A Graph-Based Neural Model for End-to-End Frame Semantic Parsing

Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 3864–3874 November 7–11, 2021. c⃝2021 Association for Computational Linguistics 3864 A Graph-Based Neural Model for End-to-End Frame Semantic Parsing Zhichao Lin1, Yueheng Sun2, Meishan Zhang 1∗ 1School of New Media and Communication, Tianjin University, China 2College of Intelligence and Computing, Tianjin University, China {chaosmyth,yhs,zhangmeishan}@tju.edu.cn Abstract Frame semantic parsing is a semantic anal- ysis task based on FrameNet which has re- ceived great attention recently. The task usu- ally involves three subtasks sequentially: (1) target identification, (2) frame classification and (3) semantic role labeling. The three sub- tasks are closely related while previous studies model them individually, which ignores their intern connections and meanwhile induces er- ror propagation problem. In this work, we pro- pose an end-to-end neural model to tackle the task jointly. Concretely, we exploit a graph- based method, regarding frame semantic pars- ing as a graph construction problem. All predi- cates and roles are treated as graph nodes, and their relations are taken as graph edges. Ex- periment results on two benchmark datasets of frame semantic parsing show that our method is highly competitive, resulting in better perfor- mance than pipeline models. 1 Introduction Frame semantic parsing (Gildea and Jurafsky, 2002) aims to analyze all sentential predicates as well as their FrameNet roles as a whole, which has received great interest recently. This task can be helpful for a number of tasks, including informa- tion extraction (Surdeanu et al., 2003), question answering (Shen and Lapata, 2007), machine trans- lation (Liu and Gildea, 2010) and others (Coyne et al., 2012; Chen et al., 2013; Agarwal et al., 2014). Figure 1 shows an example, where all predicates as well as their semantic frame and roles in the sentence are depicted. Previous studies (Das et al., 2014; Swayamdipta et al., 2017; Bastianelli et al., 2020) usually divide the task into three subtasks, including target iden- tification, frame classification and semantic role labeling (SRL), respectively. By performing the three subtasks sequentially, the whole frame seman- tic parsing can be accomplished. The majority of ∗Corresponding author. There have also been some power outages due to power lines being damaged Existence Entity Causation Effect Cause Damaging Patient Figure 1: An example involving frame semantic struc- tures, taken from the FrameNet (Baker et al., 1998). Frame-evoking predicates are highlighted in the sen- tence, and corresponding frames are shown in colored blocks below. The frame-specific roles are underlined with their frames in the same row. works focus on either one or two of the three sub- tasks, treating them separately (Yang and Mitchell, 2017; Botschen et al., 2018; Swayamdipta et al., 2018; Peng et al., 2018). The above formalization has two weaknesses. First, the individual modeling of the three subtasks is inefficient to utilize the relationship among them. Apparently, the earlier subtasks can not exploit the information from their future subtasks. Second, the pipeline strategy can suffer from the error propa- gation problem, where the errors occurring in the previous subtasks can influence …

Content truncated to 500 words