# US 2021/0240942 A1 · Electronic device for obtaining sentence corresponding to context information and operating method thereof > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2021/0240942 A1 - Document ID: US20210240942A1 - Status: Published application - Kind: A1 - Application number: 17/077,874 - Filed: 2020-10-22 - Published: 2021-08-05 - Earliest claimed: 2020-01-30 - Assignees: Samsung Electronics Co., Ltd.; New York University - Applicants: Samsung Electronics Co., Ltd.; New York University - Inventors: Yoonjung CHOI; Jaedeok KIM; Ilia KULIKOV; Sean WELLECK; Yuanzhe PANG; Kyunghyun CHO - Web page: https://patentgrep.com/patent/US20210240942A1 ## CPC - G06F 40/44: Statistical methods, e.g. probability models - G06F 40/284: Lexical analysis, e.g. tokenisation or collocates - G06F 40/56: Natural language generation - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract A method of obtaining, by an electronic device, a sentence corresponding to context information, including obtaining first output information including at least one word output by decoding the context information based on at least one data; based on detecting that a first token is not included in the first output information, determining whether a number of words included in the first output information is greater than or equal to a reference value; based on a result of the determining, replacing the at least one data with other data; and obtaining the sentence corresponding to the context information based on at least one output information obtained by decoding the context information based on the other data. ## Claims ### Claim 1 A method of obtaining, by an electronic device, a sentence corresponding to context information, the method comprising: obtaining first output information including at least one word output by decoding the context information based on at least one data; based on detecting that a first token is not included in the first output information, determining whether a number of words included in the first output information is greater than or equal to a reference value; based on a result of the determining, replacing the at least one data with other data; and obtaining the sentence corresponding to the context information based on at least one output information obtained by decoding the context information based on the other data. ### Claim 2 (depends on 1) The method of claim 1, wherein the at least one data includes at least one of a language model or a decoding algorithm. ### Claim 3 (depends on 1) The method of claim 1, wherein the replacing the at least one data with the other data increases a probability that output information including the first token is obtained. ### Claim 4 (depends on 1) The method of claim 1, wherein the at least one data includes a first language model having a first probability value for the first token, wherein the other data includes a second language model having a second probability value for the first token, and wherein the probability value is different from the second probability value. ### Claim 5 (depends on 1) The method of claim 1, wherein the at least one data includes a first decoding algorithm, and wherein the other data includes a second decoding algorithm having the first token added as a candidate word. ### Claim 6 (depends on 1) The method of claim 1, wherein the first token represents an end of the sentence, and wherein based on output information including the first token being obtained, the electronic device obtains the sentence corresponding to the context information. ### Claim 7 (depends on 1) The method of claim 1, wherein based on a number of words included in second output information being determined to be greater than or equal to a first reference value, first-first data of the at least one data is replaced with second-first data, and wherein based on a number of words included in third output information being determined to be greater than or equal to a second reference value that is greater than the first reference value, first-second data of the at least one data is replaced with second-second data. ### Claim 8 An electronic device for obtaining a sentence corresponding to context information, the electronic device comprising: a memory configured to store at least one data used for decoding the context information; at least one processor configured to: obtain first output information including at least one word output by decoding the context information based on the at least one data, based on detecting that a first token is not included in the first output information, determine whether a number of words included in the first output information is greater than or equal to a reference value, based on a result of the determining, replace the at least one data with other data and obtain the sentence corresponding to the context information based on at least one output information obtained by decoding the context information based on the other data; and an output interface configured to output information corresponding to he obtained sentence. ### Claim 9 (depends on 8) The electronic device of claim 8, wherein the at least one data includes at least one of a language model or a decoding algorithm. ### Claim 10 (depends on 8) The electronic device of claim 8, wherein the replacing the at least one data with the other data increases a probability that output information including the first token is obtained. ### Claim 11 (depends on 8) The electronic device of claim 8, wherein the at least one data includes a first language model having a first probability value for the first token, wherein the other data includes a second language model having a second probability value for the first token, and wherein the first probability value is different from the second probability value. ### Claim 12 (depends on 8) The electronic device of claim 8, wherein the at least one data includes a first decoding algorithm, and wherein the other data includes a second decoding algorithm having the first token added as a candidate word. ### Claim 13 (depends on 8) The electronic device of claim 8, wherein the first token represents an end of the sentence, and wherein based on output information including the first token being obtained, the at least one processor is further configured to obtain the sentence corresponding to the context information. ### Claim 14 (depends on 8) The electronic device of claim 8, wherein based on a number of words included in second output information being determined to be greater than or equal to a first reference value, first-first data of the at least one data is replaced with second-first data, and wherein based on a number of words included in third output information being determined to be greater than or equal to a second reference value that is greater than the first reference value, first-second data of the at least one data is replaced with second-second data. ### Claim 15 A non-transitory computer-readable medium configured to store instructions which, when executed by at least one processor of an electronic device for obtaining a sentence corresponding to context information, cause the at least one processor to: obtain first output information including at least one word output by decoding the context information based on at least one data; based on detecting that a first token is not included in the first output information, determine whether a number of words included in the first output information is greater than or equal to a reference value; based on a result of the determining, replace the at least one data with other data; and obtain the sentence corresponding to the context information based on at least one output information obtained by decoding the context information based on the other data. ### Claim 16 (depends on 15) The non-transitory computer-readable medium of claim 15, wherein the at least one data includes at least one of a language model or a decoding algorithm. ### Claim 17 (depends on 15) The non-transitory computer-readable medium of claim 15, wherein the replacing the at least one data with the other data increases a probability that output information including the first token is obtained. ### Claim 18 (depends on 15) The non-transitory computer-readable medium of claim 15, wherein the at least one data includes a first language model having a first probability value for the first token, wherein the other data includes a second language model having a second probability value for the first token, and wherein the first probability value is different from the second probability value. ### Claim 19 (depends on 15) The non-transitory computer-readable medium of claim 15, wherein the at least one data includes a first decoding algorithm, and wherein the other data includes a second decoding algorithm having the first token added as a candidate word. ### Claim 20 (depends on 15) The non-transitory computer-readable medium of claim 15, wherein based on a number of words included in second output information being determined to be greater than or equal to a first reference value, first-first data of the at least one data is replaced with second-first data, and wherein based on a number of words included in third output information being determined to be greater than or equal to a second reference value that is greater than the first reference value, first-second data of the at least one data is replaced with second-second data.