# US 10,311,146 B2 · Machine translation method for performing translation between languages > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 10,311,146 B2 - Document ID: US10311146B2 - Status: Granted - Kind: B2 - Application number: 16/045,034 - Filed: 2018-07-25 - Published: 2018-11-15 - Granted: 2019-06-04 - Earliest claimed: 2015-05-25 - Assignees: Panasonic Intellectual Property Corporation of America - Applicants: Panasonic Intellectual Property Corporation of America - Inventors: Nanami Fujiwara; Masaki Yamauchi - Web page: https://patentgrep.com/patent/US10311146B2 ## CPC - G06F 40/51: Translation evaluation - G06F 40/289: Phrasal analysis, e.g. finite state techniques or chunking - G06N 3/02: Neural networks - G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] - G06N 3/0455: Auto-encoder networks; Encoder-decoder networks - G06N 3/09: Supervised learning - G06N 3/092: Reinforcement learning - G06N 20/00: Machine learning - G06N 3/044: Recurrent networks, e.g. Hopfield networks - G06N 3/045: Combinations of networks - G10L 15/26: Speech to text systems (G10L15/08 takes precedence) - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING ## Abstract A machine translation method is provided, in which a plurality of different forward-translated sentences is generated resulting from translation of a received translation-source sentence in a first language into a second language. A plurality of backward-translated sentences is generated resulting from backward translation of the plurality of different forward-translated sentences into the first language. As translation-source sentence, voice information or text information is received. The voice information is received via a microphone that receives a voice input performed by a user. The text information is received via a text input device that receives a text input performed by a user. An output form of the forward-translated sentence corresponding to the selected backward-translated sentence is changed in accordance with which of the voice information and the text information is received as the translation-source sentence. ## Claims ### Claim 1 A machine translation method comprising: receiving a translation-source sentence in a first language; generating a plurality of different forward-translated sentences resulting from translation of the received translation-source sentence into a second language; generating a plurality of backward-translated sentences resulting from backward translation of the plurality of different forward-translated sentences into the first language; and outputting, upon receiving an operation for selecting one of the plurality of backward-translated sentences during output of the plurality of backward-translated sentences on an information output device, a forward-translated sentence corresponding to the selected backward-translated sentence to the information output device, wherein the receiving includes receiving, as the translation-source sentence, voice information via a microphone that receives a voice input performed by a user, or receiving, as the translation-source sentence, text information via a text input device that receives a text input performed by a user, wherein the method further comprising: changing an output form of the forward-translated sentence corresponding to the selected backward-translated sentence in accordance with which of the voice information and the text information is received as the translation-source sentence. ### Claim 2 (depends on 1) The machine translation method according to claim 1, wherein the information output device has a speaker and a display, and wherein the outputting includes: outputting, when the translation-source sentence is received in a form of the voice information, the forward-translated sentence corresponding to the selected backward-translated sentence via the speaker, and outputting, when the translation-source sentence is received in a form of the text information, the forward-translated sentence corresponding to the selected backward-translated sentence via the display. ### Claim 3 (depends on 1) The machine translation method according to claim 1, wherein when the text information is received as the translation-source sentence via the text input device, the plurality of different forward-translated sentences resulting from translation of the translation-source sentence into the second language, are generated based on the text information. ### Claim 4 (depends on 1) The machine translation method according to claim 1, further comprising: performing, when the voice information is received as the translation-source sentence via the microphone, voice recognition processing on the received voice information to generate text information indicating the translation-source sentence, wherein the plurality of different forward-translated sentences resulting from translation of the translation-source sentence into the second language, are generated based on the text information obtained by the voice recognition processing. ### Claim 5 A machine translation method comprising: receiving a translation-source sentence in a first language; generating a plurality of different forward-translated sentences resulting from translation of the received translation-source sentence into a second language; generating a plurality of backward-translated sentences resulting from backward translation of the plurality of different forward-translated sentences into the first language; and outputting, upon receiving an operation for selecting one of the plurality of backward-translated sentences during output of the plurality of backward-translated sentences on an information output device, a forward-translated sentence corresponding to the selected backward-translated sentence to the information output device, wherein the information output device has a display, wherein the method further comprising: displaying