence/computational results, checking the reported answer against that evidence, determining whether the answer is correct, and correcting the initial answer when verification identifies an error, Gou additionally teaches a probability of true confidence score for evaluating correctness, mapping the claimed accuracy metric used in determining factual accuracy);
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Gampa in view of Lee, Gao, and Gou. Doing so would have predictable added Gou’s verification/correction technique to the system of Lee, Gao, and Gampa, which would predictably permit the independently computed result to be compared with the generated response and used to correct factual inaccuracies.
Regarding claim 2, Gampa discloses:
A method comprising:
in connection with receiving, via an input/output path, a user query, retrieving one or more data profile attributes from memory based on the user query (Gampa, P[0022]-P[0023], P[0064]-P[0066], teaches receiving a request for information through an input/output or communication interface, processing the query to identify stored unified data types responsive to the request, and retrieving stored data having associated features, context, semantics, classifications, and metadata, mapping retrieval of profile attributes from memory based on the query);
executing the profile query model to process the user query and the one or more data profile attributes according to the activation pattern to generate a preliminary response (Gampa, P[0028]-P[0029], teaches processing the communication to identify responsive stored information and then executing a machine-learning model using the processed communication, unified data type, interface information, and user data to generate a natural-language response, mapping query and profile attributes processed according to the determined configuration to generate the preliminary response);
and
transmitting the verified response to a user device (Gampa, P[0030], teaches transmitting the generated response through the first interface to the client/user device)
Gampa does not explicitly disclose:
executing an interpretation model to process the user query and the one or more data profile attributes to compute an activation pattern for a profile query model, wherein the interpretation model is trained with historical data profile attributes to produce real-valued embeddings associated with user queries;
computing, independently of the profile query model, an expected result based on the one or more data profile attributes;
executing a corrective program that determines a comparison result based on the expected result and preliminary response to determine a verified response based on the preliminary response;
However, Lee discloses:
executing an interpretation model to process the user query and the one or more data profile attributes to compute an activation pattern for a profile query model, wherein the interpretation model is trained with historical data profile attributes to produce real-valued embeddings associated with user queries (Lee, P[0052], P[0056]-P[0065], teaches semantically parsing a user query using a trained language model based vector encoder to convert the parsed query into a numerical embedding vector, and using that vector with stored data/metadata to control downstream query processing and under BRI this embedded derived downstream processing corresponds to claimed activation pattern);
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Gampa in view of Lee. Doing so would have combined Lee’s trained query embedding technique (Lee, Abstract) with Gampa’s conversational data-response system (Gampa, Abstract) to predictable improve semantic interpretation and retrieval of query-relevant profiled data.
The combination of Gampa and Lee does not explicitly disclose:
computing, independently of the profile query model, an expected result based on the one or more data profile attributes;
executing a corrective program that determines a comparison result based on the expected result and preliminary response to determine a verified response based on the preliminary response; However, Gao discloses:
computing, independently of the profile query model, an expected result based on the one or more data profile attributes (Gao, Page 1, upper middle, Abstract, page 3 lower middle, teaches the language model generating programmatic reasoning while a separate python interpreter solver executes the generated program using the supplied data and independently produces the result));
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Gampa in view of Lee and Gao. Doing so would have combined the independent programmatic computation of Gao (Gao, Abstract) with the query-embedding of Lee (Lee, Abstract) with Gampa’s conversational data-response system (Gampa, Abstract) would have predictably improved semantic interpretation of the query while providing an independent computed result from the relevant data.
The combination of Gampa, Lee, and Gao does not explicitly disclose:
However, Gou discloses:
executing a corrective program that determines a comparison result based on the expected result and preliminary response to determine a verified response based on the preliminary response (Gou, Page 3 middle, Fig 2, Page 4 upper, teaches taking an initial response, checking it against independently obtained tool evidence/results, determining whether the initial answer is correct and generating the corrected/verified answer when the comparison reveals an error);
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Gampa in view of Lee, Gao, and Gou. Doing so would have predictable added Gou’s verification/correction technique (Gou, Abstract, Page 3) to the system of Lee, Gao, and Gampa (Lee, Abstract, Gao, Abstract, Gampa, Abstract and P[0028]-P[0029]), which would predictably permit the independently computed result to be compared with the generated response and used to correct factual inaccuracies.
Regarding claim 4, the combination of Gampa, Lee, Gao, and Gou disclose the method of claim 2.
