arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2002.07767cs.CL
收藏

基于语义相似度的学习方法提高文摘质量

Learning by Semantic Similarity Makes Abstractive Summarization Better

Korea University(高丽大学);NAVER(NAVER(纳瓦))

Wonjin Yoon, Yoon Sun Yeo, Minbyul Jeong, Bong-Jun Yi, Jaewoo Kang

摘要:

摘要模型利用预先训练好的语言模型,近年来得到了快速的发展。然而,这些模型主要是由自动评估指标,如ROUGE评估。虽然ROUGE以与人类评价分数呈正相关而闻名,但它因其脆弱性和实际质量之间的差距而受到批评。在本文中,我们比较了最近的LM,BART和基准数据集,CNN/DM,使用众包人类评价指标的参考摘要生成的摘要。有趣的是,模型生成的摘要相对于参考摘要获得更高的分数。从我们的实验结果出发,我们首先讨论了CNN/DM数据集的内在特征,预训练语言模型的进展,以及它们在训练数据上的泛化能力。最后,我们分享了我们对模型生成摘要的见解,并提出了我们对抽象摘要学习方法的想法。

英文摘要:

By harnessing pre-trained language models, summarization models had rapid progress recently. However, the models are mainly assessed by automatic evaluation metrics such as ROUGE. Although ROUGE is known for having a positive correlation with human evaluation scores, it has been criticized for its vulnerability and the gap between actual qualities. In this paper, we compare the generated summaries from recent LM, BART, and the reference summaries from a benchmark dataset, CNN/DM, using a crowd-sourced human evaluation metric. Interestingly, model-generated summaries receive higher scores relative to reference summaries. Stemming from our experimental results, we first argue the intrinsic characteristics of the CNN/DM dataset, the progress of pre-trained language models, and their ability to generalize on the training data. Finally, we share our insights into the model-generated summaries and presents our thought on learning methods for abstractive summarization.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