T2Ranking

T2Ranking是由清华大学和腾讯公司合作开发的大型中文段落排序数据集，旨在解决中文信息检索领域中数据集规模小、相关性标注粗糙和假阴性问题。该数据集包含超过30万查询和200多万个独特段落，来源于真实搜索引擎。数据集通过专家标注者提供4级细粒度相关性评分，以提高评估的准确性。T2Ranking不仅支持文本查询和段落数据，还提供了查询类型和文档的XML文件等辅助资源，以促进进一步研究。数据集的应用领域包括问答和阅读理解，旨在通过提高段落排序的准确性，增强信息检索系统的性能。T2Ranking is a large-scale Chinese paragraph ranking dataset co-developed by Tsinghua University and Tencent. It is designed to address the issues of small dataset scale, coarse relevance annotations and false negatives in the field of Chinese information retrieval. The dataset contains over 300,000 queries and more than 2 million unique paragraphs sourced from real-world search engines. Expert annotators provide 4-level fine-grained relevance scores for the dataset to improve the accuracy of evaluation. Apart from supporting text queries and paragraph data, T2Ranking also offers auxiliary resources such as XML files of query types and documents to facilitate further research. Its application scenarios include question answering and reading comprehension, aiming to enhance the performance of information retrieval systems by improving the accuracy of paragraph ranking.

搜集汇总

背景与挑战

背景概述

T2Ranking是由清华大学和腾讯公司合作开发的大型中文段落排序数据集，包含超过30万查询和200多万个独特段落，并提供了4级细粒度相关性评分，旨在解决中文信息检索中数据集规模小、标注粗糙和假阴性问题。