🤖 AI Summary
In software requirements engineering, ambiguity in natural-language requirements impedes automatic conflict detection; existing approaches suffer from poor generalizability and heavy reliance on handcrafted rules. This paper proposes a two-stage automated conflict detection framework: first, candidate conflicting requirement pairs are retrieved using sentence embeddings (BERT or Doc2Vec) and cosine similarity; second, a supervised semantic similarity model precisely classifies true conflicts. To our knowledge, this is the first work to introduce supervised semantic matching for requirement conflict detection. We further propose UnSupCDA, an unsupervised variant enabling zero-shot deployment in label-scarce scenarios. Extensive experiments across five domain-specific software requirements specification (SRS) datasets demonstrate that our method significantly outperforms keyword- and rule-based baselines, achieving high accuracy and strong cross-domain generalizability.
📝 Abstract
In the realm of software development, the clarity, completeness, and comprehensiveness of requirements significantly impact the success of software systems. The Software Requirement Specification (SRS) document, a cornerstone of the software development life cycle, delineates both functional and nonfunctional requirements, playing a pivotal role in ensuring the quality and timely delivery of software projects. However, the inherent natural language representation of these requirements poses challenges, leading to potential misinterpretations and conflicts. This study addresses the need for conflict identification within requirements by delving into their semantic compositions and contextual meanings. Our research introduces an automated supervised conflict detection method known as the Supervised Semantic Similarity-based Conflict Detection Algorithm (S3CDA). This algorithm comprises two phases: identifying conflict candidates through textual similarity and employing semantic analysis to filter these conflicts. The similarity-based conflict detection involves leveraging sentence embeddings and cosine similarity measures to identify pertinent candidate requirements. Additionally, we present an unsupervised conflict detection algorithm, UnSupCDA, combining key components of S3CDA, tailored for unlabeled software requirements. Generalizability of our methods is tested across five SRS documents from diverse domains. Our experimental results demonstrate the efficacy of the proposed conflict detection strategy, achieving high accuracy in automated conflict identification.
