- News sites should show users especially important news. Selection of important news is conducted by experts since criteria of importance are not obvious and algorithmic judgement is difficult. In order to lighten the work of experts, we propose the method that selects important news by using crowdsourcing. According to results of experiments, we could estimate the importance of news with high accuracy using the dataset collected by pairwise comparison. In addition, we applied the method that generates criteria of classification to this problem and confirmed it may create features which help to discover important news.
クラウドソーシングによる重要ニュース選択（外部サイト／External Site Link）