Original title: Construction and Natural Language Querying of a Cybersecurity Knowledge Graph
Ines Ben Brahim, Mohamed-Amine El Mortaji, Nada Haddad, Sami Rezig +3 · September 2026
Cybersecurity vulnerability data is scattered across many platforms, making it hard for experts to get a clear picture of security threats. This delay can hurt risk assessment and decision-making. Knowledge graphs help link and organize this information, but exploring them usually requires learning complex coding languages like Cypher. To solve this, researchers built a cybersecurity knowledge graph using information pulled from the National Vulnerability Database through its REST API. They mapped out connections between security flaws, software products, vendors, severity ratings, weaknesses, and references. Then, they added an artificial intelligence interface that translates plain, everyday English questions directly into graph code. The researchers found that natural language querying significantly lowers the barrier to using these security graphs. This makes exploring vulnerability data far more intuitive and practical for a wider group of cybersecurity users.
Imagine you have a huge filing cabinet with folders for every software product, but the folders are scattered across different rooms. To find anything, you need to know the exact room and drawer number. Here, the filing cabinet is the collection of vulnerability data, and the rooms are different databases. In this paper, they organize everything into one cabinet with a clear label system, and then add a search box where you can just type what you need in plain English, so you don't have to memorize the filing codes.
Cybersecurity professionals often waste time piecing together vulnerability data from multiple sources. This work makes that data accessible through plain English questions, so anyone—not just database experts—can quickly find the information they need for risk assessment and decision-making.
The work focuses on NVD data only, so it may not cover vulnerabilities from other sources. The abstract does not report quantitative results on query accuracy or system performance, so the practical effectiveness remains to be validated.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Construction and Natural Language Querying of a Cybersecurity Knowledge Graph — Ines Ben Brahim, Mohamed-Amine El Mortaji, Nada Haddad et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.