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 .
Continued Examination Under 37 CFR 1.114
2. 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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicants’ submission filed on February 26, 2026 has been entered.
Response to Amendment
3. This Office Action is in response to the applicant request for continued examination filed on February 11, 2026.
4. Claims 1-20 are pending. Claims 1, 11, and 18 are in independent form.5. Claims 1, 11, and 18 are amended.
Response to Arguments
6. Applicant’s arguments, see “Rejection Under 35 U.S.C. §101”, filed February 26, 2026. Upon reconsideration in view of Applicant’s amendment and arguments, the
rejection of claims 1-21 under 35 U.S.C. §101 is withdrawn.
7. Applicant’s arguments, see “Rejections under 35 U.S.C. § 103”, filed on July 28,
2026, have been carefully 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 § 103
8. 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.
9. 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.
10. Claims 1-4, 6-7, 10-13, 15-16, and 18-20 are rejected under 35 U.S.C. 103 as being unpatentable over Kryscinski et al. US20210124876A1 (hereinafter Kryscinski), in view of Choubey et al. US20230119109A1 (hereinafter Choubey) further in view of Madisetti et al. US20250190460A1 (hereinafter Madisetti).
Regarding claim 1, Kryscinski discloses a method comprising: deriving content from a collection of data sources to create synthetic chunks, each synthetic chunk being derived from one or more source documents and not merely an excerpt from the one or more source documents (Kryscinski [0019] e.g., “In extractive summarization, the model directly copies the salient parts of the source document into the summary. In abstractive summarization, the important parts of a source document are paraphrased to form novel sentences” This expressly distinguishes copying source excerpts from creating new. See also [0019] e.g., “In abstractive summarization, the important parts of a source document are paraphrased to form novel sentences.”. see also [0023] e.g., “… dataset examples are created by first sampling single sentences … from the source documents. The claims then pass through a set of textual transformations that output novel sentences”. The resulting abstractive summary/novel sentence is content derived from source-documents content but is not merely an excerpt because it is paraphrased/generated into a novel sentence rather than directly copied. See also [0020] e.g., “… a factually consistent summary should contain only statements that are entailed by the source document.”. It further teaches [0025] e.g., “… 1) identifying whether sentences remain factually consistent after transformation. 2) extracting a span in the source documents to support the consistency prediction” and its claim 1 expressly requires: “classifying … whether the corresponding text summary is factually consistent with the source text document” and, when consistent [Abstract] e.g., “… a span in the source text document that supports the corresponding text summary”. Thus, Kryscinski establishes the relationship between synthetically/obstructively generated content and the corresponding source content that supports it). Kryscinski does not expressly disclose generating confidence values for the synthetic chunks. the confidence values indicating a degree of confidence that the synthetic chunks have been accurately derived from corresponding portions of content present in the collection of data sources; and identifying, by the RAG assistant, a reliable subset of synthetic chunks from the subset by discarding, from the subset, one or more of the synthetic chunks of the subset for which the respective confidence value is below a threshold. Choubey discloses generating confidence values for the synthetic chunks. the confidence values indicating a degree of confidence that the synthetic chunks have been accurately derived from corresponding portions of content present in the collection of data sources (Choubey [0030] e.g., “… each of the plurality of summaries is associated with a respective first score indicative of a first factual quality, and a respective second score indicative of a second factual quality… the score associated with the plurality of summaries may be a score based on a metric … in other aspects the system determines the score.”. The respective factual-quality score teaches or at minimum suggests the claimed confidence value because it represents the factual quality/accuracy of generated summary content relative to its corresponding source document); and identifying, by the RAG assistant, a reliable subset of synthetic chunks from the subset by discarding, from the subset, one or more of the synthetic chunks of the subset for which the respective confidence value is below a threshold (Choubey [0031] e.g., “At step 210, the system filters the training dataset by removing summaries with the respective first scores below a first predetermined threshold resulting in a first training data subset. … the dataset may be filtered for the goal of lower intrinsic hallucinations by only including summaries which contain no entailment errors according to some metric”. See also [0032] e.g., “At step 215, the system filters the training dataset by removing summaries with the respective second scores below a second predetermined threshold resulting in a second training data subset”. The resulting subset can therefore contain only summaries satisfying the desired factual-quality criterion). