Mining algorithms for generic and biological data

With computer technologies developing so fast, especially since the appearance of the Internet, people have experienced the era of the data explosion. As a result, the sizes of normal databases nowadays could be hundreds of gigabytes or even terabytes. On the other hand, people's abilities to analyze the collected data are limited. This contradiction creates the need to generate new technologies and tools to analyze the collected data intelligently and efficiently, which sparks the emergence of knowledge discovery in databases (KDD) and data mining.;KDD and data mining could be applied to all kinds of data. The most common application domain of the KDD and data mining techniques is the business databases. Applying association rules and sequential pattern techniques on the market database, a store manager can infer the connections among the commodities sold by the store and thus can manage to promote the sale of various goods and obtain better marketing performance. Another common application domain is the biological databases. The techniques of data mining such as classification and pattern matching have helped people decode the human genetic code.;This thesis is concerned with the development of efficient methodologies for association rule mining. We propose to develop techniques that can be applied to a generic database independent of the application. Generic techniques need not perform uniformly well on all the applications of concern. It is often possible to design techniques specific to a given application that will outperform generic techniques. This is indeed true for data mining as well. Thus, in addition to developing generic data mining techniques, we will focus on discovering efficient analysis tools for biological data too.