INFORMATION EXTRACTION METHOD, DEVICE, AND PROGRAM

摘要

PROBLEM TO BE SOLVED: To extract a suitable candidate when information in related category is extracted from a document, and to prevent increase of learning cost.SOLUTION: An event name candidate extraction function 15 removes a substring when a text length ratio of a substring to a text length of another event name candidate does not reach to removal ratio λ when the event name candidate is the substring of another event name candidate. Thereby, a suitable event name candidate is extracted, the number of event name candidates is reduced for preventing increase of learning cost.