본문 바로가기
  • Home

Multi-stage News Classification System for Predicting Stock Price Changes

  • Journal of the Korean Society for Information Management
  • Abbr : JKOSIM
  • 2007, 24(2), pp.123~141
  • DOI : 10.3743/KOSIM.2007.24.2.123
  • Publisher : 한국정보관리학회
  • Research Area : Interdisciplinary Studies > Library and Information Science
  • Received : May 25, 2007
  • Accepted : June 23, 2007
  • Published : June 30, 2007

Woojin Paik 1 Moon Sun Shin 2 Myoung Hyoun Kyung 1 Kyung Soo Min 1 Hye Ran Oh 1 Chami Lim 1

1건국대학교
2안양대학교

Accredited

ABSTRACT

It has been known that predicting stock price is very difficult due to a large number of known and unknown factors and their interactions, which could influence the stock price. However, we started with a simple assumption that good news about a particular company will likely to influence its stock price to go up and vice versa. This assumption was verified to be correct by manually analyzing how the stock prices change after the relevant news stories were released. This means that we will be able to predict the stock price change to a certain degree if there is a reliable method to classify news stories as either favorable or unfavorable toward the company mentioned in the news. To classify a large number of news stories consistently and rapidly, we developed and evaluated a natural language processing based multi-stage news classification system, which categorizes news stories into either good or bad. The evaluation result was promising as the automatic classification led to better than chance prediction of the stock price change.

Citation status

* References for papers published after 2023 are currently being built.