Similarity Retrieval and Cluster Analysis Using R* Trees

Data mining is aimed at the extraction of interesting (i.e., nontrivial, implicit, previously unknown, and potentially useful) patterns or knowledge from huge amounts of data. In order to make data mining manageable, data mining has to be database centered. Yet, data mining goes beyond the traditional realm of database techniques; in particular, reasoning methods developed from machine learning techniques and other fields in artificial intelligence (AI) have made important contributions in data mining. Data mining thus offers an excellent opportunity to explore the interesting fundamental issue of the relationship between data and knowledge retrieval and inference and reasoning. Decades ago, researchers made an important remark stating that since knowledge retrieval must respect the semantics of the representation language, knowledge retrieval is a limited form of inference operating on the stored facts (Frisch & Allen, 1982). The inverse side of this statement has also been explored, which views inference as an extension of retrieval. For example, Chen (1996) described a computer model that is able to generate suggestions through document structure mapping based on the notion of reasoning as extended knowledge retrieval; the model was implemented using a relational approach. However, although the issue of foundations of data mining has attracted much attention among data mining researchers (ICDM, 2004), little work has been done on the important relationship between retrieval and inference (or mining). A possible reason of lacking such kind of research is the difficulty of identifying an appropriate common ground that can be used to examine both data retrieval and data mining. On the other hand, from the database perspective, an effective way to achieve efficient Chapter LVIII Similarity Retrieval and Cluster Analysis Using R* Trees