BriefGPT.xyz
Oct, 2016
阅读理解中端对端答案块提取和排序
End-to-End Reading Comprehension with Dynamic Answer Chunk Ranking
HTML
PDF
Yang Yu, Wei Zhang, Kazi Hasan, Mo Yu, Bing Xiang...
TL;DR
本文提出了一种名为DCR（dynamic chunk reader）的神经网络阅读理解模型，能够从给定文档中提取和排名一组答案候选项来回答问题，相较于之前的神经网络RC模型，DCR更加注重预测变长的答案，实验结果表明，DCR在SQuAD数据集上取得了最先进的精确匹配和F1分数。
Abstract
This paper proposes
dynamic chunk reader
(DCR), an end-to-end
neural reading comprehension
(RC) model that is able to extract and rank a set of
a
→