Neural Network Models for Paraphrase Identification, Semantic Textual Similarity, Natural Language Inference, and Question Answering
Neural Network Models for Paraphrase Identification, Semantic Textual Similarity, Natural Language Inference, and Question Answering
复制标题
DOI：
--
复制
发表时间：
2018-06
期刊：
影响因子：
--
通讯作者：
Wuwei Lan;Wei Xu
中科院分区：
文献类型：
--
作者：
Wuwei Lan;Wei Xu
在本文中，我们分析了几种用于句子对建模的神经网络设计(及其变体)，并在八个数据集上广泛比较了它们的性能，包括释义识别、语义文本相似性、自然语言推理和问答任务。虽然这些模型中的大多数都声称具有最先进的性能，但最初的论文往往只报道一两个选定的数据集。我们提供了一个系统的研究，并表明：(I)通过LSTM编码上下文信息和句子间的交互是关键的，(Ii)Tree-LSTM没有像之前声称的那样有帮助，但在Twitter数据集上的性能得到了令人惊讶的改善，(Iii)到目前为止，增强的顺序推理模型对于较大的数据集是最好的，而成对单词交互模型在数据较少的情况下获得了最佳的性能。我们以开源工具包的形式发布我们的实现。
In this paper, we analyze several neural network designs (and their variations) for sentence pair modeling and compare their performance extensively across eight datasets, including paraphrase identification, semantic textual similarity, natural language inference, and question answering tasks. Although most of these models have claimed state-of-the-art performance, the original papers often reported on only one or two selected datasets. We provide a systematic study and show that (i) encoding contextual information by LSTM and inter-sentence interactions are critical, (ii) Tree-LSTM does not help as much as previously claimed but surprisingly improves performance on Twitter datasets, (iii) the Enhanced Sequential Inference Model is the best so far for larger datasets, while the Pairwise Word Interaction Model achieves the best performance when less data is available. We release our implementations as an open-source toolkit.