REDUCING SOFTWARE REQUIREMENT AMBIGUITY THROUGH NLP-BASED REQUIREMENT QUALITY ASSESSMENT

Abstract

Software requirement ambiguity remains one of the major causes of misunderstanding, rework, delayed development, and project failure in software engineering. Ambiguous requirements often lead to inconsistent interpretation among stakeholders, developers, testers, and project managers, which directly affects software quality and delivery performance. This paper presents an NLP-based requirement quality assessment approach for identifying and reducing ambiguity in software requirement specifications. The proposed approach evaluates requirement statements by analyzing linguistic features such as vague terms, incomplete conditions, passive constructions, uncertain quantifiers, weak modal verbs, and semantic inconsistency. Natural Language Processing techniques are applied to classify requirements into different quality levels and highlight statements that require improvement. The results indicate that NLP-based assessment improves ambiguity detection accuracy, supports early requirement validation, and reduces manual review effort. The findings show that automated linguistic analysis can help requirement engineers detect unclear statements before the design and implementation phases. Overall, the study demonstrates that integrating NLP into requirement engineering practices can enhance requirement clarity, improve stakeholder agreement, and support the development of more reliable software systems.