# US 12,182,508 B2 · Natural language question answering using non-relational tables > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,182,508 B2 - Document ID: US12182508B2 - Status: Granted - Kind: B2 - Application number: 17/656,487 - Filed: 2022-03-25 - Published: 2023-09-28 - Granted: 2024-12-31 - Earliest claimed: 2022-03-25 - Assignees: International Business Machines Corporation - Applicants: International Business Machines Corporation - Inventors: Vishwajeet Kumar; Jaydeep Sen; Samarth Bharadwaj; Saneem Ahmed Chemmengath; Ioannis Katsis; Mustafa Canim - Web page: https://patentgrep.com/patent/US12182508B2 ## CPC - G06F 16/243: {Natural language query formulation} - G06F 40/279: Recognition of textual entities - G06F 16/3329: Natural language query formulation - G06F 40/106: Display of layout of documents; Previewing - G06F 40/177: of tables; using ruled lines - G06F 40/258: Heading extraction; Automatic titling; Numbering - G06F 40/35: Discourse or dialogue representation - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract A question answering bot that digests non-relational data tables is provided. A processor receives a question regarding a non-relational data table. A processor extracts at least one feature of the question using a natural language processing (NLP) model. A processor extracts at least one similar feature of the non-relational data table to the extracted at least one feature of the question. A processor determines at least one relevant cell in the non-relational data table based on the at least one feature of the question and the at least one similar feature of the non-relational data table. A processor provides an answer to the question, where the answer is based on the at least one relevant cell. ## Claims ### Claim 1 A method comprising: receiving, by one or more processors, a question regarding a non-relational data table; generating, by the one or more processors, a schema based on a physical layout of the non-relational data table, wherein the schema includes one or more predicates that indicate the physical layout of the non-relational data table; extracting, by the one or more processors, at least one feature of the question using a natural language processing (NLP) model; extracting, by the one or more processors, at least one similar feature of the non-relational data table to the extracted at least one feature of the question; determining, by the one or more processors, at least one relevant cell in the non-relational data table based on the at least one feature of the question and the at least one similar feature of the non-relational data table; and providing, by the one or more processors, an answer to the question, wherein the answer is based on the at least one relevant cell. ### Claim 2 (depends on 1) The method of claim 1, the method further comprising: identifying, by the one or more processors, a similar non-relational data table to the non-relational data table; and extracting, by the one or more processors, a feature from the similar non-relational data table based on the schema generated based on the non-relational data table. ### Claim 3 (depends on 1) The method of claim 1, wherein the physical layout of the non-relational data table includes position information of one or more selected from a group consisting of: row and column headers, row and column subheaders, and cells. ### Claim 4 (depends on 1) The method of claim 1, wherein the non-relational data table includes unstructured data. ### Claim 5 (depends on 1) The method of claim 1, wherein the NLP model is a Bidirectional Encoder Representations from Transformers (BERT) model. ### Claim 6 A computer program product comprising: one or more computer-readable storage media and program instructions stored on the one or more computer-readable storage media, the program instructions comprising: program instructions to receive a question regarding a non-relational data table; program instructions to generate a schema based on a physical layout of the non-relational data table, wherein the schema includes one or more predicates that indicate the physical layout of the non-relational data table; program instructions to extract at least one feature of the question using a natural language processing (NLP) model; program instructions to extract at least one similar feature of the non-relational data table to the extracted at least one feature of the question; program instructions to determine at least one relevant cell in the non-relational data table based on the at least one feature of the question and the at least one similar feature of the non-relational data table; and program instructions to provide an answer to the question, wherein the answer is based on the at least one relevant cell. ### Claim 7 (depends on 6) The computer program product of claim 6, the one or more computer-readable storage media further storing: program instructions to identify a similar non-relational data table to the non-relational data table; and program instructions to extract a feature from the similar non-relational data table based on the schema generated based on the non-relational data table. ### Claim 8 (depends on 6) The computer program product of claim 6, wherein the physical layout of the non-relational data table includes position information of one or more selected from a group consisting of: row and column headers, row and column subheaders, and cells. ### Claim 9 (depends on 6) The computer program product of claim 6, wherein the non-relational data table includes unstructured data. ### Claim 10 (depends on 6) The computer program product of claim 6, wherein the NLP model is a Bidirectional Encoder Representations from Transformers (BERT) model. ### Claim 11 A computer system comprising: one or more computer processors; one or more computer readable storage media; and program instructions stored on the computer readable storage media for execution by at least one of the one or more processors, the program instructions comprising: program instructions to receive a question regarding a non-relational data table; program instructions to generate a schema based on a physical layout of the non-relational data table, wherein the schema includes one or more predicates that indicate the physical layout of the non-relational data table; program instructions to extract at least one feature of the question using a natural language processing (NLP) model; program instructions to extract at least one similar feature of the non-relational data table to the extracted at least one feature of the question; program instructions to determine at least one relevant cell in the non-relational data table based on the at least one feature of the question and the at least one similar feature of the non-relational data table; and program instructions to provide an answer to the question, wherein the answer is based on the at least one relevant cell. ### Claim 12 (depends on 11) The computer system of claim 11, the one or more computer readable storage media further storing: program instructions to identify a similar non-relational data table to the non-relational data table; and program instructions to extract a feature from the similar non-relational data table based on the schema generated based on the non-relational data table. ### Claim 13 (depends on 11) The computer system of claim 11, wherein the physical layout of the non-relational data table includes position information of one or more selected from a group consisting of: row and column headers, row and column subheaders, and cells. ### Claim 14 (depends on 11) The computer system of claim 11, wherein the non-relational data table includes unstructured data. ### Claim 15 (depends on 1) The method of claim 1, wherein the one or more predicates comprise a main heading predicate and a sub heading predicate. ### Claim 16 (depends on 1) The method of claim 1, wherein the one or more predicates comprise a predicate representing cell hierarchy. ### Claim 17 (depends on 1) The method of claim 1, wherein the one or more predicates comprise a predicate representing a link to another cell. ### Claim 18 (depends on 1) The method of claim 1, wherein the one or more predicates comprise a predicate representing no header. ### Claim 19 (depends on 1) The method of claim 1, wherein the one or more predicates comprise a predicate representing a presence of a mathematical operator. ### Claim 20 (depends on 1) The method of claim 1, wherein the one or more predicates are determined based on machine vision applied to an image of the non-relational data table.