Mining Supplemental Frequent Patterns

The process of resource distribution and load balance of a distributed P2P network can be described as the process of mining Supplement Frequent Patterns (SFPs) from query transaction database. With given minimum support (min_sup) and minimum share support (min_share_sup), each SFP includes a core frequent pattern (BFP) used to draw other frequent or sub-frequent items. A latter query returns a subset of a SFP as the result. To realize the SFPs mining, this paper proposes the structure of SFP-tree along with relative mining algorithms. The main contribution includes: (1) Describes the concept of Supplement Frequent Pattern; (2) Proposes the SFP-tree along with frequencyAscending order header table FP-Tree (AFP-Tree) and Conditional Mix Pattern Tree (CMP-Tree); (3) Proposes the SFPs mining algorithms based on SFP-Tree; and (4) Conducts the performance experiment on both synthetic and real datasets. The result shows the effectiveness and efficiency of the SFPs mining algorithm based on SFP-Tree.