arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2302.12530cs.CL
收藏

感知细微冲突的语义匹配双路径建模

Dual Path Modeling for Semantic Matching by Perceiving Subtle Conflicts

School of Software, Beihang University(北京航空航天大学软件学院);Centre for Natural Language Processing, Meituan Inc.(美团自然语言处理中心)

Chao Xue, Di Liang, Sirui Wang, Wei Wu, Jing Zhang

摘要:

基于Transformer的预训练模型在语义匹配方面取得了很大的改进。然而,现有的模型仍然存在捕捉细微差异的能力不足的问题。句对中词语的修改、增删可能会使模型难以预测它们之间的关系。为了缓解这个问题,我们提出了一种新的双路径建模框架,以提高模型的能力,感知句子对的细微差异,分别建模亲和力和差异语义。在双路径建模框架的基础上,设计了双路径建模网络(DPM-Net)来识别语义关系。我们在10个语义匹配和鲁棒性测试数据集上进行了大量的实验,实验结果表明,我们提出的方法在基线上取得了一致的改进。

英文摘要:

Transformer-based pre-trained models have achieved great improvements in semantic matching. However, existing models still suffer from insufficient ability to capture subtle differences. The modification, addition and deletion of words in sentence pairs may make it difficult for the model to predict their relationship. To alleviate this problem, we propose a novel Dual Path Modeling Framework to enhance the model's ability to perceive subtle differences in sentence pairs by separately modeling affinity and difference semantics. Based on dual-path modeling framework we design the Dual Path Modeling Network (DPM-Net) to recognize semantic relations. And we conduct extensive experiments on 10 well-studied semantic matching and robustness test datasets, and the experimental results show that our proposed method achieves consistent improvements over baselines.

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。

↑