Selective association rule generator (Q626210)

The problem of mining association rules from large data sets is considered. A technology of mining is proposed where at the first step the frequencies of item sets in the data base are mined and stored in a prefix-tree structure. At the second step a rule-generation algorithm (e.g., the Apriori algorithm) produces association rules. In this approach no expensive rule filtering is necessary and the combinatorial explosion due to infrequent item sets is avoided. It allows also the inclusion of some additional information in the rules production process. Applications to artificial and real-life data sets are discussed.