An efficient pruning method for mining inter-sequence patterns based on pseudo-IDList

Anh Nguyen

doi:10.1016/j.eswa.2023.121738

2024-03-01

PRE

DOI：10.1016/j.eswa.2023.121738

原文链接

原文求助

摘要

En 中文

Mining inter-sequence patterns represents a relatively novel research direction, yielding significant contributions to the field of data mining research since its inception. Its versatile applications span various domains, with numerous research avenues emerging as a result. However, it is worth noting that inter-sequence mining algorithms tend to generate a substantial volume of frequent patterns. To date, research efforts in the realm of inter-sequence pattern mining have predominantly focused on creating, copying, and storing all candidate information. This approach has led to potential performance inefficiencies, necessitating more resources than may be inherently required. It is worth highlighting the recent introduction of the pseudo-IDList data structure, which has demonstrated promise in enhancing both runtime efficiency and memory utilization during mining tasks. Nevertheless, this innovation has thus far found application exclusively in clickstream mining and sequential pattern mining contexts.In this paper, we propose an extension of this data structure for the mining of frequent inter-sequence patterns and introduce a novel algorithm named Inter-Sequence Pattern mining based on Pseudo-Index. Additionally, our proposed algorithm incorporates a pruning method known as Inter-Sequence Pattern Mining with Index Inter-section Checking, designed to effectively curtail the proliferation of generated candidates. Our experimental results unequivocally demonstrate that the proposed algorithm excels in processing time and storage space utilization when compared to existing state-of-the-art algorithms designed for mining inter-sequence patterns.

Keyword:

Data mining

Inter-sequence pattern

Pseudo-IDList

Pruning technique