Mining frequent sequential patterns with first-occurrence forests

Abstract

In this paper, a new pattern-growth algorithm is presented to mine frequent sequential patterns using First-Occurrence Forests (FOF). This algorithm uses a simple list of pointers to the first-occurrences of a symbol in the aggregate tree [1], as the basic data structure for database representation, and does not rebuild aggregate trees for projection databases. The experimental evaluation shows that our new FOF mining algorithm outperforms the PLWAP-tree mining algorithm [2] and the FLWAP-tree mining algorithm [3], both in the mining time and the amount of memory used.

Mining Web Access Patterns using Root-set of Suffix Trees

“…First-Occurrence Forests [5] is a pattern growth mining technique used to mine Web log sequential patterns proposed by Erich A. Peterson and Peiyi Tang in March 2008. They use the simple Aggregate tree structure for mining and totally removed the concept of linked tree.…”

“…So to generate maximum number of patterns (where pattern size >= 2) the proposed approach needs number of traversing = n+n 2 +n 3 +…+n m-1 . The experimental results show that the proposed approach is faster than the WAP [2] and also FOF [5] because of its less number of tree traversing than others.…”

“…In recent years many web access pattern mining algorithms are proposed to extract significant information from web log database. The most important and popular web mining algorithms are GSP [1] which is an apriori based algorithm and WAP-tree [2], PLWAP-tree [3], FLWAP-tree [4], and FOF [5] mining algorithm etc which are pattern-growth approach. We know that, apriori based algorithms create lots of candidate sets during mining which degrades the performance of those algorithms.…”

“…There are also some approaches those avoid generating intermediate trees during mining. PLWAP [3] and FOF [5] are such algorithms which do not generate any intermediate tree during mining instead they use the concept of suffix tree. But PLWAP and FOF also have some drawback that reduces their performance.…”

“…It can generate all the patterns with same prefix by a single traverse through the tree while FOF mining [5] traverses the tree for each single pattern. Traversing for only a single pattern is an overhead of FOF mining [5] and our proposed approach overcomes this by maintaining a table which stores all the items and their corresponding counts that found during traversing for a specific prefix. After traversing, it generates the frequent patterns from the table and start traversing with new prefixes recursively.…”