Implicit Discourse Relation Detection via a Deep Architecture with Gated Relevance Network

Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, pages 1726–1735, Berlin, Germany, August 7-12, 2016. c⃝2016 Association for Computational Linguistics Implicit Discourse Relation Detection via a Deep Architecture with Gated Relevance Network Jifan Chen, Qi Zhang, Pengfei Liu, Xipeng Qiu, Xuanjing Huang Shanghai Key Laboratory of Data Science School of Computer Science, Fudan University 825 Zhangheng Road, Shanghai, China jfchen14,qz,pfliu14,xpqiu,xjhuang@fudan.edu.cn Abstract Word pairs, which are one of the most eas- ily accessible features between two text segments, have been proven to be very useful for detecting the discourse relations held between text segments. However, be- cause of the data sparsity problem, the performance achieved by using word pair features is limited. In this paper, in or- der to overcome the data sparsity problem, we propose the use of word embeddings to replace the original words. Moreover, we adopt a gated relevance network to capture the semantic interaction between word pairs, and then aggregate those se- mantic interactions using a pooling layer to select the most informative interactions. Experimental results on Penn Discourse Tree Bank show that the proposed method without using manually designed features can achieve better performance on recog- nizing the discourse level relations in all of the relations. 1 Introduction In a well-written document, no unit of the text is completely isolated, discourse relations describe how two units (e.g. clauses, sentences, and larger multi-clause groupings) of discourse are logically connected. Many downstream NLP applications such as opinion mining, summarization, and event detection, can benefit from those relations. The task of automatically identify discourse re- lation is relatively simple when explicit connec- tives such as however and because are given (Pitler et al., 2009). However, the identification becomes much more challenging when such connectives are missing. In fact, such implicit discourse relations outnumber explicit relations in naturally occurring text, and identify those relations have been shown to be the performance bottleneck of an end-to-end discourse parser (Lin et al., 2014). Most of the existing researches used rich lin- guistic features and supervised learning methods to achieve the task (Soricut and Marcu, 2003; Pitler et al., 2009; Rutherford and Xue, 2014). Among their works, word pairs are heavily used as an important feature, since word pairs like (warm,cold) might directly trigger a contrast rela- tion. However, because of the data sparsity prob- lem (McKeown and Biran, 2013) and the lack of metrics to measure the semantic relation between those pairs, which is so-called the semantic gap problem (Zhao and Grosky, 2002), the classifiers based on word pairs in the previous studies did not work well. Moreover, some text segment pairs are more complicated, it is hard to determine the re- lation held between them using only word pairs. Consider the following sentence pair with a casual relation as an example: S1: Psyllium’s not a good crop. S2: You get a rain at the wrong time and the crop is ruined. Intuitively, (good, wrong) and (good, ruined), seem to be the most informative word pairs, and it …

Content truncated to 500 words