カンファレンス (国内) ニュース記事を用いた経済専門用語のクラスタリングと極性付与

伊藤 友貴(東京大学)、坪内 孝太山下 達雄、和泉 潔(東京大学)

2016年度人工知能学会全国大会(第30回) (JSAI2019)


In the previous research, a new approach for giving the financial terms whose positive-negative polarity score is unknown, and making the feature vector of a document useful for predicting stock price trends was proposed. Our subject is to evaluate the usefulness of this approach in predicting. First, we assigned a numerical vector to a word appeared in financial news documents, and defined the feature vectors of documents. Then, we analyzed the stock price trend and the sentiment score which can be evaluated from the textual data of Yahoo! finance board using this feature vector. As a result of comparison with other traditional methods, the proposal method could forecast in higher accuracy about the stock price trends and the sentiment scores.

Paper : ニュース記事を用いた経済専門用語のクラスタリングと極性付与 (外部サイト)