Sentence Structure and Word Relationship Modeling for Emphasis Selection

Proceedings of Recent Advances in Natural Language Processing, pages 1559–1566 Sep 1–3, 2021. https://doi.org/10.26615/978-954-452-072-4_175 1559 Sentence Structure and Word Relationship Modeling for Emphasis Selection ∗ Haoran Yang, Wai Lam The Chinese University of Hong Kong {hryang, wlam} @se.cuhk.edu.hk Abstract Emphasis Selection is a newly proposed task which focuses on choosing words for emphasis in short sentences. Traditional methods only consider the sequence information of a sen- tence while ignoring the rich sentence struc- ture and word relationship information. In this paper, we propose a new framework that con- siders sentence structure via a sentence struc- ture graph and word relationship via a word similarity graph. The sentence structure graph is derived from the parse tree of a sentence. The word similarity graph allows nodes to share information with their neighbors since we argue that in emphasis selection, similar words are more likely to be emphasized to- gether. Graph neural networks are employed to learn the representation of each node of these two graphs. Experimental results demon- strate that our framework can achieve superior performance. 1 Introduction Emphasis Selection recently proposed by (Shirani et al., 2019) aims to select candidate words for em- phasis in short sentences. By emphasizing words, people’s intent can be better conveyed, which is useful in a variety of applications. For example, it can be used in spoken language processing to generate more expressive sentences and be used to enable automated design assistance in author- ing, i.e., labeling important parts in a paragraph or in a poster title. Although it seems that this task is highly similar to the task of keyword ex- traction (Gupta, 2017), these two tasks are fun- damentally different. The first difference is that keyword extraction focuses on a paragraph which is composed of multiple sentences while empha- sis selection aims to choose words from a short sentence. This difference implies that modeling ∗The work described in this paper is substantially sup- ported by a grant from the Direct Grant of the Faculty of Engineering, CUHK (Project Code: 4055093). Figure 1: Two examples of the emphasis selection. Words with darker background indicate that more peo- ple agree to emphasize. sentence structure is more effective in emphasis se- lection. The second difference is that many global word statistics methods employed in keyword ex- traction such as TF-IDF and word co-occurence frequency will not work in this task, because for short sentences, it is meaningless to count word frequency and whether the word should be empha- sized has nothing to do with the frequency of the word. In addition, keyword extraction requires that the collected keywords are diverse, which means that if two words have similar meaning, only one should be kept. However, in emphasis selection, similar words tend to be emphasized together. Em- phasis selection also shares some resemblances with entity recognition (Yadav and Bethard, 2018). But one major difference is that the parts of speech of emphasized words are more diverse and the re- lation of adjacent words is weaker in the emphasis …

Content truncated to 500 words