Data Cleaning for Process Mining with Smart Contract

Abstract

Process Mining (PM) is a special data mining technique that allows extracting information from data of critical transactions (i.e. event logs) carried out in Information Systems and monitors the patterns in these transactions. When we start to process event logs with process mining tools, we face with data quality problems such as incorrect and insufficient logging and timing. Thus, data cleaning operations must be applied to event logs before applying process mining on these logs. Being an innovative medium of distributed data processing and storage with the features of enhanced security, traceability, automated transaction verification and integration, Blockchain Technology and Smart Contracts might be a good option to process and store event logs for process mining. In this paper, we focused on the cleaning of the event logs by smart contract as data is flowing from the information systems into the blockchain, and used Hyperledger Composer by IBM to develop our solution. We tested our proposal on an open process data of 1555 records, and compared the cleaning performance of our proposal with that of DataWrangler by Stanford University. Our proposal not only cleaned all 1313 records identified and cleaned by DataWrangler, it also saved 12 additional records with a different date format that was caught and corrected by our smart contract implementation.