arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2207.04660cs.CLcs.IR
收藏

SummScore:一种基于交叉编码的摘要质量综合评价指标

SummScore: A Comprehensive Evaluation Metric for Summary Quality Based on Cross-Encoder

National University of Defense Technology(国防科技大学);PLA Academy of Military Science(中国人民解放军军事科学院)

Wuhang Lin, Shasha Li, Chen Zhang, Bin Ji, Jie Yu, Jun Ma, Zibo Yi

摘要:

文本摘要模型通常经过训练以生成满足人类质量要求的摘要。然而,现有的摘要文本的评价指标只是粗略的摘要质量代理,遭受低相关性与人类评分和抑制摘要的多样性。为了解决这些问题,我们提出了SummScore,一个综合的度量摘要质量评估的基础上交叉编码器。首先,SummScore采用原文-摘要的测量模式,通过比较原文的语义,摆脱摘要多样性的抑制。在文本匹配预训练交叉编码器的帮助下,SummScore可以有效地捕捉摘要语义之间的细微差异。其次,为了提高全面性和可解释性,SummScore由四个细粒度子模型组成,分别衡量连贯性、一致性、流畅性和相关性。我们使用半监督的多轮训练来提高我们的模型在极其有限的注释数据上的性能。大量的实验表明,SummScore在上述四个维度上与人类评分的相关性明显优于现有的评估指标。本文还提供了SummScore对16种主流摘要模型的质量评价结果,以供后续研究参考。

英文摘要:

Text summarization models are often trained to produce summaries that meet human quality requirements. However, the existing evaluation metrics for summary text are only rough proxies for summary quality, suffering from low correlation with human scoring and inhibition of summary diversity. To solve these problems, we propose SummScore, a comprehensive metric for summary quality evaluation based on CrossEncoder. Firstly, by adopting the original-summary measurement mode and comparing the semantics of the original text, SummScore gets rid of the inhibition of summary diversity. With the help of the text-matching pre-training Cross-Encoder, SummScore can effectively capture the subtle differences between the semantics of summaries. Secondly, to improve the comprehensiveness and interpretability, SummScore consists of four fine-grained submodels, which measure Coherence, Consistency, Fluency, and Relevance separately. We use semi-supervised multi-rounds of training to improve the performance of our model on extremely limited annotated data. Extensive experiments show that SummScore significantly outperforms existing evaluation metrics in the above four dimensions in correlation with human scoring. We also provide the quality evaluation results of SummScore on 16 mainstream summarization models for later research.

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。

↑