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the abstractive summarization technique of Kryscinski to incorporate the factual-quality scoring and threshold-based filtering taught by Choubey. Kryscinski teaches generating abstractive summaries in which information from a source document is paraphrased to form novel sentences rather than merely copied from a source and further teaches determining whether generated summary content is factually consistent with corresponding source document. Choubey similarly addresses factual reliability generated summaries and removing summaries whose respective scores fall below a predetermined threshold. A person of ordinary skill in the art would have been motivated to apply the factura-quality scoring and filtering of Choubey to the abstractive summaries of Kryscinski in order to quantitively evaluate the reliability of synthetically generated content relative to its source and exclude generated content determined to have insufficient factual quality, thereby reducing factual inconsistencies and hallucination in the resulting generated content. The combined teachings of Kryscinski and Choubey does not explicitly disclose: populating an index for a retrieval augmented generation (RAG) assistant with the synthetic chunks; identifying, extracting, by the RAG assistant, a subset of the synthetic chunks within the index as relevant to a query; generating, by the RAG assistant, a large language model (LLM) prompt that includes the reliable subset of the synthetic chunks from the index and the query; providing the LLM prompt to an LLM; and generating a response to the query based on output of the LLM. Madisetti populating an index for a retrieval augmented generation (RAG) assistant with the synthetic chunks (Madisetti [0183] e.g., “At step 3912 the processed and tagged chunks are indexed in a vector/graph database and/or a full-text search engine (such as, for example, Elasticsearch or Solr), enabling efficient retrieval. Indexing can be done in a combination of databases (vector/graph) and full-text search engines to enable efficient hybrid search at a later stage.”. This directly teaches putting processed chunks into an index for subsequent retrieval by the SCORE-RAG system. Thus, the reference teaches populating an index used by SCORE-RAG system with processed document chunks for subsequent retrieval); identifying, extracting, by the RAG assistant, a subset of the synthetic chunks within the index as relevant to a query (Madisetti [0183] e.g., “Upon receiving a query, the system initiates the retrieval process to find relevant information. … the system performs a hybrid search at step 3924 to retrieve relevant information.”. see also [0170] e.g., “The refined documents 3428 may then be taken by the SCORE-RAG system 3430 which may, in response to a query … carry out operations on ranking and scoring those chunks that are most relevant to a particular query … “. SCORE-RAG indexes chunks and, upon receiving a query, searches/ranks those chunks to retrieve the relevant subset); generating, by the RAG assistant, a large language model (LLM) prompt that includes the reliable subset of the synthetic chunks from the index and the query (Madisetti [0156] e.g., “These retrieval splits 2818 and relevant associated metadata and processing results are then sent to the LLM as the context information along with the query as a prompt 2814. The prompt 2814 comprises the user query 2802 and the retrieval splits 2818”. This is very strong for the claimed RAG prompt structure: retrieved information + query [Wingdings font/0xE0] LLM prompt; providing the LLM prompt to an LLM (Madisetti [0156] e.g., “These retrieval splits 2818 and relevant associated metadata and processing results are then sent to the LLM as the context information along with the query as a prompt 2814.”); and generating a response to the query based on output of the LLM (Madisetti [0156] e.g., “The LLM 2816 generates the answer 2812 based on the query 2802 and context information in the prompt 2814.”. See also [0226] e.g., “The generation module 3822 (from FIG. 40 ) utilizes the retrieved superchunks to generate a response to the user query.”. This is direct teaching of generating the answer/response using the LLM based on the query and retrieved context). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate the abstractive content-generation and source-consistency teachings of Kryscinski, as modified by the factual-quality scoring and threshold-based filtering teachings modify the abstractive summarization technique of Kryscinski to incorporate the factual-quality scoring and threshold-based filtering teachings of Choubey, into the RAG indexing, retrieval, and generation system taught by Madisetti. Madisetti teaches processing document chunks, indexing the chunks for retrieval, retrieving chunks relevant to a user query, supplying the retrieved chunks together with the query as a context in an LLM prompt, and generating an answer using the LLM. Kryscinski, teaches generating non-extractive abstractive content derived from source documents and evaluating whether such generated content is factually consistent with its source, while Choubey teaches assigning respective factual-quality scores to generated content and removing generated content having scores below predetermined thresholds. A person of ordinary skill in the art would have been motivated to use the source-grounded, factual-quality-filtered synthetic content of Kryscinski and Choubey as the indexed and retrieved content in the RAG system of Madisetti so that content determined to be insufficiently supported by its source would be excluded from the context supplied to the LLM, thereby improving the factual reliability and quality of the LLM-generated response.
