Beyond Keywords: Understanding Bug Report Features Impact on Fault Localization

Read full text Summarise with VIRAT
Authors: anonymous
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Bug Reports Analysis This repository is a replication package for the paper Beyond Keywords: Understanding Bug Report Features Impact on Fault Localization. Repository Structure bug-reports-analysis/ ├── run_full_pipeline.py # End-to-end pipeline runner ├── bug_feature_extraction/ # Feature extraction scripts │ ├── extract_bug_features.py │ ├── gemini_bug_categorization_overall.py │ ├── gemini_bug_ratings.py │ └── fine_grained_gemini_catg.py ├── final_feature_set.csv #Post pruning feature set which analysis was performed on ├── removed_features_log.txt #All features removed during pruning ├── merge_features_and_performance.py #Merges output from scripts above with tool performance results, and performs redundancy analysis ├── unified_analysis.py #Outputs upset diagrams, and all vs none and tool vs rest comparisons (RQ3) ├── predictor.py #Performs prediction tasks from RQ4 ├── requirements.txt ├── tool_comparison_summary.csv #Tool performance for each bug ├── prompts_and_regex.md #All prompts and regex used for feature extraction └── defects4j_xml/ # Defects4J XML inputs (used by extraction) Run the pipeline once to generate all results (feature integration, statistical comparisons, complementarity visualizations, and prediction outputs). Installation python -m venv venv source venv/bin/activate # On Windows: venv\Scripts\activate pip install -r requirements.txt python -m spacy download en_core_web_sm Run the pipeline python run_full_pipeline.py Preview commands without executing: python run_full_pipeline.py --dry-run Unified analysis outputs (tool_comparison_results_fixed/) The unified analysis stage (implemented in unified_analysis.py) generates outputs for Top-1, Top-5, and Top-10. Tool complementarity (UpSet diagrams): tool_comparison_results_fixed/upset_diagram_top1.png tool_comparison_results_fixed/upset_diagram_top5.png tool_comparison_results_fixed/upset_diagram_top10.png All tools vs none (feature differences): tool_comparison_results_fixed/all_vs_none_top1.csv tool_comparison_results_fixed/all_vs_none_top5.csv tool_comparison_results_fixed/all_vs_none_top10.csv Bug difficulty spectrum (by number of tools that succeed): tool_comparison_results_fixed/bug_difficulty_spectrum_top1.csv tool_comparison_results_fixed/bug_difficulty_spectrum_top5.csv tool_comparison_results_fixed/bug_difficulty_spectrum_top10.csv Each tool vs the rest (unique signals): tool_comparison_results_fixed/tool_vs_rest_top1.csv tool_comparison_results_fixed/tool_vs_rest_top5.csv tool_comparison_results_fixed/tool_vs_rest_top10.csv Predictor outputs (delta_score_outputs/) The predictor stage writes: delta_score_outputs/cv_fold_results.csv delta_score_outputs/cv_auc_summary.csv delta_score_outputs/score_vs_success_top5.csv delta_score_outputs/full_dataset_deltas.csv delta_score_outputs/full_dataset_scaling.csv delta_score_outputs/cascade_routing_topk_sweep.csv delta_score_outputs/plots/ Notes 5. Review correlations_any_tool.csv Find (Example): clarity correlates 0.45 with detection (p < 0.001) 6. Action: Update bug report template to emphasize clarity Actionable Insights For Bug Report Writers: Increase clarity and specificity (if these correlate with detection) Include stack traces (if has_stacktrace is important) Add code examples (if has_code helps) Improve readability (if readability metrics matter) For Tool Selection: Check tool-specific feature importance files Match tools to bug characteristics Use category_detection_*.csv to see which tool works best for your bug category For Tool Developers: Identify gaps: Clusters with low detection = opportunities Feature engineering: Focus on features that correlate with detection ## License © XXXX-1 Lab — XXXX-2. Distributed under the XXXX-1 Lab Research License (non-commercial use only).

Fields of Study & Subjects

Software Engineering Research Software Testing and Debugging Techniques Software Engineering Techniques and Practices

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
30 March 2026
Language
EN