Proposal of a Method Extracting Strategic Phrases from Japanese Enterprise Disclosure Documents

Abstract

Text mining is spreading in various domains due to remarkable computational capabilities and huge accumulation of digital data. However, application of text mining to enterprises' disclosure documents written in Japanese is still less common than application to SNS (Social Networking Service) messages or product reviews. In this study, we propose a new method to extract strategic phrases/words which potentially related to enterprises' various strategies, policies, initiatives, action plans, and so forth, as clue to understand the deep knowledge of business dynamic mechanism. The proposed method is able to extract strategic phrases/words from the various types of enterprises' disclosure documents written in Japanese, through the process which consists of steps as follows: (1) stratified morphological analysis; (2) classification of strategic phrases/words; (3) elimination of inadequate strategic phrases by employing dependency analysis; also (4) XBRL parsing and new sentences extraction for formal annual securities reports, as optional step. Moreover, it was demonstrated that the extracted results by the proposed method includes valuable strategic phrases/words.