The combination further discloses:
The method of claim 2, wherein the interpretation model is a language processing model trained to correspond user queries to embeddings used as input to the profile query model (Lee, P[0052], P[0056]-P[0065], teaches language processing/vector encoding model trained to convert semantic content parsed from user queries into embedding vectors and using those query embeddings as input to the downstream search/query processing operation).
Regarding claim 5, the combination of Gampa, Lee, Gao, and Gou disclose the method of claim 2.
The combination further discloses:
wherein the one or more data profile attributes are stored in association with a data profiler, and wherein the profile query model is trained to use embeddings as input (Lee, P[0056]-P[0065], embedding as input mechanism) to generate a text-based response to queries related to the data profiler (Gampa, P[0019]-P[0020], P[0024]-P[0029], teaches stored unified data representations containing semantic information, features and metadata, training machine learning response models, using feature vectors/historical communications, and executing the trained model using a processed query and stored unified data to generate a natural language response).
Regarding claim 7, the combination of Gampa, Lee, Gao, and Gou disclose the method of claim 2.
wherein the corrective program compares mathematical formulae output by the interpretation model against the preliminary response to determine accuracy, comprising (Gou, Page 3 middle, Fig 2, teaches comparing the substantive answer in an initial generated response against independently determined computational/factual evidence to determine whether the response is accurate, Gao supplies the interpreted mathematical formulae):
using the mathematical formulae and a plurality of data profiles, generating an expected result (Gao, Page 3, middle/lower , teaches executing the mathematical/programmatic solution produced form the interpreted problem using a separate solver and the supplied problem data to obtain the expected result);
extracting a reported result from an embedding of the preliminary response (Lee, P[0035]-P[0037], P[0056]-P[0059], teaches representing semantic natural-language content as mathematical embedding vectors in combination with Gou's extraction/identification of the substantive answer from the initial response, the embedding representations supplies the claimed source representation); and
comparing the expected result against the reported result to determine a measure of accuracy (Gou, Page 3, middle, Fig 2, Page 23, teaches checking the reported answer against independently obtained evidence/result to determine correctness and additionally describes probability of True as a confidence/correctness score, mapping the claimed measure of accuracy).
Regarding claim 9, the combination of Gampa, Lee, Gao, and Gou disclose the method of claim 2.
further comprising training the interpretation model, comprising:
generating a training dataset, comprising data profile descriptions in plain text (Gampa, P[0024]-P[0025], teaches historical natural-language communications and dataset information being used to define feature-vector training data for machine learning models that generate natural language responses));
training the interpretation model using a language processing algorithm to generate real-valued embeddings corresponding to input text tokens (Lee, P[0056]-P[0059], teaches training a language model based vector encoder to map natural language semantic input to numerical latent embedding vectors); and
based on the real-valued embeddings, training the interpretation model to correlate the real-valued embeddings to activation patterns (Lee, P[0057]-P[0065], teaches learning semantic relationships in embedding space and using those vector relationships to determine corresponding downstream search processing/results, mapping under BRI the learned relationship between embeddings and downstream activation patterns).
Regarding claim 12, claim 12 recites the non-transitory computer-readable media corresponding to claim 2 and is rejected for the same reasons as above
Gampa further discloses: One or more non-transitory computer-readable media comprising instructions that, when executed by one or more processors, cause operations comprising (Gampa, P[0005]):
Regarding claim 15, claim 15 recites the non-transitory computer-readable media corresponding to claim 5 and is rejected for the same reasons as above
Regarding claim 17, claim 17 recites the non-transitory computer-readable media corresponding to claim 7 and is rejected for the same reasons as above
Regarding claim 19, claim 19 recites the non-transitory computer-readable media corresponding to claim 9 and is rejected for the same reasons as above
Claims 3 and 13 are rejected under 35 USC 103 as being unpatentable over Gampa et al. (hereinafter Gampa) (US 20240211477 A1) in view of Lee et al. (hereinafter Lee) (US 20230134852 A1) in further view of Gao et al. (hereinafter Gao) (PAL: Program-aided Language Models), Gou et al. (hereinafter Gou) (CRITIC: LARGE LANGUAGE MODELS CAN SELF CORRECT WITH TOOL-INTERACTIVE CRITIQUING), and Liang et al. (hereinafter Liang) (TaskMatrix.AI: Completing Tasks by Connecting Foundation Models with Millions of APIs).
