Investigating TYPE constraint for frequent pattern mining

Abstract

Frequent pattern mining is one of the hottest fields of research in data mining. It has proven quite valuable for a variety of pattern analyses. However, a major limitation of existing or traditional frequent pattern mining techniques is the extraction of a huge number of patterns, which complicates the analysis performed by the users. To reduce these overwhelming patterns; a refined pattern mining technique that extracts only useful and to-the-point patterns for the users, is required. Finding such patterns of user’s interest from large databases is a challenging research issue, which can be mitigated via constraint-based data mining. In this scenario, a pattern mining technique that incorporates specific interests (constraints) of the user could result in a substantial reduction in the number of patterns generated. In this paper, a similar technique has been proposed and investigated. In this regard, firstly all the well-defined and categorized constraints are surveyed. From the survey, it is concluded that an important constraint “type” is referred to in the literature but not categorized based on its characteristics. One of the contributions of this paper is categorization of this important constraint “type”. Further, in this paper, the “type” constraint is divided into four modes for the sake of characterization and algorithms to employ the “type” constraints are also proposed. Moreover, the amalgamated constraints are proposed and characterized. Finally; comparative study of the proposed work with the existing techniques is presented and it is found that the proposed technique is promising in terms of reduced computational complexity.

Subject Classification:

• 35B36

Keywords:

• Pattern matching
• Data mining
• Knowledge discovery
• Frequent pattern analysis