Multifaceted LLMs for Recommendations of Software Vulnerability Repair and Assessment
Bibliographic record
Abstract
Common Weakness Enumerations (CWEs) and Common Vulnerabilities and Exposures (CVEs) are open knowledge bases that provide definitions, descriptions, and samples of code vulnerabilities. The combination of Large Language Models (LLMs) with vulnerability knowledge bases helps to enhance and automate code vulnerability repair. Meanwhile, several key factors are introduced, including 1) the retrieval of the most relevant context to a specific vulnerable code snippet; 2) augmenting the retrieved context as input to the LLMs; and 3) the form of the generated artifacts, such as code repair with natural language explanations or code repair only. Artifacts produced by these factors often lack transparency and explainability regarding the rationale behind the repair. In this paper, we propose a framework for an explainable recommendation of vulnerable code repairs with techniques addressing each factor. We design embedding and retrieval patterns based on the data characteristics of the CWE and CVE codebase and knowledge base. These patterns applied across multiple LLMs produce a set of recommendations for the same vulnerable code snippets. These recommendations are used to generate code repairs. This highlights that assessments of these recommendations become imperative in deciding the appropriate code repair and the rationale behind the repair. Through 100 experiments, we observe the inadequacy of the SOTA metrics to differentiate between lowquality retrieval and irrelevant repairs. We use the LLM-as-a-Judge framework to enhance the robustness of recommendation assessments. Compared with baselines using static code analysis and LLMs without retrieval from the CWE and CVE knowledge base, our findings have highlighted the potential that multifaceted LLM usage produces explainable and reliable recommendations under a small to mild level of self-alignment bias. Our work has been developed on open-source knowledge bases and models, which makes it reproducible and extensible to new datasets and retrieval patterns.
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 | 0.006 | 0.057 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.012 | 0.004 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.003 | 0.005 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.004 | 0.003 |
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".