A Design for Commonsense Knowledge Enhanced High-Frequency Market Analysis System

摘要

To identify the impacts of public news on security market, we propose a common-sense knowledge supported news analysis method, and design a system architecture for the news incorporated market analysis system. The graph model of common-sense knowledge is used to extend the news feature set by a random walk method. News indicators including news sentiment and news relevance are measured by common-sense knowledge supported text mining techniques. Based on these ideas, we develop a prototype system and examine the intra-day market reactions to public news on Hong Kong stock market. Our finds have shown the effectiveness of using common-sense knowledge and news in the market analysis domain. It is our belief that the common-sense knowledge incorporated market analysis system would be great helpful to market surveillance bureaus, traders and investors on the security market.