An NLP-based System for Automated Compliance Analysis and Requirement Classification in Engineering Applications

Abstract

Regulatory compliance is a critical challenge in engineering product development, particularly in industries governed by complex and frequently evolving standards. This research, conducted in collaboration with Volvo Penta, explores the use of Natural Language Processing (NLP) techniques to automate the classification and interpretation of regulatory clauses in support of early-stage requirements analysis, with a focus on ensuring traceability, usability, and transparency in the decision making process. The proposed method integrates rule-based preprocessing, domain specific keyword enrichment, semantic similarity retrieval using dense embeddings, and reasoning about individual regulatory statements using large language models (LLMs). The system is connected to Volvo Penta’s SystemWeaver platform to sup port traceable, explainable, and human-reviewable predictions at the clause level. By automating the identification and interpretation of relevant regulatory content, the system reduces manual workload, improves consistency, and enables scalable compliance workflows. Evaluation through both quantitative analysis and expert feedback indicates strong alignment with human judgment and demonstrates the system’s value as a decision-support tool in industrial engineering compliance processes. While limitations remain in handling edge cases and ambiguous regulatory language, results suggest that NLP-driven methods can meaningfully support scalable, traceable, and more efficient compliance processes in industrial engineering settings.