# US 2017/0286403 A1 · System, method, and recording medium for natural language learning > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2017/0286403 A1 - Document ID: US20170286403A1 - Status: Published application - Kind: A1 - Application number: 15/087,050 - Filed: 2016-03-31 - Published: 2017-10-05 - Earliest claimed: 2016-03-31 - Assignees: International Business Machines Corporation - Applicants: International Business Machines Corporation - Inventors: Octavian Popescu; Vadim Sheinin - Web page: https://patentgrep.com/patent/US20170286403A1 ## CPC - G06F 40/211: Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars - G06F 40/284: Lexical analysis, e.g. tokenisation or collocates - G06F 40/30: Semantic analysis - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract A natural language learning method, system, and non-transitory computer readable medium include analyzing a corpus of sentences stored in a database to identify an internal structure of words in the corpus of sentences, creating a plurality of new words that are a combination of the internal structure of a word of the words in the corpus of sentences and the word, clustering the plurality of new words created by the creating that match into a plurality of cluster groups, filtering the plurality of cluster groups to create a partial set of each of the plurality of cluster groups, and performing word embedding processing on the partial set of each of the plurality of cluster groups to obtain vectors for new words. ## Claims ### Claim 1 A natural language learning system comprising: an analyzing circuit configured to analyze a corpus of sentences stored in a database to identify an internal structure of words in the corpus of sentences; a new word creating circuit configured to create a plurality of new words that comprise a combination of the internal structure of a word of the words in the corpus of sentences and the word; a clustering circuit configured to cluster the plurality of new words created by the new word creating circuit that match into a plurality of cluster groups; a filtering circuit configured to filter the plurality of cluster groups to create a partial set of each of the plurality of cluster groups; and a word embedding processing circuit configured to obtain vectors for new words by performing word embedding processing on the partial set of each of the plurality of cluster groups. ### Claim 2 (depends on 1) The system of claim 1, wherein the internal structure of the words comprises a syntactical feature of the words of the corpus of sentences. ### Claim 3 (depends on 1) The system of claim 1, wherein the new word includes a key cite to a sentence of the corpus of sentences in which the word is originally included. ### Claim 4 (depends on 2) The system of claim 2, wherein the clustering circuit clusters the plurality of new words based on the syntactical feature matching. ### Claim 5 (depends on 1) The system of claim 1, wherein the plurality of new words are created by combining lemmas and syntactic functions. ### Claim 6 (depends on 1) The system of claim 1, wherein the vectors obtained by the word embedding processing circuit encode syntactic structure and concepts of the plurality of new words. ### Claim 7 (depends on 1) The system of claim 1, wherein the word embedding processing comprises a Recurrent Neural Network (RNN) processing. ### Claim 8 A natural language learning method, comprising: analyzing a corpus of sentences stored in a database to identify an internal structure of words in the corpus of sentences; creating a plurality of new words that comprise a combination of the internal structure of a word of the words in the corpus of sentences and the word; clustering the plurality of new words created by the creating that match into a plurality of cluster groups; filtering the plurality of cluster groups to create a partial set of each of the plurality of cluster groups; and performing word embedding processing on the partial set of each of the plurality of cluster groups to obtain vectors for new words. ### Claim 9 (depends on 8) The method of claim 8, wherein the internal structure of the words comprises a syntactical feature of the words of the corpus of sentences. ### Claim 10 (depends on 8) The method of claim 8, wherein the new word includes a key cite to a sentence of the corpus of sentences in which the word is originally included. ### Claim 11 (depends on 9) The method of claim 9, wherein the clustering clusters the plurality of new words based on the syntactical feature matching. ### Claim 12 (depends on 8) The method of claim 8, wherein the creating creates the plurality of new words by combining lemmas and syntactic functions. ### Claim 13 (depends on 8) The method of claim 8, wherein the vectors obtained by the performing encode syntactic structure and concepts of the plurality of new words. ### Claim 14 (depends on 8) The method of claim 8, wherein the word embedding processing comprises a Recurrent Neural Network (RNN) processing. ### Claim 15 A non-transitory computer-readable recording medium recording a natural language learning program, the program causing a computer to perform: analyzing a corpus of sentences stored in a database to identify an internal structure of words in the corpus of sentences; creating a plurality of new words that comprise a combination of the internal structure of a word of the words in the corpus of sentences and the word; clustering the plurality of new words created by the creating that match into a plurality of cluster groups; filtering the plurality of cluster groups to create a partial set of each of the plurality of cluster groups; and performing word embedding processing on the partial set of each of the plurality of cluster groups to obtain vectors for new words. ### Claim 16 (depends on 15) The non-transitory computer-readable recording medium of claim 15, wherein the internal structure of the words comprises a syntactical feature of the words of the corpus of sentences. ### Claim 17 (depends on 15) The non-transitory computer-readable recording medium of claim 15, wherein the new word includes a key cite to a sentence of the corpus of sentences in which the word is originally included. ### Claim 18 (depends on 16) The non-transitory computer-readable recording medium of claim 16, wherein the clustering clusters the plurality of new words based on the syntactical feature matching ### Claim 19 (depends on 15) The non-transitory computer-readable recording medium of claim 15, wherein the creating creates the plurality of new words by combining lemmas and syntactic functions. ### Claim 20 (depends on 15) The non-transitory computer-readable recording medium of claim 15, wherein the vectors obtained by the performing encode syntactic structure and concepts of the plurality of new words.