Research On Ontology-based Case Base Building And Reasoning For Theme Events In The Stock Markets

There are often occurring some events that will affect the stock price and profit in Chinese stock markets. In a similar market environment, similar theme events will have similar effects on the markets. Investors often need the help of knowledge and experience which comes from historical similar events to make decisions. However the events’information in stock markets is often formalized as unstructured texts, which can’t be represented by simple rules or models it is also difficult for investors to efficiently use this information to quickly retrieve similar events and further make decisions by the experimental knowledge they got. Thus, it is essential to collect and standardize the information of events.Cased-based reasoning (CBR), defines the addressed problems in the past as cases, and uses them to deal with new and similar ones. By retrieving the similar cases, CBR helps to make decisions by predicting the effects of the new events on the markets. Ontology, on the other hand, clearly defines the commonly recognized conceptions and terms in one certain field, uses the semantic model of domain knowledge to express the conceptions, and regulates the relationships between two different conceptions within the hierarchy, thus it has strong abilities to express knowledge and do reasoning. The combination of ontology and CBR integrates the knowledge expressed by cases and the general domain knowledge models given by ontology, which implies the ontology-based case reasoning can take full advantages of domain knowledge to obtain a more accurate and comprehensive result.In this dissertation ontology and case-based reasoning was integrated:firstly ontology was used to express the structures of cases, protege was applied to build case ontology and domain knowledge ontology; then a ontology-based case base was built for theme events, and finally a case reasoning experimental system was designed, where market volatility trends forecasting, cases management, cases learning and other modules were emphasized. In this system, the basic info of events and the market fluctuation info are stored in the case base, which can be helpful for investors to accumulate the empirical knowledge and further make decisions. The empirical results show that when a similar event happens in stock markets, this system can find out a similar case and further analyze the detailed information to help to predict the possible fluctuations and eventually give recommendations for investors.