Regarding claim 3, the combination of Gampa, Lee, Gao, Gou discloses the method of claim 2.
The combination further discloses:
The method of claim 2, wherein executing the interpretation model to compute the activation pattern for the profile query model comprises:
determining supplemental data, wherein the supplemental data is computed based on a plurality of data profiles (Gampa, P[0018]-P[0020], teaches augmenting stored unified data representations with additional features derived from received data, its source, associated user, context, etc. and providing supplemental information computed form stored data representations for subsequent processing);
determining a mathematical formula from a set of mathematical formulae, wherein the mathematical formula is selected for applicability to the user query (Gao, Page 3, middle portion, teaches interpreting the natural language problem to determine the mathematical/programmatic operations appropriate to that particular query and supplying those operations to an external solver).
The combination does not explicitly disclose:
determining an algorithm for the profile query model;
However, Liang discloses: determining an algorithm for the profile query model (Liang, Page 4, upper-middle portion, teaches selecting the API/model whose functionality satisfies the interpreted task and executing the corresponding algorithm)
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Gampa in view of Lee, Gao, Gou, Liang. Doing so will provide the task specific API/model selection of Liang (Liang, Abstract) with Gampa’s conversational data-response system (Gampa, Abstract), Lee’s semantic query embeddings (Lee, Abstract), Gao’s independent programmatic computation (Gao, Abstract), and Gou’s verification/correction to predictable enable selection of the appropriate processing functionality (Gou, Abstract), improve query interpretation, independently computer an expected result, and correct inaccurate generated responses.
Regarding claim 13, claim 13 recites the non-transitory computer-readable media corresponding to claim 3 and is rejected for the same reasons as above
Claims 6, 10, 11, 16 and 20 are rejected under 35 USC 103 as being unpatentable over Gampa et al. (hereinafter Gampa) (US 20240211477 A1) in view of Lee et al. (hereinafter Lee) (US 20230134852 A1) in further view of Gao et al. (hereinafter Gao) (PAL: Program-aided Language Models), Gou et al. (hereinafter Gou) (CRITIC: LARGE LANGUAGE MODELS CAN SELF CORRECT WITH TOOL-INTERACTIVE CRITIQUING), and Kovachev et al. (hereinafter Kovachev) (US 20240078248 A1).
Regarding claim 6, the combination of Gampa, Lee, Gao, and Gou discloses the method of claim 2.
The combination further discloses:
based on the measure of correspondence, determining a feasibility score of the activation pattern (Gou, Page 4, upper, Page 5, upper, teaches evaluating an output for criteria including feasibility and using the verification determination to decide whether the initial response is accepted or corrected)); and
determining the verified response based on the feasibility score (Gou, page 4 upper, see above mapping)).
The combination does not explicitly disclose:
determining a null type and a null value distribution associated with a dataset represented by the one or more data profile attributes;
based on the null type and the null value distribution, determining a measure of correspondence with the activation pattern, wherein the measure of correspondence indicates an extent of null values and null types in data required to complete the activation pattern
However, Kovachev discloses:
wherein executing the corrective program comprises:
determining a null type and a null value distribution associated with a dataset represented by the one or more data profile attributes (Kovachev, P[0080]-P[0083], P[0093]-P[0095], teaches profiling database columns according to data type and calculating null/missing value counts, percentages, distributions, and statistical representations for profiled dataset));
based on the null type and the null value distribution, determining a measure of correspondence with the activation pattern, wherein the measure of correspondence indicates an extent of null values and null types in data required to complete the activation pattern (Kovachev, P[0080]-P[0083], teaches quantitatively determining null counts/percentages and data types for selected data features thus, quantifying the extent to which data required for the configured profiling operation is absent, selected/configured profiling operation can be read as claimed activation pattern under BRI);
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Gampa in view of Lee, Gao, Gou, Kovachev. Doing so will provide the statistical data-profiling techniques of Kovachev with Gampa’s conversational data-response system (Gampa, Abstract), Lee’s semantic query embeddings (Lee, Abstract), Gao’s independent programmatic computation (Gao, Abstract), and Gou’s verification/correction to predictably provide dataset statistics and metadata useful for computing and verifying data-responsive results.
Regarding claim 10, the combination of Gampa, Lee, Gao and Gou disclose the method of claim 2.
