A Non-Autoregressive Edit-Based Approach to Controllable Text Simplification

Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 3757–3769 August 1–6, 2021. ©2021 Association for Computational Linguistics 3757 A Non-Autoregressive Edit-Based Approach to Controllable Text Simplification Sweta Agrawal, Weijia Xu, Marine Carpuat Department of Computer Science University of Maryland {sweagraw, weijia, marine}@cs.umd.edu Abstract We introduce a new approach for the task of Controllable Text Simplification, where sys- tems rewrite a complex English sentence so that it can be understood by readers at differ- ent grade levels in the US K-12 system. It uses a non-autoregressive model to iteratively edit an input sequence and incorporates lexi- cal complexity information seamlessly into the refinement process to generate simplifications that better match the desired output complexity than strong autoregressive baselines. Analysis shows that our model’s local edit operations are combined to achieve more complex sim- plification operations such as content deletion and paraphrasing, as well as sentence splitting. 1 Introduction Text simplification (TS) aims to automatically rewrite text so that it is easier to read. What makes text simple depends on its target audience (Xu et al., 2015): replacing complex or specialized terms with simpler synonyms might be helpful for non-native speakers (Petersen and Ostendorf, 2007; Allen, 2009) whereas restructuring text into short sentences with simple words might better match the literacy skills of children (Watanabe et al., 2009). Studies of simplification tools for deaf or hard-of- hearing users also show that they prefer lexical sim- plification to be applied on-demand (Alonzo et al., 2020). Yet, research in TS has mostly focused on developing models that generate a generic simpli- fied output for a given source text (Xu et al., 2015; Zhang and Lapata, 2017; Alva-Manchego et al., 2020). We contrast this Generic TS with Control- lable TS which specifies desired output properties. Prior work has addressed Controllable TS for ei- ther high-level properties, such as the target reading grade level for the entire text (Scarton and Specia, 2018; Nishihara et al., 2019), or low-level proper- ties, such as the compression ratio or the nature of Grade Text 10 Tesla is a maker of electric cars, which do not need gas and can be charged by being plugged into a wall socket. 5 Tesla cars can be charged by being plugged in, like a phone. They do not need any gas. 3 Tesla builds cars that do not need gas. Table 1: Simplified text changes depending on the read- ing grade level of the target audience. The bold font highlights changes compared to the grade 10 version. the simplification operation to use (Mallinson and Lapata, 2019; Martin et al., 2020; Maddela et al., 2020). Specifying the desired reading grade level might be more intuitive for lay users. However, it provides only weak control over the nature of sim- plification. As illustrated in Table 1, simplifying text to different grade levels results in diverse edits. To rewrite the grade 10 original for grade 5, the complex text is split into two sentences and para- phrased. When simplifying for grade 3, phrases are …

Content truncated to 500 words