Mining Maximal Sequential Patterns without Candidate Maintenance

摘要

Sequential pattern mining is an important data mining task with wide applications. However, it may present too many sequential patterns to users, which degrades the performance of the mining task in terms of execution time and memory requirement, and makes it difficult for users to comprehend the results. The problem becomes worse when dealing with dense or long sequences. As a solution, several studies were performed on mining maximal sequential patterns. However, previous algorithms are not memory efficient since they need to maintain a large amount of intermediate candidates in main memory during the mining process. To address these problems, we present a both time and memory efficient algorithm to efficiently mine maximal sequential patterns, named MaxSP (Maximal Sequential Pattern miner), which computes all maximal sequential patterns without storing intermediate candidates in main memory. Experimental results on real datasets show that MaxSP serves as an efficient solution for mining maximal sequential patterns.