Tempest-OSINT: Cybersecurity Documents and QA Pairs for RAG Evaluation

TEMPEST-OSINT provides a collection of cybersecurity documents and paired question-answer sets designed for Retrieval-Augmented Generation (RAG) evaluation. Created by Costa de Moura and Manoel Malon and hosted on Harvard Dataverse, the data supports the RRAG approach for searching technical security documentation as of March 2026.

Use Cases

• Benchmarking RAG system performance using the provided question-answer pairs
• Fine-tuning embedding models on the cybersecurity ingestion dataset
• Evaluating document retrieval accuracy for technical security queries

Strengths

• Includes both ingestion and evaluation subsets for end-to-end RAG benchmarking
• Domain-specific focus on cybersecurity documentation
• Provides explicit question-answer pairs for model validation

Limitations

• Total record count and document volume are not specified in the metadata
• File formats and data encoding standards are undocumented
• Lack of explicit license information may restrict usage rights

Provenance

Source

Harvard Dataverse

Collection Method

Curated for the RRAG research approach

Freshness

Last updated March 2026.