IslamicFaithQA: Bilingual Benchmark for Faithful Islamic Question Answering

IslamicFaithQA is a bilingual Arabic/English generative question answering benchmark designed for faithfulness-oriented evaluation. The dataset accompanies the paper 'From RAG to Agentic RAG for Faithful Islamic Question Answering' and was authored by QCRI. It was last updated on 2026-02-25.

Use Cases

• Evaluate faithfulness and correctness of generative AI models based on the benchmark's Islamic QA structure.
• Research hallucination reduction techniques using the dataset's bilingual question-answer pairs.
• Test retrieval-augmented generation (RAG) and agentic approaches for Islamic knowledge tasks.
• Benchmark model abstention behavior when faced with questions on Islamic topics.

Strengths

• Designed specifically for faithfulness-oriented evaluation in Islamic QA.
• Bilingual content in Arabic and English, supporting cross-lingual research.
• Accompanies a peer-reviewed paper (arxiv.org/abs/2601.07528), providing research context.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and exact size are unknown, which may limit suitability assessment.

Provenance

Source

QCRI

Collection Method

Likely constructed as a benchmark for research, as described in the accompanying paper.

Time Range

null

Freshness

Last updated 2026-02-25 08:59:21; freshness should be verified.

Geography

null