Dear All,
SinaLab is pleased to release a new open-source dataset for
conflict detection and resolution in software requirements in Arabic.
26,545 requirement pairs manually annotated for conflict detection.
Beyond Software Engineering, the dataset can also support research on conflict detection in Arabic NLP more broadly.
We also developed an LLM-based conflict detection and resolution methods tested with ChatGPT, LLaMA Fanar, and DeepSeek, achieving an F1 score of 94%.