Generation of Process Sequence Based on Implicit Temporal Overlap Function

Abstract

Activities within processes occur in sequence, and the discovering of these sequences is an essential step and of great significance to process mining. This paper is aimed at intelligently discovering process sequences that lie within the helpdesk unit event log, which was primarily obtained from the 4TU repository. Explicit approaches have mostly being applied to mining rules and little attention given to sequences that can be generated via implicit approach. Hence, an implicit approach to association rule discovery was adopted using the modified temporal overlap scoring module (TOSM). The module was implemented using Java programming language. The experimental results showed that the temporal overlap module discovered sequences in an intelligent manner by factoring in the overlap property and identifying hidden dependencies. The resulting association rule generated for each sequence, as represented in the lift value, was recorded as significant to the entire log as compared to that of the explicit approach.

Keywords: Hierarchical Temporal Memory, Overlap, Process Aware Information System, Event Logs, Process Sequence.

DOI: 10.7176/CEIS/11-4-04

Publication date:June 30^th 2020