International Conference on Computer and Knowledge Engineering

Home / 15th International Conference on Computer and Knowledge Engineering

Enhanced Duplicate Bug Report Detection in Anonymized Environments: A Parallelized Multi-Task Learning Framework

Authors :

Alireza Shorafa¹ Abolfazl Zarghani²

1- Shiraz university 2- Ferdowsi university of mashhad

Keywords :

Duplicate Bug Report Detection،Anonymized Environments،Multi-Task Learning،BERT

Abstract :

Duplicate Bug Report Detection (DBRD) is pivotal for streamlining software maintenance, particularly in anonymized environments where user identifiers are absent. We introduce an advanced, parallelized architecture that integrates clustering, multi-modal feature extraction, and multi-task learning to tackle these challenges. By employing BERT for semantic text embeddings, HDBSCAN for density-based clustering, Latent Dirichlet Allocation (LDA) for topic modeling, and a multi-task learning (MTL) framework for simultaneous retrieval and classification, our approach achieves an F1-Score of 0.90 and Recall@5 of 0.88 on anonymized Eclipse and Mozilla datasets. Parallel processing enhances computational efficiency, while privacy-preserving techniques ensure compliance with ethical standards. Comparative evaluations against state-of-the-art methods (e.g., Cupid, REP, SABD) reveal 5-15\% performance improvements, establishing our framework as a leading solution for academic research and large-scale bug tracking systems.