A New Trace Similarity Measurement Based On Constraints Of Process Mainstream And It's Applications

Business processes are precious intellectual assets of every organization as they control the revenue potential by shaping the resource profile of an organization.That's why enterprises and academics strive to model and optimize the inter-organization and cross-organization processes.It's deserved that the accumulated processes have achieved a marvelous number.The growth of process data makes it possible to analyze events and excavate the rules of business process.Process mining which aims at discovering,monitoring and improving real processes by extracting knowledge from event logs recorded by information systems is one of the most promising research in process modeling and optimization.The innovative work of this thesis mainly manifests in the following several aspects:(1)To solve the problem that contemporary trace data processing techniques ignore the different importance of different activities,this thesis proposes the constrained trace similarity measure.Firstly,some typical causal sequences that reflecting the process backbone are identified.Subsequently,combined with the common subsequences,these sequences are exploited as constraint conditions to guarantee the priority of important activities in traces and retain the information implied in the original data as much as possible.(2)To solve the problem that currently available discovery algorithms are unable to generate accurate and comprehensible process models out of event logs stemming from highly complex environments,this thesis proposes the constrained trace clustering.It uses the constrained trace similarity measure as clustering bias and then adapts two applicable clustering strategies that are agree with the process mining perspective.Experimental results show that our constrained trace clustering delivers compact and comprehensive synopses of flexible process to support their analysis and improvement.(3)To solve the problem that there is always a gap between the structure bias and the behavior bias from which contemporary process similarity measures suffer,this thesis proposes the constrained process similarity measure.Based on the constrained trace similarity measure,it bypasses the process models and directly calculates the process similarity by comparing the corresponding log data.Experimental results show that our constrained process similarity measure provides a refined measurement.