A Frame-based Sentence Representation for Machine Reading Comprehension

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 891–896 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 891 A Frame-based Sentence Representation for Machine Reading Comprehension Shaoru Guo1, Ru Li 1,2∗, Hongye Tan1,2, Xiaoli Li3, Yong Guan1, Hongyan Zhao1 and Yueping Zhang1 1. School of Computer and Information Technology, Shanxi University, Taiyuan, China 2. Key Laboratory of Computational Intelligence and Chinese Information Processing of Ministry of Education, Shanxi University, Taiyuan, China 3. Institute for Infocomm Research, A*Star, Singapore {guoshaoru0928,guanyong0130,13233587117}@163.com {liru,tanhongye}@sxu.edu.cn,xlli@ntu.edu.sg hongyanzhao@tyust.edu.cn Abstract Sentence representation (SR) is the most cru- cial and challenging task in Machine Reading Comprehension (MRC). MRC systems typi- cally only utilize the information contained in the sentence itself, while human beings can leverage their semantic knowledge. To bridge the gap, we proposed a novel Frame-based Sentence Representation (FSR) method, which employs frame semantic knowledge to facili- tate sentence modelling. Specifically, differen- t from existing methods that only model lexi- cal units (LUs), Frame Representation Models, which utilize both LUs in frame and Frame-to- Frame (F-to-F) relations, are designed to mod- el frames and sentences with attention schema. Our proposed FSR method is able to inte- grate multiple-frame semantic information to get much better sentence representations. Our extensive experimental results show that it per- forms better than state-of-the-art technologies on machine reading comprehension task. 1 Introduction Machine Reading Comprehension (MRC) requires machines to read and understand a text passage, and answer relevant questions about it. Human beings can easily understand the meaning of a sentence based on their semantic knowledge. For instance, given a sentence Katie bought some chocolate cook- ies, people know Katie is a buyer, chocolate cook- ies are goods and belong to Food class etc. Existing machine learning approaches, however, face great challenges to address complicated MRC questions, as they do not have above semantic knowledge. Nevertheless, FrameNet (Fillmore, 1976; Bak- er et al., 1998), as a knowledge base, provides schematic scenario representation that could be po- tentially leveraged to better understand sentences. ∗Corresponding author: Ru Li. F Commerce buy FEs Buyer, Goods, ... LUs buy.v, buy.n, buyer.n, purchase.n,... T [Katie]Buyer boughtCommerce buy [some chocolate cookies]Goods F-to-F Commerce buy—-Shopping— Seeking—Locating Table 1: Example of F, FEs, LUs, T and F-to-F. It enables the development of wide-coverage frame parsers (Gildea and Jurafsky, 2002; Das et al., 2014), as well as various real-world application- s, ranging form event recognition (Liu et al., 2016), textual entailment (Burchardt et al., 2009), question answering (Ofoghi et al., 2009), narrative schemas (Chambers and Jurafsky, 2010) and paraphrase i- dentification (Zhang et al., 2018), etc. In particular, Frame (F) is defined as a composition of Lexical Units (LUs) and a set of Frame Elements (FEs). Given a sentence, if its certain word evokes a frame by matching a LU, then it is called Target (T). It is worth mentioning that FrameNet arranges different relevant frames into a network by defining Frame- to-Frame (F-to-F) relations. Table 1 provides an example of F, FEs, LUs, T and …

Content truncated to 500 words