从文件中提取信息：实际场景中的问答与记号分类（英文版）

英文标题：Information Extraction from Documents: Question Answering vs Token Classification in real-world setups中文摘要：通过与传统标记分类方法的比较，我们发现当处理较短、干净的实体时，最好使用基于标记分类的方法，而 QA 方法可以成为嘈杂环境或长实体用例的良好替代方案。英文摘要：Research in Document Intelligence and especially in Document Key InformationExtraction (DocKIE) has been mainly solved as Token Classification problem.Recent breakthroughs in both natural language processing (NLP) and computervision helped building document-focused pre-training methods, leveraging amultimo