# US 11,443,118 B2 · Word embedding method and apparatus, and word search method > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,443,118 B2 - Document ID: US11443118B2 - Status: Granted - Kind: B2 - Application number: 16/822,436 - Filed: 2020-03-18 - Published: 2021-04-15 - Granted: 2022-09-13 - Earliest claimed: 2019-10-14 - Assignees: Samsung Electronics Co., Ltd. - Applicants: Samsung Electronics Co., Ltd. - Inventors: Sanghyun Yoo; Young-Seok Kim; Junhwi Choi - Web page: https://patentgrep.com/patent/US11443118B2 ## CPC - G06F 40/30: Semantic analysis - G16C 20/70: Machine learning, data mining or chemometrics - G06F 40/274: Converting codes to words; Guess-ahead of partial word inputs - G06F 40/295: Named entity recognition - G06N 3/0442: characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] - G06N 3/045: Combinations of networks - G06N 3/0464: Convolutional networks [CNN, ConvNet] - G06N 3/09: Supervised learning - G06N 20/00: Machine learning - G16C 60/00: Computational materials science, i.e. ICT specially adapted for investigating the physical or chemical properties of materials or phenomena associated with their design, synthesis, processing, characterisation or utilisation - G06N 3/044: Recurrent networks, e.g. Hopfield networks - G16C 20/80: Data visualisation - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G16C: COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract A word embedding method and apparatus and a word search method are provided, wherein the word embedding method includes training a word embedding model based on characteristic information of a chemical substance, and acquiring an embedding vector of a word representing the chemical substance from the word embedding model, wherein the word embedding model is configured to predict a context word of an input word. ## Claims ### Claim 1 A word embedding method comprising: training a word embedding model based on characteristic information of a chemical substance to output a context word of a word representing the chemical substance, in response to an input of composition information of the chemical substance; and acquiring an embedding vector of the word representing the chemical substance from the word embedding model, wherein the word representing the chemical substance is split into letters or elements and the letters or the elements are sequentially input to the word embedding model. ### Claim 2 (depends on 1) The word embedding method of claim 1, wherein the training of the word embedding model comprises training the word embedding model based on any one or any combination of structure information, composition information, and physical property information of the chemical substance. ### Claim 3 (depends on 1) The word embedding method of claim 1, wherein the training of the word embedding model further comprises training the word embedding model to output a context word of the word representing the chemical substance, in response to an input of structure information of the chemical substance to the word embedding model. ### Claim 4 (depends on 3) The word embedding method of claim 3, wherein the structure information of the chemical substance is determined based on a format of one of a fingerprint, a simplified molecular-input line-entry system (SMILES), a graph, or an image. ### Claim 5 (depends on 1) The word embedding method of claim 1, wherein the composition information of the chemical substance is acquired from the word representing the chemical substance. ### Claim 6 (depends on 1) The word embedding method of claim 1, wherein the training of the word embedding model further comprises training the word embedding model to output physical property information of the chemical substance from the word embedding model. ### Claim 7 (depends on 6) The word embedding method of claim 6, wherein the physical property information comprises information about any one or any combination of a mass, a volume, a color, a melting point, and a boiling point of the chemical substance. ### Claim 8 (depends on 1) The word embedding method of claim 1, further comprising: inputting the embedding vector to a portion corresponding to the word representing the chemical substance in a word embedding matrix corresponding to the word embedding model. ### Claim 9 (depends on 1) The word embedding method of claim 1, further comprising: determining whether a word with an embedding vector to be generated represents a chemical material. ### Claim 10 (depends on 1) A non-transitory computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform the method of claim 1. ### Claim 11 A word search method comprising: receiving characteristic information of a chemical substance or a word representing the chemical substance; and outputting a word representing a substance having a characteristic similar to the chemical substance based on a word embedding matrix, wherein the word embedding matrix is acquired from a word embedding model trained based on characteristic information of a plurality of chemical substances to output a context word of a word representing the chemical substance, in response to an input of composition information of the chemical substance, and wherein the word representing the chemical substance is split into letters or elements and the letters or the elements are sequentially input to the word embedding model is configured to predict a context word of an input word. ### Claim 12 (depends on 11) The word search method of claim 11, wherein the characteristic information of the chemical substance comprises any one or any combination of structure information, composition information, and physical property information of the chemical substance. ### Claim 13 The word embedding apparatus comprising: a processor configured to: train a word embedding model to output physical property information of a chemical substance based on characteristic information of the chemical substance; and acquire an embedding vector of a word representing the chemical substance from the word embedding model, and the word embedding model is configured to predict a context word of an input word, wherein the physical property information comprises information about any one or any combination of a mass, a volume, a color, a melting point, and a boiling point of the chemical substance. ### Claim 14 (depends on 13) The word embedding apparatus of claim 13, wherein the processor is further configured to train the word embedding model based on any one or any combination of structure information, composition information, and the physical property information of the chemical substance. ### Claim 15 (depends on 13) The word embedding apparatus of claim 13, wherein the processor is further configured to train the word embedding model to output a context word of the word representing the chemical substance, in response to an input of structure information of the chemical substance to the word embedding model. ### Claim 16 (depends on 13) The word embedding apparatus of claim 13, wherein the processor is further configured to train the word embedding model to output a context word of the word representing the chemical substance, in response to an input of composition information of the chemical substance to the word embedding model. ### Claim 17 (depends on 13) The word embedding apparatus of claim 13, wherein the processor is further configured to input the embedding vector to a portion corresponding to the word representing the chemical substance in a word embedding matrix corresponding to the word embedding model.