An Algorithm of Constraint Frequent Neighboring Class Set Mining Based on Filling Class Set

摘要

Since the algorithms of constraint frequent neighboring class set mining based on Apriori is unsuitable for mining any length constraint frequent neighboring class set and has some redundancy computing, this paper proposes an algorithm of constraint frequent neighboring class set mining based on filling class set, which may efficiently extract any length constraint frequent neighboring class set from large spatial database. The algorithm uses binary conversion to turn neighboring class set into integer, and regards these integers as mining spatial database, and it uses double search strategy to generate constraint frequent neighboring class set, namely, one is that the algorithm uses two k-constraint frequent neighboring class sets to connect (k+1)-candidate constraint frequent neighboring class set, the other is that it also uses filling virtual class set of (k+1)-candidate constraint frequent neighboring class set to generate another candidate. In whole mining course the algorithm need only scan database once. The experimental result indicates that the algorithm is more efficient than the constraint frequent neighboring class set mining algorithm based on Apriori when mining any length constraint frequent neighboring class set.