US 12,242,477 B2
Semantic search based on a graph database
Teng Sun, Beijing (CN); Tong Liu, Xi'an (CN); Si Tong Zhao, Xi'an (CN); XueLiang Zhao, Shanghai (CN); Frank Feng, Beijing (CN); Yu Zui Wy You, Beijing (CN); and Zhong Fang Yuan, Xi'an (CN)
Assigned to International Business Machines Corporation, Armonk, NY (US)
Filed by International Business Machines Corporation, Armonk, NY (US)
Filed on Sep. 7, 2021, as Appl. No. 17/467,516.
Prior Publication US 2023/0076923 A1, Mar. 9, 2023
Int. Cl. G06F 16/2453 (2019.01); G06F 16/21 (2019.01); G06F 16/242 (2019.01); G06F 16/901 (2019.01); G06N 20/00 (2019.01)
CPC G06F 16/24542 (2019.01) [G06F 16/212 (2019.01); G06F 16/243 (2019.01); G06F 16/9024 (2019.01); G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented, operable on a computer system, the method comprising:
identifying, automatically by the computer system, a keyword from words in a natural language query;
selecting, automatically by the computer system, a first set of nodes semantically matching the keyword, a first node of the first set of nodes having at least one matched element corresponding to the keyword, wherein the selecting of the first set of nodes comprises:
determining that the keyword is associated with a further keyword related to entities in the natural language query based on at least one of syntax analysis or part of speech analysis;
binding words in the natural language query based on the determining that the keyword is associated with a further keyword; and
updating the first set of nodes using a number of nodes semantically matching both the keyword and the further keyword based on the binding words in the natural language query;
determining, automatically by the computer system, a query target from the natural language query;
identifying, automatically by the computer system, one or more target nodes included in the first set of nodes, the one or more target nodes corresponding to the query target;
selecting, automatically by the computer system, a path from candidate paths based on similarities between the candidate paths and a plurality of paths in the graph database, wherein:
a candidate path traverses at least one target node of the one or more target nodes and at least one node from the first set of nodes; and
a path of the plurality of paths traverses a subset of the plurality of nodes; and
generating, automatically by the computer system, a graph query for retrieving information from the graph database based on the selected path and the query target;
wherein the generating the graph query further comprises creating clauses in the graph query that correspond to the natural language query.