The combination does not explicitly disclose:
wherein the one or more data profile attributes comprise metadata attributes generated for a feature of a dataset, and wherein computing the expected result comprises selecting, based on the activation pattern, one or more of a range, median, and quartile values of the dataset, a standard deviation and skewness values of the dataset, and a variable type for each feature in the dataset However, Kovachev discloses:
wherein the one or more data profile attributes comprise metadata attributes generated for a feature of a dataset, and wherein computing the expected result comprises selecting, based on the activation pattern, one or more of a range, median, and quartile values of the dataset, a standard deviation and skewness values of the dataset, and a variable type for each feature in the dataset (Kovachev, P[0078]-P[0083], P[0087], P{0093]-P[0097], teaches configuration driven profiling of selected dataset columns/features and computing/storing metadata including min, max ranges and q1/q2/q3 quartiles/median, standard deviation, skewness, and data/variable type)).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Gampa in view of Lee, Gao, Gou, Kovachev. Doing so will provide the statistical data-profiling techniques of Kovachev with Gampa’s conversational data-response system (Gampa, Abstract), Lee’s semantic query embeddings (Lee, Abstract), Gao’s independent programmatic computation (Gao, Abstract), and Gou’s verification/correction to predictably provide dataset statistics and metadata useful for computing and verifying data-responsive results.
Regarding claim 11, the combination of Gampa, Lee, Gao and Gou disclose the method of claim 2.
The combination further discloses:
using the comparison result to determine the verified response (Gou, Page 4, upper portion, teaches using the verification determination to return the response when correct or generate corrected response when the comparison indicates an error)).
The combination does not explicitly disclose:
wherein executing the corrective program comprises:
comparing the expected result to a reported value in the preliminary response, the expected result being computed from a histogram, an inferred distribution, or a linear regression model identified by the one or more data profile attribute
However, Kovachev discloses:
wherein executing the corrective program comprises:
comparing the expected result to a reported value in the preliminary response, the expected result being computed from a histogram, an inferred distribution, or a linear regression model identified by the one or more data profile attributes (Kovachev, P[0092]-P[0098], teaches calculating and presenting distributions and category value-frequency information for dataset features, satisfying the histogram distribution alternative, comparison);
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Gampa in view of Lee, Gao, Gou, Kovachev. Doing so will provide the statistical data-profiling techniques of Kovachev with Gampa’s conversational data-response system (Gampa, Abstract), Lee’s semantic query embeddings (Lee, Abstract), Gao’s independent programmatic computation (Gao, Abstract), and Gou’s verification/correction to predictably provide dataset statistics and metadata useful for computing and verifying data-responsive results.
Regarding claim 16, claim 16 recites the non-transitory computer-readable media corresponding to claim 6 and is rejected for the same reasons as above
Regarding claim 20, the combination of Gampa, Lee, Gao and Gou disclose the non-transitory computer-readable medium of claim 12.
The combination further discloses:
generating the response based on the expected result (Gampa, P[0028]-P[0029], teaches using retrieved/processed stored data corresponding to the query as input to a trained model to generate the natural-language response)).
The combination does not explicitly disclose:
wherein the library produces one or more metadata attributes based on a dataset, the one or more metadata attributes comprising at least one of a number of null values in the dataset[[;]],a range, median, and quartile values of the dataset[[;]],a standard deviation and skewness values of the dataset[[;]], or a variable type for each feature in the dataset, the operations further comprising:
computing an expected result based on the one or more metadata attributes However, Kovachev discloses:
wherein the library produces one or more metadata attributes based on a dataset, the one or more metadata attributes comprising at least one of a number of null values in the dataset[[;]],a range, median, and quartile values of the dataset[[;]],a standard deviation and skewness values of the dataset[[;]], or a variable type for each feature in the dataset, the operations further comprising (Kovachev, P[0078] - P[0083], P[0087], P[0093]-P[0097], teaches data profiling system with null counts/percentages, min/max range, median/quartile values, standard deviation, skewness, and data/variable type):
computing an expected result based on the one or more metadata attributes (Kovachev, P[0080]-P[0084], teaches executing parameterized statistical profiling queries over dataset features to calculate results based on those metadata/statistical attributes))
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Gampa in view of Lee, Gao, Gou, Kovachev. Doing so will provide the statistical data-profiling techniques of Kovachev with Gampa’s conversational data-response system (Gampa, Abstract), Lee’s semantic query embeddings (Lee, Abstract), Gao’s independent programmatic computation (Gao, Abstract), and Gou’s verification/correction to predictably provide dataset statistics and metadata useful for computing and verifying data-responsive results.
