US 12,277,180 B2
Information processing apparatus and method
Kazuhiro Miyajima, Tokyo (JP)
Assigned to SK TSUSHIN CORPORATION
Filed by SK TSUSHIN CORPORATION, Tokyo (JP)
Filed on Dec. 29, 2022, as Appl. No. 18/147,992.
Application 18/147,992 is a continuation of application No. PCT/JP2022/012257, filed on Mar. 17, 2022.
Claims priority of application No. 2021-088053 (JP), filed on May 26, 2021.
Prior Publication US 2023/0153364 A1, May 18, 2023
Int. Cl. G06F 16/95 (2019.01); G06F 16/953 (2019.01); G06N 20/00 (2019.01)
CPC G06F 16/953 (2019.01) [G06N 20/00 (2019.01)] 11 Claims
OG exemplary drawing
 
1. An information processing apparatus comprising a processing circuitry, wherein
the information processing apparatus stores in a storage unit a trained model generated by training using as training data multiple types of words and a result of searching Internet with a search engine using the multiple types of words as a search query, the multiple types of words being selected from a first type of words indicating areas, a second type of words indicating industry types, a third type of words indicating products or services, a fourth type of words describing or modifying areas, a fifth type of words describing or modifying industry types, or a sixth type of words describing or modifying products or services;
the trained model receives, as an input, designated words designated by a user, and thereby outputs an article content, and
the information processing apparatus causes the processing circuitry to execute:
receiving an input of at least a first word and a second word different in type from the first word as first designated words, the first word being one of the first, second, third, fourth, fifth, or sixth types of words; and
outputting a first article content corresponding to the received first designated words, based on the received first word and second word and the trained model,
wherein the trained model is generated by training based on a result in a predetermined search ranking range among results of a search on the Internet based on the search query;
wherein, the trained model is generated by training using as training data the multiple types of words and generalized sentences included in an article content included in the results of searching Internet with a search engine using the multiple types of words as a search query; and
the processing circuitry outputs the first article content separated from the generalized sentences included in the training data held in the storage unit by a predetermined range with respect to a distance between sentences included or a similarity between words included, based on the received first word and second words and the trained model.