Most of the current data mining algorithms handle databases consisting of a single table. Recently few algorithms are proposed that mine data stored in multiple tables for association rules. The classical association rule mining implicitly assumes that all items in the data are of the same nature and/or have similar frequencies in the data. In real life applications this is not the case. So use of only one minimum support value for the whole database may not be sufficient. If minimum support is set high the rules that involve infrequent items will not be found and if minimum support is set very low then this may result in flood of rules. In this paper we present a technique that can allow a user to specify multiple minimum supports to take care of the natures of items and their varied frequencies in multiple-table database environment. We discuss previous approaches that partly address the related issues to present a couple of algorithms to address the issue of multiple support constraints in multiple table environment.
The key idea here is to use formal concept analysis and fuzzy membership criterion to partition the data space into clusters and provide knowledge through fuzzy lattices. The procedures, written here, are regarded as mapping or transform of the original space (samples) onto concepts. The mapping is further given the fuzzy membership criteria for clustering from which the clustered concepts of various degrees are found. Bucket hashing measure has been used as a measure of similarity in the proposed algorithm. The concepts are evaluated on the basis of this criterion and then they are clustered. The intuitive appeal of this approach lies in the fact that once the concepts are clustered, the data analyst is equipped with the concept measure as well as the identification of the bridging points. An interactive concept map visualization technique called Fuzzy Conceptual Frame Lattice or Fuzzy Concept Lattices is presented for user-guided knowledge discovery from the knowledge base.