MITRE RAGAS: Cybersecurity Ingestion Corpus and QA Evaluation Pairs

Cybersecurity document corpus and question-answer pairs designed for evaluating Retrieval-Augmented Generation (RAG) systems. Developed by Manoel Malon Costa de Moura and hosted on Harvard Dataverse, the collection supports the RRAG search methodology as of March 2026. It contains two distinct subsets: an ingestion dataset for document indexing and an evaluation dataset for performance testing.

Use Cases

• Benchmarking RAG performance using the provided question-answer pairs
• Testing document retrieval accuracy against the cybersecurity ingestion corpus
• Evaluating the RRAG search approach for domain-specific information retrieval

Strengths

• Includes both ingestion and evaluation components for end-to-end RAG testing
• Domain-specific focus on cybersecurity documentation
• Designed specifically to support the RRAG search methodology

Limitations

• Unknown record count and total file size
• Lack of detailed schema or column-level metadata
• Potential narrow focus on MITRE-related cybersecurity content

Provenance

Source

Costa de Moura, Manoel Malon via Harvard Dataverse

Collection Method

Curated

Freshness

Last updated March 2026.