Yahoo! For Amazon: Sentiment Extraction from Small Talk on the Web by Sanjiv R. Das, Mike Y. Chen
Abstract
The volume of discussion on the Web is growing rapidly. Stock message boards have been in vogue for almost a decade now, and discussion on these boards continues to increase in volume. Firms may follow such discussions to determine how investors feel about their stocks, regulators may observe them to ensure no market manipulation occurs, and investors may use the boards to express opinions and gather information for asset allocation decisions. To cope with the overwhelming volume of text on the boards we develop an algorithm for parsing out the sentiment about a stock from Web postings. Our algorithm consists of various classifier algorithms coupled together by a voting scheme. Empirical application of this algorithm suggests that tech-sector postings are related to stock index levels and to volumes and volatility. This algorithm may also be used to assess the impact on investor opinion of management announcements, press releases, third-party news, and regulatory changes.
Source: MANAGEMENT SCIENCE Vol. 53, No. 9, September 2007
Comments