Claims 11 and 19 incorporate substantively all the limitations of claim 1 in a system and one or more tangible processor-readable storage media form and are rejected under the same rationale.
Regarding claim 2, the rejection of claim 1 hereby incorporated by reference, Kryscinski, Choubey, and Madisetti discloses a method, wherein deriving the content from the collection of data sources comprises changing content of a data source or changing a format of the data source (Kryscinski [0047] e.g., “At a process 230, transform module 134 of data generation module 130 performs one or more text transformations T on the text or single sentences sampled from source documents S … the transformations generate novel claim sentences ... paraphrase transformation, entity and number swapping transformation, pronoun swapping data augmentation, sentence negation transformation, and injection of noise.”. For paraphrasing specifically, see [0048] e.g., “Paraphrasing: In a paraphrasing transformation, one or more sentences from a source document are rephrased … minor syntactic and lexical changes”. Transforming/rephrasing source-document text into a novel sentence plainly constitutes changing content and/or format of content from the data source).
Regarding claim 3, the rejection of claim 1 hereby incorporated by reference, Kryscinski, Choubey, and Madisetti discloses a method, further comprising: generating, for a synthetic chunk of the synthetic chunks, a respective annotation that associates the synthetic chunk with a respective portion of the content, the response comprising the respective annotation (Kryscinski [0067] e.g., “… , factual consistency module 150 extracts, highlights, or otherwise identifies a span in the source documents to support the consistency prediction … start and end indices for selection and transformation spans in the source document and claim sentence” and identifies the particular source-document span that supports the generated claim sentence. Its claims similarly recite: [claim 1] e.g., “… a span in the source text document that supports the corresponding text summary”. So conceptually we have generated summary/claim [Wingdings font/0xDF] associated supporting source span. This is close to: synthetic chunk, annotation associating it with corresponding source content).
Regarding claim 4, the rejection of claim 1 hereby incorporated by reference, Kryscinski, Choubey, and Madisetti discloses a method, further comprising generating an explanation for a first synthetic chunk of the synthetic chunks, the explanation describing a derivation of the first synthetic chunk from a respective portion of the content, wherein the response includes the explanation (Kryscinski [0026] e.g., “… the systems and methods of the present disclosure add specialized modules to the factual consistency model that explain which portions of both the source document and generated text summary are pertinent to the model's decision. … explanatory modules that augment the factual consistency model provide useful assistance to humans as they verify the factual consistency between a source document and generated summaries.”. see also [0067] e.g., “This embodiment of the model can be referred to as the factual consistency checking (FactCC) model” and identifies/highlights the source span supporting the generated summary).
Regarding claim 5, the rejection of claim 1 hereby incorporated by reference, Kryscinski, Choubey, and Madisetti discloses a method, wherein a select synthetic chunk of the synthetic chunks is representative of a table and deriving the content includes expanding a table within a data source of the collection of data sources to create an expanded table, wherein expanding the table includes adding one or more columns or rows storing information that is not explicit but implied by formatting of the table .
Regarding claim 6, the rejection of claim 1 hereby incorporated by reference, Kryscinski, Choubey, and Madisetti discloses a method, wherein a select synthetic chunk of the synthetic chunks is a translation of a text from a data source of the collection of data sources and deriving the content comprises at least translating the text from a first language to a second language (Kryscinski [0048] e.g., “… paraphrases are produced by backtranslation using Neural Machine Translation (NMT) systems … an original sentence in English language is translated to an intermediate, non-English language, and the translated back to English” with French, German, Chinese, Spanish, and Russian given as examples of intermediate language. The first translation operation itself satisfies the claimed: text in first language [Wingdings font/0xE0] translate into second language. And because Kryscinski’ s resulting transformed sentence is derived from the original source-document sentence).
