gPROFIT: A Tool to Assist the Automatic Extraction of Business Knowledge From Legacy Information Systems

Abstract

Business digitization is a crucial strategy for business growth in the 21st century. Its benefits include improving business process automation, customer satisfaction, productivity, decision-making, turnover, and adaptation to market changes. However, digitization is not a trivial task. As a major paradigm and mindset shift, it involves a lot of effort within an organization and therefore requires commitment from employees and managers. This is especially critical in companies whose business processes are mostly reliant on legacy information systems (LIS), which are usually specialized and based on technological architectures that could be considered obsolete. The replacement of these systems by more recent, process-oriented technologies, the building up of employees’ know-how and the continued use of outdated documentation are difficult, expensive tasks that hinder the initiation of continuous improvement processes in companies. This paper proposes techniques for finding and extracting process models from legacy databases. Specifically, it ( ${i}$ ) lays the theoretical foundations of a model-driven framework for systematically extracting business process models (conform to standard BPMN notation) from LIS considering process time perspective, and (ii) proposes a technological tool called gPROFIT, which uses machine learning techniques to support that theoretical framework, facilitate its use in real environments and extract the business knowledge embedded in such legacy systems. The paper also presents proofs-of-concept showing how our proposal has been validated in several legacy systems.

Keywords

• Business digitization
• model-driven engineering
• process mining
• legacy information systems
• machine learning techniques