Entity and relation extractions for threat intelligence knowledge graphs
Извлечение сущностей и отношений для графов знаний о киберугрозах
2024-09-19
SCID: 54.1/zhh6fp4w
Discuss with AI
SecureBERTattention-based BiLSTMcyber threat intelligenceentity and relation extractionthreat intelligence knowledge graphs
Figures from the paper
Abstract (AI)
Advanced persistent threats (APTs) represent a complex challenge in cybersecurity as they infiltrate networks stealthily to conduct espionage, steal data, and maintain a long-term presence. To combat these threats, security professionals increasingly depend on cyber knowledge graphs (CKGs), which provide scalable solutions to analyze and structure vast amounts of cyber threat intelligence (CTI) from diverse sources in real time. This aids in developing and automating proactive security measures. Developing CKGs requires extracting entity and their relationships from unstructured CTI reports. However, existing approaches face significant limitations, such as difficulties with the nuances of cybersecurity language and high rates of error propagation, resulting in low accuracy and poor generalizability.This paper introduces a comprehensive Threat Intelligence Knowledge Graph (TiKG) framework designed to overcome these existing shortcomings. It leverages a domain-specific transformer-based model, SecureBERT, optimized for cybersecurity contexts. The TiKG framework features an advanced extraction pipeline that integrates SecureBERT embeddings with an attention-based BiLSTM, effectively capturing the context and nuances of security texts. This reduces the likelihood of error propagation and improves the model's extraction accuracy. Additionally, we incorporate a domain-specific ontology within our pipeline, coupled with an inference model, to ensure precise relation mapping and achieve high accuracy in relation extraction. Leveraging a scalable TI open-source dataset (DNRTI) and a curated CTI dataset, we demonstrate through extensive evaluations and analysis the effectiveness of our framework, showing significant improvements over existing methods in detecting and linking cyber threats. The contributions of this study offer a robust platform for security professionals to analyze and predict potential attacks, develop effective defenses, and enhance the strategic capabilities of cybersecurity operations.
Key Findings
1
A domain-specific ontology and inference model support precise relation mapping in threat intelligence relation extraction.
2
Evaluations on DNRTI, STUCCO, CYNER, and a curated CTI dataset show significant improvements over existing methods in detecting and linking cyber threats.
3
The framework is intended to strengthen automated threat analysis, attack prediction, defensive planning, and cybersecurity operations.
4
The paper introduces a Threat Intelligence Knowledge Graph pipeline for extracting entities and relationships from unstructured cyber threat intelligence reports.
5
The pipeline combines the cybersecurity-specific SecureBERT model with an attention-based BiLSTM to capture domain context and reduce error propagation during extraction.
Research Object
unstructured cyber threat intelligence reports concerning advanced persistent threats
Research Subject
accurate extraction and semantic linking of cybersecurity entities and their relationships for threat intelligence knowledge graphs
Publication Details
Publication Date
2024-09-19
Journal
Publisher
ISSN
Cited by
46
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest