SUBJQA

SUBJQA是一个基于客户评论的英语问答数据集，由哥本哈根大学计算机科学系等机构创建，包含超过10,000个跨6个领域的示例，涵盖产品和服务的评价。数据集通过最新的意见提取和矩阵分解技术构建，特别关注问题和答案中的主观性标注。SUBJQA旨在解决自然语言处理中主观性表达的问题，特别是在问答系统中的应用，帮助研究者开发能够理解和处理主观性内容的模型。

SUBJQA is an English question answering dataset based on customer reviews, created by institutions including the Department of Computer Science at the University of Copenhagen and others. It contains over 10,000 examples spanning six domains, covering product and service reviews. The dataset is constructed using state-of-the-art opinion extraction and matrix factorization techniques, with particular focus on subjective annotation in both questions and answers. SUBJQA aims to address the issue of subjective expressions in natural language processing, especially for applications in question answering systems, and helps researchers develop models capable of understanding and processing subjective content.