Explainable Prediction of Text Complexity: The Missing Preliminaries for Text Simplification

Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pages 1086–1097 August 1–6, 2021. ©2021 Association for Computational Linguistics 1086 Explainable Prediction of Text Complexity: The Missing Preliminaries for Text Simplification Cristina Gˆarbacea1, Mengtian Guo4, Samuel Carton3, Qiaozhu Mei1,2 1Department of EECS, University of Michigan, Ann Arbor 2School of Information, University of Michigan, Ann Arbor 3Department of CS, University of Colorado, Boulder 4School of Information and Library Science, University of North Carolina, Chapel Hill Abstract Text simplification reduces the language com- plexity of professional content for accessibil- ity purposes. End-to-end neural network mod- els have been widely adopted to directly gen- erate the simplified version of input text, usu- ally functioning as a blackbox. We show that text simplification can be decomposed into a compact pipeline of tasks to ensure the trans- parency and explainability of the process. The first two steps in this pipeline are often ne- glected: 1) to predict whether a given piece of text needs to be simplified, and 2) if yes, to identify complex parts of the text. The two tasks can be solved separately using either lexical or deep learning methods, or solved jointly. Simply applying explainable complex- ity prediction as a preliminary step, the out-of- sample text simplification performance of the state-of-the-art, black-box simplification mod- els can be improved by a large margin. 1 Introduction Text simplification aims to reduce the language complexity of highly specialized textual content so that it is accessible for readers who lack adequate literacy skills, such as children, people with low education, people who have reading disorders or dyslexia, and non-native speakers of the language. Mismatch between language complexity and lit- eracy skills is identified as a critical source of bias and inequality in the consumers of systems built upon processing and analyzing professional text content. Research has found that it requires on average 18 years of education for a reader to prop- erly understand the clinical trial descriptions on ClinicalTrials.gov, and this introduces a potential self-selection bias to those trials (Wu et al., 2016). Text simplification has considerable potential to improve the fairness and transparency of text information systems. Indeed, the Simple English Wikipedia (simple.wikipedia.org) has been con- structed to disseminate Wikipedia articles to kids and English learners. In healthcare, consumer vo- cabulary are used to replace professional medical terms to better explain medical concepts to the public (Abrahamsson et al., 2014). In education, natural language processing and simplified text gen- eration technologies are believed to have the poten- tial to improve student outcomes and bring equal opportunities for learners of all levels in teaching, learning and assessment (Mayfield et al., 2019). Ironically, the definition of “text simplification” in literature has never been transparent. The term may refer to reducing the complexity of text at var- ious linguistic levels, ranging all the way through replacing individual words in the text to generat- ing a simplified document completely through a computer agent. In particular, lexical …

Content truncated to 500 words