Context Tracking Network: Graph-based Context Modeling for Implicit Discourse Relation Recognition

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 1592–1599 June 6–11, 2021. ©2021 Association for Computational Linguistics 1592 Context Tracking Network: Graph-based Context Modeling for Implicit Discourse Relation Recognition Yingxue Zhang1, Fandong Meng1, Peng Li1, Ping Jian2, Jie Zhou1 1Pattern Recognition Center, WeChat AI, Tencent Inc, China 2Beijing Institute of Technology, China {yxuezhang, fandongmeng, patrickpli, withtomzhou}@tencent.com pjian@bit.edu.cn Abstract Implicit discourse relation recognition (IDRR) aims to identify logical relations between two adjacent sentences in the discourse. Existing models fail to fully utilize the contextual in- formation which plays an important role in in- terpreting each local sentence. In this paper, we thus propose a novel graph-based Context Tracking Network (CT-Net) to model the dis- course context for IDRR. The CT-Net firstly converts the discourse into the paragraph as- sociation graph (PAG), where each sentence tracks their closely related context from the intricate discourse through different types of edges. Then, the CT-Net extracts contextual representation from the PAG through a spe- cially designed cross-grained updating mech- anism, which can effectively integrate both sentence-level and token-level contextual se- mantics. Experiments on PDTB 2.0 show that the CT-Net gains better performance than mod- els that roughly model the context. 1 Introduction Implicit discourse relation recognition (IDRR) aims to identify logical relations between two ad- jacent sentences in discourse without the guidance of connectives (e.g., because, but), which is one of the major challenges in discourse parsing. With the rise of deep learning, lots of sentence-modeling based methods (Liu and Li, 2016; Rönnqvist et al., 2017; Bai and Zhao, 2018; Xu et al., 2019; Shi and Demberg, 2019) have emerged in the field of IDRR. These methods typically focus on modeling the local semantics of these two sentences, without considering wider discourse context. Contextual information plays an important role in understanding sentences. Take the paragraph P = {S1, S2, S3, S4} in Figure 1 as an example, the ground-truth relation between S3 and S4 is “Comparison”. Combining the contextual infor- mation carried by S1 and S2, we can more eas- ily identify the “Comparison” relation reflected by The manufacturer went public at $15.75 a share in August 1987. Mr. Sim’s goal then was a $29 per-share price. Strong earnings growth helped achieve that price far ahead of schedule. The stock has since softened, trading around $25 a share !": !$: !%: !&: !" !$ !% !& Figure 1: The paragraph association graph (PAG) (right) built for a case (left) of PDTB 2.0. “achieve that price” (rising: “$15.75 a share” to “$29 per-share”) and “softened” (falling: “$29 per- share” to “$25 a share”). Dai and Huang (2018) move one step on utilizing wider discourse context, where they use a hierarchical BiLSTM (H-LSTM) to model the whole paragraph rather than only the two sentences, to obtain context-aware sentence representation. However, there are still two limita- tions in their model. First, they roughly merge all the information in the paragraph, which dilutes the role of key …

Content truncated to 500 words