论文泛读136使用图形进行文本摘要的提取方法

Posted 2021-06-29 及时行樂_

贴一下汇总贴：论文阅读记录

论文链接：《Extractive approach for text summarisation using graphs》

一、摘要

自然语言处理是一门重要的学科，其目的是通过数字表示来理解文本，由于我们写作和说话的方式多种多样，通常不够准确。我们的论文探讨了可用于使用提取方法解决文本摘要问题的不同图相关算法。我们考虑两个指标：句子重叠和编辑距离来衡量句子相似度。

二、结论

Our extracted summaries are evaluated using the RecallOriented Understudy for Gisting Evaluation - ROGUE metric. We have incorporated three types of ROGUE evaluation metric. Rogue-1, the overlap of unigrams (each word) between extracted summary and ground truth summary provided in the dataset, Rogue-2, the overlap of bigrams in the respective relation, and Rogue-L which reports on the longest common subsequence statistics.

三、模型

模型：