JDocQA: Japanese Document Question Answering Dataset for Generative Language Models
This dataset addresses the challenge of document QA for Japanese language applications, but it is incremental as it adapts existing QA tasks to a new language and format.
The authors introduced JDocQA, a large-scale Japanese document question answering dataset with 5,504 PDF documents and 11,600 annotated QA instances, requiring both visual and textual information, and evaluated it with LLMs and multimodal models, finding that including unanswerable questions in fine-tuning helps reduce hallucination.
Document question answering is a task of question answering on given documents such as reports, slides, pamphlets, and websites, and it is a truly demanding task as paper and electronic forms of documents are so common in our society. This is known as a quite challenging task because it requires not only text understanding but also understanding of figures and tables, and hence visual question answering (VQA) methods are often examined in addition to textual approaches. We introduce Japanese Document Question Answering (JDocQA), a large-scale document-based QA dataset, essentially requiring both visual and textual information to answer questions, which comprises 5,504 documents in PDF format and annotated 11,600 question-and-answer instances in Japanese. Each QA instance includes references to the document pages and bounding boxes for the answer clues. We incorporate multiple categories of questions and unanswerable questions from the document for realistic question-answering applications. We empirically evaluate the effectiveness of our dataset with text-based large language models (LLMs) and multimodal models. Incorporating unanswerable questions in finetuning may contribute to harnessing the so-called hallucination generation.