A double search mining algorithm in frequent neighboring class set

Abstract

This paper addresses the existing problems that present frequent neighboring class set mining algorithms is inefficient to extract long frequent neighboring class set in spatial data mining, and introduces a double search mining algorithm in frequent neighboring class set, which is suitable for mining any frequent neighboring class set in large spatial data through down-top search strategy and top-down search strategy. Firstly, the algorithm turns neighboring class set of right instance into digit to create database of neighboring class set, and then generates candidate frequent neighboring class set via double search strategy, namely, one is that it gains (k+1)-neighboring class set as candidate frequent items by computing (k+1)-superset of k-frequent neighboring class set, the other is that it gains l-neighboring class set as candidate frequent item by computing l-subset of (l+1)-non frequent neighboring class set. The mining algorithm computes support of candidate frequent neighboring class set by AND operation. The algorithm improves mining efficiency through these methods. The result of experiment indicates that the algorithm is faster and more efficient than present algorithms when mining frequent neighboring class set in large spatial data.