Research On Error Mining Technology Of Business Process Control Flow

With the advance of the cloud computing and big data technology,the business processes supported by information systems have generated massive event logs.Based on these event logs,companies can perform applications and analysis such as process mining to provide intelligent decision making for the enterprise.In contrast to traditional process analysis and current big data analysis techniques,process mining particularly aims to discover useful process knowledge from actual execution histories(event logs)of business process,which has recently received a growing attention.However,discovering control flow errors from event logs has not been considered in the state-of-the-art process mining techniques.For example,deadlock and lack of synchronization.In addition,existing process mining techniques presuppose that event logs are generated by sound business process execution,so existing process mining methods cannot be directly applied to business process event logs with control flow errors.This paper aims to find two types of control flow errors,deadlocks and lack of synchronization from the event logs.The specific research contents include:1.We propose a business process control flow errors mining method.First of all,we preprocess the event log to obtain sub-event logs for detecting deadlocks and lack of synchronization.Secondly,we mine the sub-event logs to obtain two sets of event relationships.Finally,according to the definition of deadlock and lack of synchronization,the event relationships and ? algorithm are used to obtain the process fragments with control flow errors.2.We designed and implemented the Java language-based control flow errors mining prototype tool DCFE and packaged it into a ProM plugin.The DCFE input is an event log in XES format,and the output is a process fragment containing control flow errors in PNML format.3.Based on 25 event logs generated by business process execution with deadlocks or lack of synchronization,the prototype tool DCFE was used to conduct mining experiments,and case analysis is performed on the real log.The results of experiments and case analysis clarify the effectiveness of our method.