US 11,934,783 B2
Systems and methods for enhanced review comprehension using domain-specific knowledgebases
Yoshihiko Suhara, Los Altos, CA (US); Behzad Golshan, Mountain View, CA (US); Yuliang Li, Cupertino, CA (US); Chen Chen, Sunnyvale, CA (US); Xiaolan Wang, Mountain View, CA (US); Jinfeng Li, Mountain View, CA (US); Wang-Chiew Tan, San Jose, CA (US); çagatay Demiralp, Mountain View, CA (US); and Aaron Traylor, Providence, RI (US)
Assigned to RECRUIT CO., LTD., Toyko (JP)
Filed by Recruit Co., Ltd., Tokyo (JP)
Filed on Apr. 4, 2023, as Appl. No. 18/295,735.
Application 18/295,735 is a continuation of application No. 17/008,572, filed on Aug. 31, 2020, granted, now 11,620,448.
Prior Publication US 2023/0281390 A1, Sep. 7, 2023
Int. Cl. G06F 40/284 (2020.01); G06F 16/35 (2019.01); G06F 18/211 (2023.01); G06N 7/01 (2023.01)
CPC G06F 40/284 (2020.01) [G06F 16/35 (2019.01); G06F 18/211 (2023.01); G06N 7/01 (2023.01)] 20 Claims
OG exemplary drawing
 
1. A non-transitory computer readable storage medium storing instructions that are executable by a review comprehension system that includes one or more processors to cause the review comprehension system to perform operations for natural language processing, the operations comprising:
receiving input text;
extracting, from the input text, a modifier and aspect pair;
generating a first data structure based on the modifier and aspect pair;
generating a second data structure based on the modifier and aspect pair;
computing a dense representation of the modifier and aspect pair based on the first data structure and the second data structure;
generating candidate premise-conclusion pair based on the cosine similarity of the modifier and aspect pair to other modifier and aspect pairs; and
including premise-conclusion pair in a database, wherein the candidate premise-conclusion pairs relationship is verified using a machine learning model to other candidates for marking for inclusion in the database.