Cybersecurity QAA: 1,563 Question-Answer-Assertion Triples for UK SMEs

1,563 question-answer-assertion triples evaluate large language models on cybersecurity advice for UK small and medium-sized enterprises. The dataset covers topics like network security, data protection, and user access management. It was created by Rowden and last updated on the Hugging Face platform in December 2024.

Use Cases

• Evaluate LLM accuracy in cybersecurity advice based on question-answer-assertion triples.
• Benchmark model performance on SME-specific security topics like data protection.
• Fine-tune models for cybersecurity information retrieval using the provided QAA structure.

Strengths

• Contains 1,563 structured question-answer-assertion triples.
• Focuses on cybersecurity topics relevant to UK small and medium-sized enterprises.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is known, but the specific topics and assertion details require inspection of the full dataset.
• Last updated 2024-12-16 17:57:01; freshness should be verified.

Provenance

Source

Rowden on Hugging Face.

Collection Method

Likely curated for evaluating LLMs in cybersecurity.

Freshness

Last updated 2024-12-16 17:57:01.

Geography

Primarily United Kingdom (UK SMEs).