Regarding claim 7, the rejection of claim 1 hereby incorporated by reference, Kryscinski, Choubey, and Madisetti discloses a method, wherein a select synthetic chunk of the synthetic chunks is a summarization of a text from a data source of the collection of data sources and deriving the content comprises at least summarizing the text (Madisetti [0191] e.g., “… derived forms 4224, such as a summary/abstract of a document 4226, a combination of summaries/chunks 4228, or a combination of multiple documents 4230”. Madisetti teaches a selected derived chunk comprising a summarization of content from a source document. See also Kryscinski [0045] e.g., “… an unannotated collection or set S of source documents”, see also [0047] e.g., “… transform module 134 of data generation module 130 performs one or more text transformations T on the text or single sentences sampled from source documents S… paraphrase transformation …” see also [0048] e.g., “… one or more sentences from a source document are rephrased…”).
(Canceled)
(Canceled)
Regarding claim 10, the rejection of claim 1 hereby incorporated by reference, Kryscinski, Choubey, and Madisetti discloses a method, wherein the response comprises a reference to a particular data source associated with a select synthetic chunk of the subset of the synthetic chunks (Madisetti [0152] e.g., “The data source 2600 comprises documents (such as company-specific documents or public documents)… related metadata is inferred”, see also [0163] e.g., “… public and private legal documents obtained from courts and other sources … legal citation…”, see also [0183] e.g., “… topic tags, citation information, and other relevant attributes…”. See also [0027] e.g., “… find the relevant subset of the documents being evaluated in Retrieval Augmented Generation (RAG) pipelines”. This is a good for retrieval of source document. See also [0030] e.g., “… advanced document (including chunk) processing, intelligent information retrieval, and adaptive response generation mechanisms”. Finaly, see [0183] e.g., “… the aggregate response or its derivation may be presented as a result through an API or an Ul…”).
Regarding claim 12, the rejection of claim 11 hereby incorporated by reference, Kryscinski, Choubey, and Madisetti discloses a system, further comprising: generating, for a synthetic chunk of the synthetic chunks, a respective annotation that associates the synthetic chunk with a respective portion of the content, the response comprising the respective annotation (Kryscinski [0067] e.g., “… , factual consistency module 150 extracts, highlights, or otherwise identifies a span in the source documents to support the consistency prediction … start and end indices for selection and transformation spans in the source document and claim sentence” and identifies the particular source-document span that supports the generated claim sentence. Its claims similarly recite: [claim 1] e.g., “… a span in the source text document that supports the corresponding text summary”. So conceptually we have generated summary/claim [Wingdings font/0xDF] associated supporting source span. This is close to: synthetic chunk, annotation associating it with corresponding source content).
Regarding claim 13, the rejection of claim 11 hereby incorporated by reference, Kryscinski, Choubey, and Madisetti discloses a system, further comprising generating an explanation for a first synthetic chunk of the synthetic chunks, the explanation describing a derivation of the first synthetic chunk from a respective portion of the content, wherein the response includes the explanation (Kryscinski [0026] e.g., “… the systems and methods of the present disclosure add specialized modules to the factual consistency model that explain which portions of both the source document and generated text summary are pertinent to the model's decision. … explanatory modules that augment the factual consistency model provide useful assistance to humans as they verify the factual consistency between a source document and generated summaries.”. see also [0067] e.g., “This embodiment of the model can be referred to as the factual consistency checking (FactCC) model” and identifies/highlights the source span supporting the generated summary).
Regarding claim 15, the rejection of claim 11 hereby incorporated by reference, Kryscinski, Choubey, and Madisetti discloses a system, wherein deriving the content from the collection of data sources to create a select data chunk of the synthetic chunks comprises translating a text from a first language to a second language (Kryscinski [0048] e.g., “… paraphrases are produced by backtranslation using Neural Machine Translation (NMT) systems … an original sentence in English language is translated to an intermediate, non-English language, and the translated back to English” with French, German, Chinese, Spanish, and Russian given as examples of intermediate language. The first translation operation itself satisfies the claimed: text in first language [Wingdings font/0xE0] translate into second language. And because Kryscinski’ s resulting transformed sentence is derived from the original source-document sentence).
Regarding claim 16, the rejection of claim 11 hereby incorporated by reference, Kryscinski, Choubey, and Madisetti discloses a system, wherein deriving the content from the collection of data sources to create a select synthetic chunk of the synthetic chunks comprises summiting a text in a data source and the select data chunk comprises a summary of the text (Madisetti [0191] e.g., “… derived forms 4224, such as a summary/abstract of a document 4226, a combination of summaries/chunks 4228, or a combination of multiple documents 4230”. Madisetti teaches a selected derived chunk comprising a summarization of content from a source document. See also Kryscinski [0045] e.g., “… an unannotated collection or set S of source documents”, see also [0047] e.g., “… transform module 134 of data generation module 130 performs one or more text transformations T on the text or single sentences sampled from source documents S… paraphrase transformation …” see also [0048] e.g., “… one or more sentences from a source document are rephrased…”).