Claims 8 and 21 are rejected under 35 USC 103 as being unpatentable over Gampa et al. (hereinafter Gampa) (US 20240211477 A1) in view of Lee et al. (hereinafter Lee) (US 20230134852 A1) in further view of Gao et al. (hereinafter Gao) (PAL: Program-aided Language Models), Gou et al. (hereinafter Gou) (CRITIC: LARGE LANGUAGE MODELS CAN SELF CORRECT WITH TOOL-INTERACTIVE CRITIQUING), and Lim et al. (hereinafter Lim) (US 20240086566 A1).
Regarding claim 8, the combination of Gampa, Lee, Gao, Gou discloses the method of claim 2.
The combination does not explicitly disclose:
wherein the one or more data profile attributes are stored in association with a data profiler, further comprising the corrective program using metadata of the data profiler to determine data integrity:
based on privacy metadata associated with a data profile, determining a disclosable dataset, wherein the disclosable dataset comprises data in the data profile suitable for answering the user query; and
modifying the preliminary response to contain only data from the disclosable dataset
However, Lim discloses:
wherein the one or more data profile attributes are stored in association with a data profiler, further comprising the corrective program using metadata of the data profiler to determine data integrity (Lim, P[0158]-P[0165], teaches using database-object, query, user/context, and policy metadata to evaluate whether requested data can properly be accessed and returned)):
based on privacy metadata associated with a data profile, determining a disclosable dataset, wherein the disclosable dataset comprises data in the data profile suitable for answering the user query (Lim, P[0164]-P[0168], P[0177]-P[0179], teaches selecting applicable privacy/access policies and determining the permitted rows, columns, or information that may be returned for the requesting operation)); and
modifying the preliminary response to contain only data from the disclosable dataset (Lim, P[0177[P[0182], teaches modifying results through filtering unauthorized rows, masking/redacting protected columns, limiting returned information, or blocking data before it is supplied to the requesting application).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Gampa in view of Lee, Gao, Gou, Kovachev. Doing so will provide the policy-based filtering of Lim (Lim, Abstract) with Gampa’s conversational data-response system (Gampa, Abstract), Lee’s semantic query embeddings (Lee, Abstract), Gao’s independent programmatic computation (Gao, Abstract), and Gou’s verification/correction to predictably restrict generated and verified responses to data authorized for disclosure.
Regarding claim 21, the combination of Gampa, Lee, Gao, Gou discloses the non-transitory computer-readable of claim 12.
The combination further discloses:
the operations further comprising: computing an expected result based on the one or more data profile attributes, wherein a corrective program uses the activation pattern computed by the interpretation model to select a mathematical formula for computing the expected result (Gao, Page 3, middle to lower, teaches interpreting the natural language problem to determine/generate the particular mathematical or programmatic operations applicable to that problem and executing those selected operations with an external solver to computer the result, mapping activation pattern controlled formula selection and independent expected result computation)); and
executing the corrective program to compare the expected result against the response for verifying factual accuracy (Gou, Page 3-5, Fig 2, teaches comparing an initial response against independently obtained evidence/results, determining factual correctness and correcting the response when verification reveals an error)
The combination does not explicitly disclose:
and confidentiality of the response to generate a verified response
However, Lim discloses:
and confidentiality of the response to generate a verified response (Lim, P[0177]-P[0185], teaches confidentiality-verification aspect).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Gampa in view of Lee, Gao, Gou, Kovachev. Doing so will provide the policy-based filtering of Lim (Lim, Abstract) with Gampa’s conversational data-response system (Gampa, Abstract), Lee’s semantic query embeddings (Lee, Abstract), Gao’s independent programmatic computation (Gao, Abstract), and Gou’s verification/correction to predictably restrict generated and verified responses to data authorized for disclosure.
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 SHASHIDHAR S MANOHARAN whose telephone number is (571)272-6772. The examiner can normally be reached M-F 8:00-4:00.
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, Andrew Flanders can be reached at 571-272-7516. 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.
/SHASHIDHAR SHANKAR MANOHARAN/ Examiner, Art Unit 2655
/ANDREW C FLANDERS/ Supervisory Patent Examiner, Art Unit 2655
Prosecution Insights