| Summary: | 碩士 === 中國文化大學 === 資訊管理研究所 === 90 === With prevalence of Internet, users can easily retrieve the information what they want from Internet. Information explosion shows that efficient information summarization is aspired to all users. Therefore, an efficient knowledge management methodology becomes very important. Some technologies, such as text mining, for acquiring knowledge from huge amount of electronic documents are recognized as important technology in this field.
This work focuses on the applications of text-mining on Chinese industrial news and knowledge discovery. We use information extract method to extract news into companies, event keywords, time, locations, and persons categories based on the characteristics of news. The set of five extracted slots is called information template. The templates are summarized by rule induction. Such that, we can discover unexpected knowledge from these summarized rules. We built an integrated industrial news text mining model by using induction rule learner. This model is suitable to manipulate rules in bag-of-word form. Furthermore, we proposed interestingness to measure interesting strength of rules and accuracy to measure the rule confidence. The users can analyze the discovered rules based these two measures. These are helpful to discover unexpected knowledge. It is meaningful to commercial activities if we can discover valuable rules.
This work intents to build an automatic Chinese news text-mining model. This model induce large amount of news to produce a rule set for querying purpose. Furthermore, we introduce interestingness and accuracy to help users finding valuable rules. Besides industrial news application, we believe this model is suitable for knowledge discovery application in other fields.
|
|---|
