USING SOFT COMPUTING TECHNIQUES TO INTEGRATE MULTIPLE KINDS OF ATTRIBUTES IN DATA MINING

摘要

Data mining discouvers interesting information from a data set. Mining incorporates different methods and considers different kinds of information. Granulation is an important task of mining. Mining methods include: association rule discouvery, classification, partitioning, clustering, and sequence discouvery. The data sets can be extremely large with multiple kinds of data in high dimensionality. Most current clustering algorithms deal with either quantitative or qualitative data, but not both. However, many data sets contain a mixture of quantitative and qualitative data. We are considering how to best group records containing multiple kinds of data. It is difficult to do this. Even grouping based on different quantitative metrics has its difficulties. There are many partially successful strategies as well as several different possible differential geometries. Adding in various qualitative elements is exceedingly difficult. We expect to use a mixture of scalar methods, soft computing (rough sets, fuzzy sets), as well as methods using other metrics. To cluster records in a data set, it would be useful to have a similarity measure. Unfortunately, few exist that account meaningfully for any combination of kinds of data. The more meaningful metrics known are restrictive to a particular area or science. The method of combining magnitude difference and simple matching so that it is general enough for mining is a topic that is yet to be reasonably solved. We will present several strategies for integrating multiple metrics.