FAQ CANDIDATE EXTRACTION SYSTEM AND FAQ CANDIDATE EXTRACTION PROGRAM

摘要

PROBLEM TO BE SOLVED: To provide a FAQ candidate extracting system which is suitable for discourse data characteristics and extracts a FAQ candidate query from discourse data based on discourse semantics without specifying the framework of a sentence structure.SOLUTION: Discourse data 101 and discourse semantics 200 including flow information 21 of each statement are input into the FAQ candidate extracting system 1, and the FAQ candidate query from the discourse data 101 is extracted by and output from the FAQ candidate extracting system 1. The FAQ candidate extracting system 1 comprises: a query extracting part 60 for extracting query/request statements, to which a flow indicating query or request is set, uttered by a customer from the discourse data 101; a candidate extracting part 80 for extracting the query/request statement including specified keyword from the query/request statements; and a clustering part 90 for executing clustering of the query/request statements and outputting the typical query/request statement of each cluster as a FAQ candidate 300.