Graph Neural Network vs. Large Language Model: A Comparative Analysis for Bug Report Priority and Severity Prediction
Bibliographic record
Abstract
A vast number of incoming bug reports demand effective methods to identify priority and severity for bug triaging. With increased technological advancement, machine learning and deep learning have been extensively examined to address this problem. Although Large Language Models (LLMs) such as Fine-tuned BERT (early generation LLM) have proven to capture context in the underlying textual data, severity and priority prediction demand additional features for understanding the relationships with other bug reports. This work utilizes the graph-based approach to model the bug reports and their other attributes, such as component, product and bug type information. It utilizes the relational intelligence of Graph Neural Network (GNN) to address the prioritization and severity assessment of bug reports in the Bugzilla bug tracking system. Initial tests on the Mozilla project dataset indicate that a project-wise predictive approach using GNNs yields higher accuracy in determining the priority and severity of bug reports compared to LLMs across multiple Mozilla projects, contributing to a notable advancement in the automation of bug severity and priority prediction tasks. Specifically, GNNs demonstrated a remarkable improvement over LLMs, increasing the priority prediction accuracy by 37% & 30% and severity prediction accuracy by 43% & 30% for Core and Firefox projects, respectively. Overall, GNN outperformed the Fine-tuned BERT (LLM) in predicting priority and severity for all the Mozilla projects.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.009 |
| Meta-epidemiology (narrow) | 0.002 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".