Building Fuzzy Blocks from Data Cubes
Résumé
Multidimensional databases have become very popular for decision making frameworks. In this context, huge amounts of data are stored in data warehouses and decision makers try and navigate through this data using OLAP tools in order to visualize and analyze it. Although navigating through the data is one of the key issues, many issues are still open, and users are still not provided with intelligent tools for automatically identifying relevant parts from the data. In this paper, we address this problem and we propose to mine homogeneous areas of the data, which we call blocks. In previous work, we have defined a level-wise method to automatically mine such blocks. However, these blocks are crisp in their definition, although they are described by fuzzy rules. We extend our previous work by proposing ways to mine fuzzy blocks, and we compare the three approaches, showing that fuzzifying blocks leads to more clearly defined areas from the data.