Efficient Mining of Gap-Constrained Subsequences and Its Various Applications

Abstract

Mining frequent subsequence patterns is a typical data-mining problem and various efficient sequential pattern mining algorithms have been proposed. In many application domains (e.g., biology), the frequent subsequences confined by the predefined gap requirements are more meaningful than the general sequential patterns. In this article, we propose two algorithms, Gap-BIDE for mining closed gap-constrained subsequences from a set of input sequences, and Gap-Connect for mining repetitive gap-constrained subsequences from a single input sequence. Inspired by some state-of-the-art closed or constrained sequential pattern mining algorithms, the Gap-BIDE algorithm adopts an efficient approach to finding the complete set of closed sequential patterns with gap constraints, while the Gap-Connect algorithm efficiently mines an approximate set of long patterns by connecting short patterns. We also present several methods for feature selection from the set of gap-constrained patterns for the purpose of classification and clustering. Our extensive performance study shows that our approaches are very efficient in mining frequent subsequences with gap constraints, and the gap-constrained pattern based classification/clustering approaches can achieve high-quality results.

Mining sequential patterns with periodic wildcard gaps

“…(2) In some studies, only one or two positions of the given sequence are considered [23], whereas we consider that a group of positions are used to denote a support of a pattern in the given sequence. This means that each position p j should be considered.…”

“…Researchers have explored several major approaches to untangle this hard problem: (1) The set of length-m candidate patterns is used to generate a set of length-(m+1) candidate patterns. Then select frequent patterns from the set of candidate patterns and iterate this process till it is done [20,22,23]. (2) The Apriori-like property is employed to reduce the size of the set of candidate patterns [20].…”

“…Table 1 shows a comparison of related work. [26] No Apriori One-off All frequent patterns Ding et al [30] No Apriori Non-overlap All/closed frequent patterns Xie et al [27] Yes Apriori One-off All frequent patterns Li et al [23] Yes Apriori First-last Closed/long frequent patterns Ji et al [21] Yes Apriori-like All Minimal distinguishing subsequence patterns Min et al [24] Yes Apriori All All frequent patterns Zhang et al [20] Yes Apriori-like All All frequent patterns Zhu and Wu [25] Yes Apriori-like All All frequent patterns This paper Yes Apriori-like All All/Top-K frequent patterns for each length…”

“…Ji et al [21] proposed an algorithm to mine all minimal distinguishing subsequences satisfying the minimum and maximum gap constraints and a maximum length constraint. Li et al [23] proposed Gap-BIDE to discover the complete set of closed sequential patterns with gap constraints, and Gap-Connect to mine the approximate set of long patterns. They further explored several feature selection methods from the set of gap-constrained patterns on the issue of classification and clustering.…”

“…Some of the patterns with excessively high frequency have been determined as the suspected cause of some diseases [20,21]. In data mining, researchers found that some frequent periodic patterns can be used in feature selection for the purpose of classification and clustering [22,23]. So an extraordinary task is how to mine all these frequent patterns and the Top-K patterns for each length from a mass of data to improve its efficiency in each application field.…”