Conclusions and Future Work
- 97 Downloads
Research into multi-database mining has become important, imperative, and challenging with the increasing development of multi-databases. Although data in multi-databases can be merged into a single dataset for knowledge discovery, such merging can lead to many problematic issues such as tremendous amounts of data, the destruction of data distributions, and the infiltration of uninteresting attributes, as well as security and privacy concerns. In particular, some concepts, such as regularity, causal relationships, and rules cannot be discovered if we simply search the merged single dataset, since the knowledge is essentially hidden within the multi-databases (Zhong-Yao-Ohsuga 1999). To deal with multi-database mining problems, this book has designed a new and effective strategy for multi-database mining. This strategy consists of a process for analyzing local patterns, which allows (a) the achievement of high performance in novel and quality pattern discovery; and (b) pattern discovery to become a realistic support for dual-level applications. We now conclude by outlining the key techniques presented in this book and looking towards the future.
Unable to display preview. Download preview PDF.