Towards automatic business process redesign: an NLP based approach to extract redesign suggestions

• Published in: Automated software engineering Vol. 29; no. 1; p. 12
• Main Authors: Mustansir, Amina, Shahzad, Khurram, Malik, Muhammad Kamran
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.05.2022; Springer Nature B.V
• Subjects: Annotations; Artificial Intelligence; Business process management; Business process reengineering; Computer Science; Customer feedback; Effectiveness; Feedback; Natural language; Natural language processing; Software Engineering/Programming and Operating Systems; Special Issue on Software Engineering Automation: A Natural Language Perspective; Subject specialists; User feedback; Supervised learning techniques; Business process redesign; Application of NLP in BPR; Deep learning techniques; Software engineering
• ISSN: 0928-8910; 1573-7535
• DOI: 10.1007/s10515-021-00316-8

Business process redesign (BPR) is widely recognized as a key phase of the business process management lifecycle. However, the existing studies have focused on proposing theoretical models, methodologies, and redesign patterns, whereas, the BPR activity remains dependent upon domain experts with little or no consideration to end-user feedback. To facilitate these experts, in this study, we have proposed a natural language processing (NLP) based approach to identify redesign suggestions from end-user feedback in natural language text. The proposed approach includes a novel set of annotation guidelines that can be used to generate computational resources for business process redesign. Secondly, to demonstrate the effectiveness of the proposed approach, we have generated computational resources which are composed of three real-world business processes and end-user feedback containing 8421 sentences. Finally, we have performed 270 experiments using six traditional and three deep learning techniques to evaluate their effectiveness for the identification of redesign suggestions from raw text. The classified suggestions can be used by domain experts to prioritize the redesign possibilities, without going through the details of the customer feedback.