HistoryBankQA: Multilingual Temporal Question Answering on Historical Events

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Expert, medium

Summary

HistoryBankQA introduces a new multilingual database, HistoryBank, containing over 10 million historical events from Wikipedia timelines and infoboxes, covering 10 languages. This resource addresses limitations in existing benchmarks for temporal reasoning in large language models (LLMs), which often lack multilingual coverage and historical depth. The project also presents a comprehensive benchmark with 6 temporal question answering (QA) tasks across all supported languages. Evaluations of models like LLaMA-3-8B, Mistral-7B, Gemma-2-9B, Qwen3-8B, and GPT4o show GPT-4o consistently performs best, while Gemma-2 leads among smaller models. This work provides a rich, publicly available resource for advancing multilingual, temporally-aware language understanding of historical events.

Key takeaway

For NLP engineers and AI scientists developing or evaluating LLMs for temporal reasoning, HistoryBankQA offers a critical new resource. You should integrate this comprehensive, multilingual benchmark to rigorously test your models' understanding of historical events across 10 languages. This allows for more robust evaluation beyond recent or English-centric data, revealing performance gaps and guiding improvements in temporal NLP capabilities.

Key insights

HistoryBankQA provides a large-scale, multilingual benchmark and database for temporal question answering on historical events.

Principles

• Temporal reasoning is crucial for diverse NLP tasks.
• Existing LLM benchmarks for temporal reasoning are limited in scope.
• Multilingual and historically deep datasets are essential for robust evaluation.

Method

HistoryBank was built by extracting 10M+ historical events from Wikipedia timelines and infoboxes across 10 languages. A benchmark was created covering 6 temporal QA tasks.

In practice

• Evaluate LLMs on historical temporal reasoning using the HistoryBankQA benchmark.
• Utilize the HistoryBank database for multilingual NLP research.
• Access public code and datasets for further development.

Topics

• Temporal Question Answering
• Multilingual NLP
• Historical Events
• LLM Benchmarking
• HistoryBank Database
• Wikipedia Data

Code references

• mandalbiswadip/history-bank

Best for: Research Scientist, AI Scientist, NLP Engineer

Related on AIssential

• HistoriQA-ThirdRepublic: Multi-Hop Question Answering Corpus for Historical Research, Parliamentary Debates from the French Third Republic (1870-1940) — HistoriQA-ThirdRepublic bridges NLP benchmarks and historical scholarship through complex multi-hop QA.
• IndicDB -- Benchmarking Multilingual Text-to-SQL Capabilities in Indian Languages — IndicDB benchmarks multilingual Text-to-SQL for Indic languages, revealing a significant performance gap compared to English.
• Data Augmentation for Historical NER: A Systematic Comparison of Lexical and LLM-based Approaches — This paper systematically compares lexical and LLM-based data augmentation for Historical Named Entity Recognition.
• Causal_QA.PT: A Human–LLM Co-Curated Benchmark for Causal Question Answering in Portuguese Language — A new human-LLM co-curated benchmark improves causal reasoning evaluation for Portuguese LLMs.
• SemEval-2026 Task 7: Everyday Knowledge Across Diverse Languages and Cultures — Evaluating LLM adaptability across diverse, low-resource language-culture pairs reveals critical performance and misalignment issues.
• EHRNote-ChatQA: A Benchmark for Evidence-Grounded Multi-Turn Clinical Question Answering over Longitudinal Discharge Summaries — EHRNote-ChatQA benchmarks LLMs on evidence-grounded, multi-turn clinical QA over longitudinal discharge summaries, revealing current limitations.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.