Multi-granule Association Rule Mining Based on Quantitative Concept Lattice

It is important to mine different granule knowledge because high-level and more generational information is getting more interesting for users in knowledge discovery. As the main data reduction method, attribute-oriented induction (AOI) generally takes the simply statistical information from original databases into account, and it can only mine the knowledge with single attribute. However, in Hasse diagram of quantitative concept lattice, the generalization and specialization relationships between concepts can be expressed clearly, so it is easy to find attributes with the same cardinality in relational databases, as well as the appropriate thresholds. Association rule mining algorithm based on quantitative concept lattice can extract multi-granule knowledge in the multi-level and multi-attribute way. Therefore, different granule knowledge can be easily focused, and then the relationships of transforms between different granule knowledge can be discovered rapidly.