Graph-Based Modelling of Concurrent Sequential Patterns

Abstract

Structural relation patterns have been introduced recently to extend the search for complex patterns often hidden behind large sequences of data. This has motivated a novel approach to sequential patterns post-processing and a corresponding data mining method was proposed for Concurrent Sequential Patterns (ConSP). This article reines the approach in the context of ConSP modelling, where a companion graph-based model is devised as an extension of previous work. Two new modelling methods are presented here together with a construction algorithm, to complete the transformation of concurrent sequential patterns to a ConSP-Graph representation. Customer orders data is used to demonstrate the effectiveness of ConSP mining while synthetic sample data highlights the strength of the modelling technique, illuminating the theories developed.

Discovering exclusive patterns in frequent sequences

“…Synchronization comes into play once the control node receives input on one of its incoming branches, at a point in the workflow process where multiple parallel activities converge into one single thread of execution. The concurrent sequential patterns mining and modelling proposed by Lu, Chen et al (2008, 2010 can discover and represent parallel split and synchronization patterns, i.e. it is applicable to this first aspect of workflow model design.…”

“…In this model, nodes of SPG correspond to items in a sequential pattern and directed edges are used to denote the sequence relation between two elements. The idea has been adapted for modelling concurrent sequential patterns (and called ConSP-Graph) with two special nodes, called fork nodes and synchronizer nodes respectively, to represent the concurrent relationship (Lu, Chen et al, 2010).…”

“…Structural relation patterns mining has motivated the graph-based modelling of concurrent sequential patterns (Lu, Chen et al, 2010) and has also been applied in the web mining context for the identification of concurrency across web access patterns (Lu, Keech et al, 2009). This paper will focus on exclusive patterns, proposing specifically exclusive sequential patterns, and a corresponding mining method is described and explained through a running worked example.…”

Applications of Concurrent Access Patterns in Web Usage Mining

“…The idea was adapted in [7] for modelling concurrent sequential patterns. Each page view can be represented as a node in a graph and the directed edge between two nodes indicates a sequential relation, i.e.…”

“…Without dwelling on the detail, in this context concurrent access patterns graph (CAP-Graph) is thus a graphical representation of CAPs which maps a function from a set of directed edges to a set of pairs of nodes (see [7]). The definition of CAPGraph is an extension of that for sequential patterns graph and therefore the method to construct SPG can be adapted for CAP modelling.…”