BenchIEFL
收藏资源简介:
BenchIEFL是由蒙特利尔大学的RALI实验室创建的一个用于开放信息抽取(OIE)的新基准数据集。该数据集通过重新注释BenchIE数据集,修正了常见错误和不一致性,提高了注释的精确性和相关性。BenchIEFL包含新的匹配函数,能更灵活地捕捉有效抽取,从而产生更公平的系统评估排名。数据集主要用于评估OIE系统在下游任务中的性能,如问答和文本理解,旨在通过提供更高质量的基准来推动OIE技术的发展。
BenchIEFL is a novel benchmark dataset for Open Information Extraction (OIE), developed by the RALI Laboratory at the University of Montreal. It is constructed by re-annotating the original BenchIE dataset, correcting common errors and inconsistencies, and improving the accuracy and relevance of annotations. BenchIEFL includes new matching functions that can flexibly capture valid extractions, thereby generating more equitable system evaluation rankings. The dataset is primarily used to evaluate the performance of OIE systems in downstream tasks such as question answering and text understanding, aiming to advance the development of OIE technology by providing higher-quality benchmarks.




