Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

202 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MulitaMiner logo

Vulnerability Extraction from Security Reports using LLMs


Artifact Index

⚠️ This branch (main) is not an evaluation artifact. It is an index. The repository hosts the artifacts of two different papers, each one on its own branch.

Paper Artifact branch Frozen tag
SBSeg 2026 · Main Track
MulitaMiner: A Multi-Version Evaluation of LLM-Based Vulnerability Report Extraction
Pipeline progression V1 → V2 → V3, evaluated with cloud LLMs across 450 runs.
V3 sbseg2026-artifact
WTICG 2026
On-Premise vs. Cloud: Local LLMs for Vulnerability Extraction from Security Scanner Reports
Nine local models (4B to 21B) compared against a DeepSeek cloud reference.
slms wticg2026-artifact

Both papers deal with LLM-based vulnerability extraction, but they are distinct works: the SBSeg one measures how far pipeline engineering carries extraction quality using cloud models, while the WTICG one measures the quality cost of running local models to keep scanner reports confidential. Check the title before starting the review.

Every artifact branch carries its own complete README.md, with considered badges, basic information, dependencies, installation, minimum test and the claims of its paper.

Branch or tag?

The branch always holds the latest state of an artifact. The tag is frozen: it points at the exact commit submitted for evaluation and never moves, so the code under review cannot shift while you are reviewing it. Either link works, and both render the same README.

Previous material

Content Reference
Artifact of the previous version of the tool (dataset of 6,700 vulnerabilities extracted from 129 OpenVAS reports), archived on Zenodo v2-dataset-artifact

That tag preserves the state of this branch from before it became an index, so already published references remain valid.

About the tool

MulitaMiner is an automated tool for extracting and structuring vulnerabilities from heterogeneous PDF reports produced by security scanners (OpenVAS, Tenable WAS). Its LLM-based pipeline combines scanner-aware adaptive segmentation and specialized prompting to turn unstructured findings into consistent, analysis-ready records, with standardized outputs and quality validation.

LICENSE

Distributed under the MIT License. See LICENSE.

About

MulitaMiner: a tool developed to extract and process vulnerabilities from security PDF reports using Large Language Models (LLMs).

Resources

Stars

4 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages