license: cc-by-nc-nd-4.0
task_categories:
- question-answering
language:
- zh
tags:
- art
size_categories:
- 1K<n<10K
pretty_name: A Comprehensive Benchmark for Chinese Long Historical Document Understanding
configs:
- config_name: default
data_files:
- split: test
path: LongHisDoc_IMG/**/*.jpg
LongHisDoc
📖 Introduction
In this repo, we present LongHisDoc, a pioneering benchmark specifically designed to evaluate the capabilities of LLMs and LVLMs in long-context historical document understanding tasks. This benchmark includes 101 historical documents across 10 categories, with 1,012 expert-annotated question-answer pairs covering four types, and the evidence for the questions is drawn from three modalities. Here, we provide all page screenshots of 101 PDF-formatted historical documents (./LongHisDoc_IMG), their OCR results (./OCR_res), and QA pairs (./LongHisDoc.json).
- The Q&A pairs are provided in json format (LongHisDoc.json) and contain the following attributes:
{
"doc_id": "9242",
"question": "成公三年，诸侯伐郑，军队驻扎在哪里？",
"answer": "伯牛",
"evidence_pages": "[3]",
"type": "Single Page",
"answer_type": "Str",
"evidence_modal": "layout",
"input_pages": "[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30]",
"item_id": "9242_1",
"book_name": "《01/24音註全文春秋括例始末左伝句読直解著者林堯叟（宋）／訓点者松永昌易[数量]25冊[書誌事項]刊本,寛文01年[旧蔵者]昌平坂学問所/M2015073110264263265_0001》",
"guji_type_zh": "史藏",
"guji_type_en": "History"
}
🔎 Evaluate
The data is hosted in this Hugging repo. See LongHisDoc Eval Toolkit for evaluation using the corresponding evaluation toolkit.
📜 License
The data should be used and distributed under (CC BY-NC-ND 4.0) for non-commercial research purposes.