# US 12,182,497 B2 · Utterance sentence extension apparatus, utterance sentence generator apparatus, utterance sentence extension method, and program > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,182,497 B2 - Document ID: US12182497B2 - Status: Granted - Kind: B2 - Application number: 17/603,418 - Filed: 2020-04-10 - Published: 2022-08-18 - Granted: 2024-12-31 - Earliest claimed: 2019-04-16 - Assignees: Nippon Telegraph and Telephone Corporation - Applicants: Nippon Telegraph and Telephone Corporation - Inventors: Ko Mitsuda; Ryuichiro Higashinaka; Junji Tomita - Web page: https://patentgrep.com/patent/US12182497B2 ## CPC - G06F 40/166: Editing, e.g. inserting or deleting - G06F 40/242: Dictionaries - G06F 40/253: Grammatical analysis; Style critique - G06F 40/268: Morphological analysis - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract An expanded utterance that is used to output a more appropriate output utterance for an utterance can be generated. An utterance sentence expansion device includes an expansion unit that inserts, for an utterance that is an utterance to be expanded that includes a noun and is morphologically analyzed in advance, by using information of an expansion dictionary, which includes higher-level categories of the noun, one or more higher-level categories of the expansion dictionary corresponding to the noun included in the utterance into a position before the noun of the utterance to generate an expanded utterance. ## Claims ### Claim 1 An utterance sentence expansion device comprising an expander configured to insert, for an utterance that is an utterance to be expanded that includes a noun and an unknown input and are morphologically analyzed in advance, by using information of an expansion dictionary, the expansion dictionary including a plurality of higher-level categories of the noun, one or more higher-level categories of the plurality of higher-level categories of the expansion dictionary corresponding to the noun included in the utterance into a position before a first noun of the utterance to generate an expanded utterance, wherein the first noun is automatically indicated by a space and a symbol. ### Claim 2 (depends on 1) The utterance sentence expansion device according to claim 1, wherein in a case of inserting the plurality of higher-level categories of the expansion dictionary corresponding to the noun included in the utterance, the expander inserts, into a position before the noun of the utterance, the plurality of higher-level categories in such an order that, among the plurality of higher-level categories, a category in a higher-level hierarchy is closer to a beginning of a sentence to generate the expanded utterance. ### Claim 3 (depends on 2) The utterance sentence expansion device according to claim 2, further comprising: providing an output utterance by using an utterance sentence as an input and using the expanded utterance generated from the utterance sentence as an input to an utterance generation model that is trained in advance, wherein the utterance generation model is trained in advance by the expander by using an input utterance sentence included in training data as an input and the expanded utterance generated from the input utterance sentence as an input utterance sentence of the training data. ### Claim 4 (depends on 1) The utterance sentence expansion device according to claim 1, further comprising: providing an output utterance by using an utterance sentence as an input and using the expanded utterance generated from the utterance sentence as an input to an utterance generation model that is trained in advance, wherein the utterance generation model is trained in advance by the expander by using an input utterance sentence included in training data as an input and the expanded utterance generated from the input utterance sentence as an input utterance sentence of the training data. ### Claim 5 An utterance sentence expansion method, comprising inserting, for an utterance that is an utterance to be expanded that includes a noun and an unknown input and are morphologically analyzed in advance, by an expander using information of an expansion dictionary, the expansion dictionary including a plurality of higher-level categories of the noun, one or more higher-level categories of the plurality of higher-level categories of the expansion dictionary corresponding to the noun included in the utterance into a position before a first noun of the utterance to generate an expanded utterance, wherein the first noun is automatically indicated by a space and a symbol. ### Claim 6 (depends on 5) The utterance sentence expansion method according to claim 5, wherein in a case of inserting the plurality of higher-level categories of the expansion dictionary corresponding to the noun included in the utterance, the expander inserts, into a position before the noun of the utterance, the plurality of higher-level categories in such an order that, among the plurality of higher-level categories, a category in a higher-level hierarchy is closer to a beginning of a sentence to generate the expanded utterance. ### Claim 7 (depends on 6) The utterance sentence expansion method according to claim 6, further comprising: providing an output utterance by using an utterance sentence as an input and using the expanded utterance generated from the utterance sentence as an input to an utterance generation model that is trained in advance, wherein the utterance generation model is trained in advance by the expander by using an input utterance sentence included in training data as an input and the expanded utterance generated from the input utterance sentence as an input utterance sentence of the training data. ### Claim 8 (depends on 5) The utterance sentence expansion method according to claim 5, further comprising: providing an output utterance by using an utterance sentence as an input and using the expanded utterance generated from the utterance sentence as an input to an utterance generation model that is trained in advance, wherein the utterance generation model is trained in advance by the expander by using an input utterance sentence included in training data as an input and the expanded utterance generated from the input utterance sentence as an input utterance sentence of the training data. ### Claim 9 A computer-readable non-transitory recording medium storing computer-executable program instruction that when executed by a processor cause a computer system to: insert, for an utterance that is an utterance to be expanded that includes a noun and an unknown input and are morphologically analyzed in advance, by an expander using information of an expansion dictionary, the expansion dictionary including a plurality of higher-level categories of the noun, one or more higher-level categories of the plurality of higher-level categories of the expansion dictionary corresponding to the noun included in the utterance into a position before the noun of the utterance to generate an expanded utterance, wherein the first noun is automatically indicated by a space and a symbol. ### Claim 10 (depends on 9) The computer-readable non-transitory recording medium according to claim 9, wherein in a case of inserting the plurality of higher-level categories of the expansion dictionary corresponding to the noun included in the utterance, the expander inserts, into a position before the noun of the utterance, the plurality of higher-level categories in such an order that, among the plurality of higher-level categories, a category in a higher-level hierarchy is closer to a beginning of a sentence to generate the expanded utterance. ### Claim 11 (depends on 10) The computer-readable non-transitory recording medium according to claim 10, further comprising: providing an output utterance by using an utterance sentence as an input and using the expanded utterance generated from the utterance sentence as an input to an utterance generation model that is trained in advance, wherein the utterance generation model is trained in advance by the expander by using an input utterance sentence included in training data as an input and the expanded utterance generated from the input utterance sentence as an input utterance sentence of the training data. ### Claim 12 (depends on 9) The computer-readable non-transitory recording medium according to claim 9, further comprising: providing an output utterance by using an utterance sentence as an input and using the expanded utterance generated from the utterance sentence as an input to an utterance generation model that is trained in advance, wherein the utterance generation model is trained in advance by the expander by using an input utterance sentence included in training data as an input and the expanded utterance generated from the input utterance sentence as an input utterance sentence of the training data.