17. (Cancelled)
Regarding claim 18, the rejection of claim 11 hereby incorporated by reference, Kryscinski, Choubey, and Madisetti discloses a system, wherein the response comprises a reference to a particular data source associated with a select synthetic chunk of the subset of the synthetic chunks (Madisetti [0152] e.g., “The data source 2600 comprises documents (such as company-specific documents or public documents)… related metadata is inferred”, see also [0163] e.g., “… public and private legal documents obtained from courts and other sources … legal citation…”, see also [0183] e.g., “… topic tags, citation information, and other relevant attributes…”. See also [0027] e.g., “… find the relevant subset of the documents being evaluated in Retrieval Augmented Generation (RAG) pipelines”. This is a good for retrieval of source document. See also [0030] e.g., “… advanced document (including chunk) processing, intelligent information retrieval, and adaptive response generation mechanisms”. Finaly, see [0183] e.g., “… the aggregate response or its derivation may be presented as a result through an API or an Ul…”), wherein the query answering system is further configured to perform operations comprising displaying, via a user interface, the user interface object (Madisetti [0183] e.g., “… metadata … including topic tags, citation information, and other relevant attributes. … the aggregate response or its derivation may be presented as a result through an API or an Ul.”, see also [0163] e.g., “The exemplary interface 3200 is designed for legal use case” That allows users to query “public and private legal documents obtained from courts and other sources”).
Regarding claim 20, the rejection of claim 19 hereby incorporated by reference, Kryscinski, Choubey, and Madisetti discloses a one or more tangible processor-readable storage media, further comprising: generating, for a synthetic chunk of the synthetic chunks, a respective annotation that associates the synthetic chunk with a respective portion of the content, the response comprising the respective annotation (Kryscinski [0067] e.g., “… , factual consistency module 150 extracts, highlights, or otherwise identifies a span in the source documents to support the consistency prediction … start and end indices for selection and transformation spans in the source document and claim sentence” and identifies the particular source-document span that supports the generated claim sentence. Its claims similarly recite: [claim 1] e.g., “… a span in the source text document that supports the corresponding text summary”. So conceptually we have generated summary/claim [Wingdings font/0xDF] associated supporting source span. This is close to: synthetic chunk, annotation associating it with corresponding source content).
11. Claims 5, 14, and 21 are rejected under 35 U.S.C. 103 as being unpatentable over Kryscinski et al. US20210124876A1 (hereinafter Kryscinski), in view of Choubey et al. US20230119109A1 (hereinafter Choubey) further in view of Madisetti et al. US20250190460A1 (hereinafter Madisetti). as applied to claims 1-4, 6-7, 10-13, 15-16, and 18-20 above, and further in view of Hays et al. US. 2014/0019437 A1 (hereinafter Hays).
Regarding claim 5, the rejection of claim 1 hereby incorporated by reference, Kryscinski, Choubey, and Madisetti does not clearly disclose a method, wherein a select synthetic chunk of the synthetic chunks is representative of a table and deriving the content includes expanding a table within a data source of the collection of data sources to create an expanded table, wherein expanding the table includes adding one or more columns or rows storing information that is not explicit but implied by formatting of the table. Hayes discloses wherein a select synthetic chunk of the synthetic chunks is representative of a table and deriving the content includes expanding a table within a data source of the collection of data sources to create an expanded table, wherein expanding the table includes adding one or more columns or rows storing information that is not explicit but implied by formatting of the table (Hays [0036] e.g., “The report definition 120 may be modified to support such calculations. For example, another column may be added to the report to calculate those overall total sales figures. … The column may have a row header labeled "Total" … compute the total sales for each distinct combination of Country, Region, and Year values”. This expressly teaches information that is implicit in the table/report structure. See also [0033] e.g., “The data scope may be specified explicitly or implicitly … the data scope may be implicitly specified via the context or location of the aggregation operation, .. in the report definition … “ This determines different information based on where the cells/expressions occur within the row/column hierarchy. See also [0035] e.g., “… the overall total sales figures in each year for a Country and Region combination are not provided in the report..” see also [0036] e.g., “The report definition 120 may be modified … another column may be added to the report to calculate those overall total sales figures.”. Hays teaches expanding a tabular report by adding a column containing information derived from implicit contextual/positional relationships represented by the organization of rows, columns, and cells). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the RAG document/chunk processing system of Madisetti to process tabular source content by expanding the table to include one or more additional rows or columns containing information derived from implicit contextual or positional relationships represented by the table, as taught by the Hayes reference. One would have been motivated to make such a modification to make information implicit in the organization and scope of the tabular data explicit and available for subsequent processing and retrieval, thereby facilitating more complete and useful processing of the source data by the RAG system. Such a modification would have amounted to the predictable use of a known table-processing technique according to its established function.
