IDENTIFICATION PROCESS RELATIONSHIP OF PROCESS MODEL DISCOVERY BASED ON WORKFLOW-NET

• Main Author: Rahmadi, Ferdi
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/22144
• Institution: Institut Teknologi Bandung

Process mining can be used to perform process analysis on service-based systems. One way is to do the discovery process model of the event log on the service. Unfortunately process models like workflow-net and process discovery algorithm used by process mining can not describe the interaction between processes in the service. Therefore, this thesis aims to provide a solution for process mining to model interaction between process on a service-based system by using Proclet as process modeling language. <br /> <br /> <br /> The first step to answer this problem is to perform an analysis of event log data model that can meet thecharacteristics of proclet. Then the second step is to <br /> <br /> perform analysis to identify the process relation from the data model event data log filed in the first stage. Then the third stage is to design software calledProcshipFinder to be able to implement the method produced in the second stage. <br /> <br /> <br /> There are two case studies to test the proposed method. The first case study is to reverse engineer the proclet model of Meeting and Personal Entry that already existed. The purpose of this case study is to prove that the method can be remodeled identically using the proposed event log data model. While in the second case study, will use the data in the interview process by using video recording. The system in the second case study will use three services, namely Recruiter, Candidate and Server. Event logs in each service are taken from various sources such as email logs, transaction data and event log data. The purpose of the second case study is to test ProcshipFinder with data on the real environment. <br /> <br /> <br /> From the results of this study, it can be concluded that Proclet can be used to model the interaction of processes on the service because port components in Proclet have the capability to model cardinality and multiplicity processes, the next conclusion is the identification of process relations with cardinality ports (*), (+) and (1)can be determined by looking at the number of unique performative recipients on all process instances. Port cardinality with relation with (?) Notation in this thesis can not be identified because the method proposed in the thesis does not check the event that has no interaction. The last conclusion is that there are many challenges in doing event correlations between proclet. This is because each information system has a different structure in storing event information and knowledge needed on the domain of problems that will be done process mining. <br /> <br />