TFIDFを用いた災害・危機に関する言語資料体からのキーワード自動抽出手法の外的妥当性の検証

概要

論文の詳細を見る
In this study, the external validity of the automatic keyword detection method using TFIDF (Term Frequency/Inversed Document Frequency) which we developed for identifying chronological changes in keywords from a total of 2,623 newspaper articles on the 2004 Niigata-ken Chuetsu earthquake. We applied this keyword extraction method for the 19 corpuses of the digital news text data taken from 13 crisis events including both natural hazards and social hazards. As a result, a total of at least 500 news articles should be acquired as the corpus to be analyzed before this method detect a "meaningful set of keywords", which can be automatically determined with a 89.5% accuracy by taking into account the maximum number of keywords to be detected and the threshold of TFIDF deviation value.
地域安全学会の論文