Automated Extraction of Compliance Elements in Software Engineering Contracts Using Natural Language Generation
Gokul Rejithkumar, Preethu Rose Anish, Pratik Sonar, Smita S. Ghaisas · 2024
Software Engineering (SE) contracts are legally binding agreements governing software development, usage, related responsibilities, and rights between the parties involved. SE contracts contain obligatory clauses which encompass compliance elements such as Regulations, Standards, and Policies. Identifying these elements is essential for ensuring compliance by aiding in contract negotiation, assignment of the obligatory clauses to relevant departments within the organization and for identification of high-level software requirements. Non-compliance may lead to penalties, damaging an organization's reputation and trust. In this paper, we automate the extraction of compliance elements from SE contracts. We employed a text-to-text generation approach and conducted experiments using the T5 model for the automated extraction of compliance elements from obligatory contractual clauses. The text-to-text generation approach yielded a mean F1-score of 0.92 for ROUGE-L.