A PROPOSING FOR MINING K SEQUENTIAL PATTERNS

Abstract

The problem of finding k sequential patterns from a sequence database has been researched and proposed to solve the problem of how users can choose a minimum support threshold to find the number k desired user patterns in the sequential pattern mining problem. This solution is highly appreciated because the implementation cost is much lower than the traditional sequential pattern mining problem. However, currently, the process of finding sequential patterns in the problem of finding k sequential patterns is built according to the pattern development and projection database approach, so it costs a lot of money to perform projections. To reduce execution time, this paper proposes to use prism block encoding method to represent pattern information and calculate pattern support. Experimental results show that the proposed method has better execution time than the projection method on projection database.