TriagerX: Dual Transformers for Bug Triaging Tasks with Content and Interaction Based Rankings

2025-08-23 · Source: cs.AI updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Expert, quick

Summary

TriagerX is a novel bug triaging system designed to improve developer and component recommendations by addressing limitations in existing Pretrained Language Models (PLMs). Unlike single-transformer architectures, TriagerX employs a dual-transformer setup, generating content-based recommendations from the last three layers of each transformer. This approach enhances the capture of token semantics in bug reports. The system then refines these recommendations using an interaction-based ranking methodology, which incorporates developers' historical engagement with similar fixed bugs. Across five datasets, TriagerX surpassed nine transformer-based methods, improving Top-1 and Top-3 developer recommendation accuracy by over 10%. Deployed with an industry partner, TriagerX also outperformed state-of-the-art baselines by up to 10% for component recommendations and 54% for developer recommendations on the partner's dataset.

Key takeaway

For research scientists developing bug triaging systems, TriagerX demonstrates that combining dual-transformer architectures with interaction-based ranking significantly boosts recommendation accuracy. You should consider integrating historical developer interaction data and multi-model content analysis to overcome the limitations of single-transformer PLMs, especially when aiming for high-precision developer and component assignments in large-scale development environments.

Key insights

TriagerX uses dual transformers and interaction history to improve bug triaging recommendations.

Principles

• Dual transformers enhance token semantics.
• Interaction history refines content-based rankings.

Method

TriagerX uses a dual-transformer architecture for content-based ranking, then refines these recommendations with an interaction-based ranking considering historical developer engagement with similar fixed bugs.

In practice

• Improve developer recommendation accuracy by >10%.
• Enhance component recommendations by up to 10%.
• Deploy for developer and component assignments.

Topics

• TriagerX
• Dual Transformers
• Bug Triaging
• Content-based Ranking
• Interaction-based Ranking

Best for: Research Scientist, AI Scientist, Machine Learning Engineer, Software Engineer

Related on AIssential

• Metis: Bridging Text and Code Memory for Self-Evolving Agents — Self-evolving agents benefit from a hierarchical dual-representation memory that bridges text and code for optimal performance.
• TriSearch: Learning to Optimize Triangulations via Bistellar Flips — TriSearch uses RL with circuit-supported subtriangulation to optimize polytope triangulations via bistellar flips, generalizing across dimensions.
• A beginner’s guide to Tmux: a multitasking superpower for your terminal — Tmux enhances terminal multitasking by enabling persistent, split-pane workspaces, crucial for multi-agent workflows.
• Working Notes on Late Interaction Dynamics: Analyzing Targeted Behaviors of Late Interaction Models — Late Interaction models exhibit length bias, especially in causal multi-vector architectures, while MaxSim effectively uses top token similarity.
• The Firefox security bug spike wasn't just about the model—it was about the harness — Improved LLM harnesses, not just models, reduced inaccurate AI-generated bug reports in open-source projects.
• Does Traversal Order Matter? A Systematic Study of Tree Traversal Methods in Transformer Grammars — Systematically exploring Breadth-First and a novel Production-Rule Traversal significantly impacts Transformer Grammar performance.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.AI updates on arXiv.org.