# US 11,308,274 B2 · Word grouping using a plurality of models > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,308,274 B2 - Document ID: US11308274B2 - Status: Granted - Kind: B2 - Application number: 16/415,576 - Filed: 2019-05-17 - Published: 2020-11-19 - Granted: 2022-04-19 - Earliest claimed: 2019-05-17 - Assignees: International Business Machines Corporation - Applicants: International Business Machines Corporation - Inventors: Ryosuke Kohita; Issei Yoshida; Tetsuya Nasukawa; Hiroshi Kanayama - Web page: https://patentgrep.com/patent/US11308274B2 ## CPC - G06F 40/205: Parsing - G06F 40/30: Semantic analysis - G06F 17/15: Correlation function computation {including computation of convolution operations (arithmetic circuits for sum of products per se, e.g. multiply-accumulators G06F7/5443; digital filters, e.g. FIR, IIR, adaptive filters H03H17/00)} - G06F 40/274: Converting codes to words; Guess-ahead of partial word inputs - G06F 40/284: Lexical analysis, e.g. tokenisation or collocates - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract A computer-implemented method is provided. The method includes acquiring a seed word; calculating a similarity score of each of a plurality of words relative to the seed word for each of a plurality of models to calculate a weighted sum of similarity scores for each of the plurality of words; outputting a plurality of candidate words among the plurality of words; acquiring annotations indicating at least one of preferred words and non-preferred words among the plurality of the candidate words; updating weights of the plurality of models in a manner to cause weighted sums of similarity scores for the preferred words to be relatively larger than the weighted sums of the similarity scores for the non-preferred words, based on the annotations; and grouping the plurality of candidate words output based on the weighted sum of similarity scores calculated with updated weights of the plurality of models. ## Claims ### Claim 1 A computer-implemented method comprising: acquiring a seed word; calculating a similarity score of each of a plurality of words relative to the seed word for each of a plurality of models to calculate a weighted sum of the similarity scores for each of the plurality of words; outputting a plurality of candidate words among the plurality of words; acquiring annotations indicating one or more preferred words among the plurality of candidate words; updating weights of the plurality of models in a manner to cause weighted sums of similarity scores for the preferred words to be relatively larger than weighted sums of similarity scores for non-annotated words, the non-annotated words being non-preferred words; and grouping the plurality of candidate words output based on the weighted sum of similarity scores calculated with updated weights of the plurality of models. ### Claim 2 (depends on 1) The computer-implemented method of claim 1, wherein the plurality of models extract different feature quantities. ### Claim 3 (depends on 2) The computer-implemented method of claim 2, wherein calculating the similarity scores includes using a weighted average of cosine similarities of the plurality of models to calculate, the weighted sum of the similarity scores. ### Claim 4 (depends on 3) The computer-implemented method of claim 3, wherein the weighted average of the cosine similarities is calculated using a function in which a total weight of the plurality of models is normalized to become a constant value. ### Claim 5 (depends on 1) The computer-Implemented method of claim 1, wherein updating weights of the plurality of models includes updating the weights of the plurality of models in a manner to increase a smallest weighted sum of the similarity scores, among the weighted sums of the similarity scores of the plurality of preferred words. ### Claim 6 (depends on 1) The computer-implemented method of claim 1, wherein updating weights of the plurality of models includes updating the weights of the plurality of models in a manner to decrease a randomly selected weighted sum of the similarity scores, among the weighted sums of the similarity scores of the plurality of non-preferred words. ### Claim 7 (depends on 1) The computer-implemented method of claim 1, wherein outputting a plurality of candidate words includes outputting the plurality of candidate words in order of magnitude of the weighted sums of the similarity scores. ### Claim 8 (depends on 1) The computer-implemented method of claim 1, wherein acquiring annotations includes treating candidate words that are not indicated as being the preferred words, among the plurality of candidate words, as the non-preferred words. ### Claim 9 An apparatus comprising: a processor or a programmable circuitry; and one or more computer readable mediums collectively including instructions that, in response to being executed by the processor or the programmable circuitry, cause the processor or the programmable circuitry to: acquire a seed word; calculate a similarity score of each of a plurality of words relative to the seed word for each of a plurality of models to calculate a weighted sum of the similarity scores for each of the plurality of words; output a plurality of candidate words among the plurality of words; acquire annotations indicating one or more preferred words among the plurality of candidate words; update weights of the plurality of models in a manner to cause weighted sums of similarity scores for the preferred words to be relatively larger than weighted sums of the similarity scores for non-annotated words, the non-annotated words being non-preferred words; and group the plurality of candidate words output based on the weighted sum of similarity scores calculated with updated weights of the plurality of models. ### Claim 10 (depends on 9) The apparatus of claim 9, wherein the plurality of models extract different feature quantities. ### Claim 11 (depends on 9) The apparatus of claim 9, wherein updating weights of the plurality of models includes updating the weights of the plurality of models in a manner to increase a smallest weighted sum of the similarity scores, among the weighted sums of the similarity scores of the plurality of preferred words. ### Claim 12 (depends on 9) The apparatus of claim 9, wherein updating weights of the plurality of models includes updating the weights of the plurality of models in a manner to decrease a randomly selected weighted sum of the similarity scores, among the weighted sums of the similarity scores of the plurality of non-preferred words. ### Claim 13 A computer program product including one or more computer readable storage mediums collectively storing program instructions that are executable by a processor or programmable circuitry to cause the processor or the programmable circuitry to perform operations comprising: acquiring a seed word; calculating a similarity core of each of a plurality of words relative to the seed word for each of a plurality of models to calculate a weighted sum of the similarity scores for each of the plurality of words; outputting a plurality of candidate words among the plurality of words; acquiring annotations indicating one or more preferred words among the plurality of candidate words; updating weights of the plurality of models in a manner to cause weighted sums of similarity scores for the preferred to be relatively larger than weighted sums of similarity scores for non-annotated words, the non-annotated words being non-preferred words; and grouping the plurality of candidate words output based on the weighted of similarity scores calculated with updated weights of the plurality of models. ### Claim 14 (depends on 13) The computer program product of claim 13, wherein the plurality of models extract different feature quantities. ### Claim 15 (depends on 13) The computer program product of claim 13, wherein updating weights of the plurality of models includes updating the weights of the plurality of models in a manner to increase a smallest weighted sum of the similarity scores, among the weighted sums of the similarity scores of the plurality of preferred words. ### Claim 16 (depends on 13) The computer program product of claim 13, wherein the updating weights of the plurality of models includes updating the weights of the plurality of models in a manner to decrease a randomly selected weighted sum of the similarity scores, among the weighted sums of the similarity scores of the plurality of non-preferred words.