the plurality of backward-translated sentences in a first area on the display; and displaying the translation-source sentence in a second area on the display, the second area being different from the first area. ### Claim 6 (depends on 5) The machine translation method according to claim 5, further comprising: displaying, in a third area on the display, the forward-translated sentence corresponding to the selected backward-translated sentence. ### Claim 7 (depends on 6) The machine translation method according to claim 6, further comprising: displaying the forward-translated sentence corresponding to the selected backward-translated sentence in a direction different from a direction of the plurality of backward-translated sentences displayed in the first area. ### Claim 8 A machine translation method comprising: receiving a translation-source sentence in a first language; generating a plurality of different forward-translated sentences resulting from translation of the received translation-source sentence into a second language; generating a plurality of backward-translated sentences resulting from backward translation of the plurality of different forward-translated sentences into the first language; and outputting, upon receiving an operation for selecting one of the plurality of backward-translated sentences during output of the plurality of backward-translated sentences on an information output device, a forward-translated sentence corresponding to the selected backward-translated sentence to the information output device, wherein the receiving includes receiving, as the translation-source sentence, voice information via a microphone that receives a voice input performed by a user, or receiving, as the translation-source sentence, text information via a text input device that receives a text input performed by a user, wherein the method further comprising: generating a forward-translated sentence group that is a collection of forward-translated sentences resulting from translation of the received translation-source sentence into the second language, the forward-translated sentence group including the plurality of different forward-translated sentences; determining a subject or a predicate in each forward-translated sentence included in the forward-translated sentence group; and determining the plurality of different forward-translated sentences in the forward-translated sentence group, based on the determined subject or predicate. ### Claim 9 A machine translation method comprising: receiving a translation-source sentence in a first language; generating a plurality of different forward-translated sentences resulting from translation of the received translation-source sentence into a second language; generating a plurality of backward-translated sentences resulting from backward translation of the plurality of different forward-translated sentences into the first language; and outputting, upon receiving an operation for selecting one of the plurality of backward-translated sentences during output of the plurality of backward-translated sentences on an information output device, a forward-translated sentence corresponding to the selected backward-translated sentence to the information output device, wherein the receiving includes receiving, as the translation-source sentence, voice information via a microphone that receives a voice input performed by a user, or receiving, as the translation-source sentence, text information via a text input device that receives a text input performed by a user, the method further comprising: generating a backward-translated sentence group that is a collection of at least one backward-translated sentence generated for each of the plurality of different forward-translated sentences, the backward-translated sentence group including the plurality of backward-translated sentences; calculating evaluation values obtained by evaluating similarities between the translation-source sentence and the at least one backward-translated sentence included in the backward-translated sentence group; and selecting the plurality of backward-translated sentences from the backward-translated sentence group, based on the evaluation values. ### Claim 10 A machine translation method comprising: receiving a translation-source sentence in a first language; generating a plurality of different forward-translated sentences resulting from translation of the received translation-source sentence into a second language; generating a plurality of backward-translated sentences resulting from backward translation of the plurality of different forward-translated sentences into the first language; and outputting, upon receiving an operation for selecting one of the plurality of backward-translated sentences during output of the plurality of backward-translated sentences on an information output device, a forward-translated sentence corresponding to the selected backward-translated sentence to the information output device, wherein the receiving includes receiving, as the translation-source sentence, voice information via a microphone that receives a voice input performed by a user, or receiving, as the translation-source sentence, text information via a text input device that receives a text input performed by a user, wherein the method further comprising: generating a backward-translated sentence group that is a collection of at least one backward-translated sentence generated for each of the plurality of different forward-translated sentences, the backward-translated sentence group including the plurality of backward-translated sentences; determining a subject or a predicate of each of the at least one backward-translated sentence included in the backward-translated sentence group; and selecting the plurality of backward-translated sentences from the backward-translated sentence group, based on the determined subject or predicate.