BIOSSES (Biomedical Semantic Similarity Estimation System)

BIOSSES 数据集包含总共 100 个句子对，所有句子对均选自“TAC2 Biomedical Summarization Track Training Data Set”。句子对由五位不同的人类专家评估，他们判断它们的相似性并给出 [0-4] 范围内的分数。我们的指南是根据 SemEval 2012 任务 6 指南编写的。The BIOSSES dataset contains a total of 100 sentence pairs, all selected from the "TAC2 Biomedical Summarization Track Training Data Set". These sentence pairs were evaluated by five distinct human experts, who judged their similarity and assigned scores within the range of [0, 4]. Our evaluation guidelines were developed based on the SemEval 2012 Task 6 guidelines.

搜集汇总

背景与挑战

背景概述

BIOSSES是一个用于生物医学领域语义相似度估计的数据集，包含100个选自TAC2数据集的句子对，每个句子对由五位专家根据SemEval 2012指南标注0-4分的相似度分数。