Regarding claim 14, the rejection of claim 11 hereby incorporated by reference, Kryscinski, Choubey, Madisetti, and Hays discloses a system, wherein a select synthetic chunk of the synthetic chunks is representative of a table and deriving the content includes expanding a table within a data source of the collection of data sources to create an expanded table, wherein expanding the table includes adding one or more columns or rows storing information that is not explicit but implied by formatting of the table (Hays [0036] e.g., “The report definition 120 may be modified to support such calculations. For example, another column may be added to the report to calculate those overall total sales figures. … The column may have a row header labeled "Total" … compute the total sales for each distinct combination of Country, Region, and Year values”. This expressly teaches information that is implicit in the table/report structure. See also [0033] e.g., “The data scope may be specified explicitly or implicitly … the data scope may be implicitly specified via the context or location of the aggregation operation, .. in the report definition … “ This determines different information based on where the cells/expressions occur within the row/column hierarchy. See also [0035] e.g., “… the overall total sales figures in each year for a Country and Region combination are not provided in the report..” see also [0036] e.g., “The report definition 120 may be modified … another column may be added to the report to calculate those overall total sales figures.”. Hays teaches expanding a tabular report by adding a column containing information derived from implicit contextual /positional relationships represented by the organization of rows, columns, and cells). The motivation for the proposed combination is maintained.
Regarding claim 21, the rejection of claim 1 hereby incorporated by reference, Kryscinski, Choubey, Madisetti, and Hays discloses a method, wherein a select synthetic chunk of the synthetic chunks is derived by one or more of: translating text from a first language to a second language, wherein the language model is trained on a corpus that includes a larger volume of text in the second language than the first language (Kryscinski [0048] e.g., “… paraphrases are produced by backtranslation using Neural Machine Translation (NMT) systems … an original sentence in English language is translated to an intermediate, non-English language, and the translated back to English” with French, German, Chinese, Spanish, and Russian given as examples of intermediate language. The first translation operation itself satisfies the claimed: text in first language [Wingdings font/0xE0] translate into second language. And because Kryscinski’ s resulting transformed sentence is derived from the original source-document sentence); expanding a table within a data source of the collection of data sources to create an expanded table, wherein expanding the table includes adding one or more columns or rows storing information that is not explicit but implied by formatting of the table (Hays [0036] e.g., “The report definition 120 may be modified to support such calculations. For example, another column may be added to the report to calculate those overall total sales figures. … The column may have a row header labeled "Total" … compute the total sales for each distinct combination of Country, Region, and Year values”. This expressly teaches information that is implicit in the table/report structure. See also [0033] e.g., “The data scope may be specified explicitly or implicitly … the data scope may be implicitly specified via the context or location of the aggregation operation, .. in the report definition … “ This determines different information based on where the cells/expressions occur within the row/column hierarchy); or summarizing text from a data source of the collection of data sources to create a data chunk smaller than input prompt limits of the language model. The motivation for the proposed combination is maintained.
Conclusion
12. Any inquiry concerning this communication or earlier communications from the examiner should be directed to BERHANU MITIKU whose telephone number is (571)270-1983. The examiner can normally be reached Monday – Friday 8:30AM – 4:00PM.
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, Ajay Bhatia can be reached at 571-272-3906. 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.
/BERHANU MITIKU/Examiner, Art Unit 2156
/AJAY M BHATIA/Supervisory Patent Examiner, Art Unit 2156
Prosecution Insights