# US 11,443,199 B2 · Program storage medium, apparatus and method provided with ruleset-selectable inference engine > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,443,199 B2 - Document ID: US11443199B2 - Status: Granted - Kind: B2 - Application number: 16/464,752 - Filed: 2017-12-06 - Published: 2019-12-19 - Granted: 2022-09-13 - Earliest claimed: 2016-12-07 - Assignees: Takeoka Lab Corporation - Applicants: Takeoka Lab Corporation - Inventors: Shozo Takeoka - Web page: https://patentgrep.com/patent/US11443199B2 ## CPC - G06F 16/00: Information retrieval; Database structures therefor; File system structures therefor - G06N 5/027: {Frames} - G06N 5/042: {Backward inferencing} - G06N 20/10: using kernel methods, e.g. support vector machines [SVM] - G06Q 10/107: Computer-aided management of electronic mailing [e-mailing] - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR ## Abstract An apparatus, a program-stored storage medium and a method with an inference engine can execute inference using a minimum ruleset in various applications. The apparatus includes: a machine learning engine being a classifying-type engine configured to include adapted-to-category learning models each generated by using each adapted-to-category set of teacher data, the adapted-to-category set being obtained by classifying teacher data for each category, and to use the learning models to output a category data corresponding to the inputted object data; a ruleset selector configured to select, from rulesets each prepared for each category and stored in a rule base, a ruleset corresponding to the category data outputted from the machine learning engine; and a rule engine configured to execute inference to the inputted object data by using the ruleset selected by the ruleset selector, and to output the inference result. ## Claims ### Claim 1 A non-transitory computer-readable storage medium on which is stored a program to be executed by a computer mounted on an inference apparatus outputting a result of inferring inputted object data, the program causing the computer to: function as a machine learning engine that is a classifying-type engine configured to include adapted-to-category learning models each generated by using each adapted-to-category set of teacher data, each of the adapted-to-category set being obtained by classifying teacher data for each category, the computer, functioning as the machine learning engine, receives the inputted object data and uses the learning models to output a category data corresponding to the inputted object data; select, from a plurality of rulesets each prepared for each category and stored in a rule database, a ruleset corresponding to the category outputted from the machine learning engine corresponding to the inputted object data; and function as a rule engine by executing inference to the inputted object data by using the selected ruleset as a ruleset of the category outputted from the machine learning engine that corresponds to the inputted object data, and outputting the inference result to execute processing based on the inference result. ### Claim 2 (depends on 1) The non-transitory computer-readable storage medium as claimed in claim 1, wherein: the machine learning engine includes a plurality of support vector machines each of which corresponds to each category, the computer is configured to, as the support vector machine in a learning phase, input pairs of a value showing whether to belong to a target category or not and teacher data, and generate a learning model, and the computer is configured to, as the support vector machine in an operating phase, input object data, and output a value showing whether the object data belongs to the target category or not by using the learning model. ### Claim 3 (depends on 1) The non-transitory computer-readable storage medium as claimed in claim 1, wherein: the object data and the teacher data are text data, the program further causes the computer to output a set of morphemes obtained from individual text data of the object data and the teacher data, the set of morphemes obtained from the object data is inputted in both of the machine learning engine and the rule engine, and the set of morphemes obtained from the teacher data is inputted in the machine learning engine. ### Claim 4 (depends on 3) The non-transitory computer-readable storage medium as claimed in claim 3, wherein the program further causes the computer to: receive, functioning as a middle arguments generator, the set of morphemes obtained from the object data, and output, to the rule engine, a plurality of middle arguments as the object data, the plurality of middle arguments corresponding to each of different purposes, wherein the computer, as the rule engine, executes backward chaining of the middle arguments as the object data by using the selected ruleset, and outputs the inference result. ### Claim 5 (depends on 4) The non-transitory computer-readable storage medium as claimed in claim 4, wherein the computer, as the middle arguments generator, is described in LISP which is an expression-oriented and procedural functional-programming language. ### Claim 6 (depends on 3) The non-transitory computer-readable storage medium as claimed in claim 3, wherein: the object data and the teacher data are email text data, and the category is data showing whether the text data is automatically-distributed email data or not, whether the text data is business email data or not, or whether the text data is private email data or not. ### Claim 7 (depends on 1) The non-transitory computer-readable storage medium as claimed in claim 1, wherein the computer, as the rule engine, is described in Prolog which is an expression-oriented and nonprocedural logical-programming language. ### Claim 8 (depends on 1) The non-transitory computer-readable storage medium as claimed in claim 1, wherein: the object data and the teacher data are multimedia data, the program further causes the computer to output a set of text elements given to each multimedia data of the object data and the teacher data, the set of text elements given to the object data is inputted in both of the machine learning engine and the rule engine, and the set of text elements given to the teacher data is inputted in the machine learning engine. ### Claim 9 (depends on 1) The non-transitory computer-readable storage medium as claimed in claim 1, wherein the object data and the teacher data are sensor type data and measurement values, a set of sensor type data and measurement values of the object data is inputted in both of the machine learning engine and the rule engine, and a set of sensor type data and measurement values of the teacher data is inputted in the machine learning engine. ### Claim 10 An inference apparatus outputting a result of inferring inputted object data, the inference apparatus comprising: a processor programmed to: function as a machine learning engine that is a classifying-type engine configured to include adapted-to-category learning models each generated by using each adapted-to-category set of teacher data, the adapted-to-category set being obtained by classifying teacher data for each category, the processor, functioning as the machine learning engine, receives the inputted object data and uses use the learning models to output a category corresponding to the inputted object data; select, from a plurality of rulesets each prepared for each category and stored in a rule database, a ruleset corresponding to the category data outputted from the machine learning engine corresponding to the inputted object data; and function as a rule engine by executing inference to the inputted object data by using the selected ruleset as a ruleset of the category outputted from the machine learning engine that corresponds to the inputted object data, and outputting the inference result to execute processing based on the inference result. ### Claim 11 An inference method executed by a processor of an inference apparatus outputting a result of inferring inputted object data, the processor of the inference apparatus functioning as a machine learning engine that includes adapted-to-category learning models each generated by using each adapted-to-category set of teacher data, the adapted-to-category set being obtained by classifying teacher data for each category, and the inference apparatus including a memory that stores a rule database that stores, in advance, a plurality of rulesets each prepared for each category, the inference method comprising: a first step, executed by the processor as the machine learning engine, of receiving the inputted object data and using the learning models to output a category data corresponding to the inputted object data; a second step, by the processor, of selecting, from the rulesets stored in the rule database in the memory, a ruleset corresponding to the category outputted from the machine learning engine corresponding to the inputted object data; and a third step, by the processor, of executing inference to the inputted object data by using the ruleset selected in the second step as a ruleset of the category outputted from the machine learning engine that corresponds to the inputted object data, and then outputting the inference result to execute processing based on the inference result.