Lexi: A tool for adaptive, personalized text simplification

Proceedings of the 27th International Conference on Computational Linguistics, pages 245–258 Santa Fe, New Mexico, USA, August 20-26, 2018. 245 Lexi: A tool for adaptive, personalized text simplification Joachim Bingel1 Gustavo H. Paetzold2 Anders Søgaard1 1 Department of Computer Science, University of Copenhagen, Denmark 2 Federal University of Technology - Paraná, Brazil {bingel,soegaard}@di.ku.dk, ghpaetzold@utfpr.edu.br Abstract Most previous research in text simplification has aimed to develop generic solutions, assuming very homogeneous target audiences with consistent intra-group simplification needs. We argue that this assumption does not hold, and that instead we need to develop simplification systems that adapt to the individual needs of specific users. As a first step towards personalized simplification, we propose a framework for adaptive lexical simplification and introduce Lexi, a free open-source and easily extensible tool for adaptive, personalized text simplification. Lexi is easily installed as a browser extension, enabling easy access to the service for its users. 1 Introduction Many a research paper on text simplification starts out by sketching the problem of text simplification as rewriting a text such that it becomes easier to read, changing or removing as little of its informational content as possible (Zhu et al., 2010; Coster and Kauchak, 2011; De Belder and Moens, 2010; Paetzold and Specia, 2015; Bingel and Søgaard, 2016). Such a statement may describe the essence of simplifica- tion as a research task, but it hides the fact that it is not always easy to decide what is easy for a particular user. This paper discusses why we need custom-tailored simplifications for individual users, and argues that previous research on non-adaptive text simplification has been too generic to unfold the full potential of text simplification. Even when limiting ourselves to lexical substitution, i.e. the task of reducing the complexity of a document by replacing difficult words with easier-to-read synonyms, we see plenty of evidence that, for instance, dyslexics are highly individual in what material is deemed easy and complex (Ziegler et al., 2008). Lexi, which we introduce in this paper, is a free, open-source and easily extensible tool for adaptively learning what items specific users find difficult, using this information to provide better (lex- ical) simplification. Our system initially serves Danish, but is easily extended to further languages. For surveys of text simplification, including resources across languages, see Siddharthan (2014), Shardlow (2014b) and Collins-Thompson (2014). 1.1 There is no one-size-fits-all solution to text simplification Text simplification is a diverse task, or perhaps rather a family of tasks, with a number of different target audiences that different papers and research projects have focused on. Among the most prominent tar- get audiences are foreign language learners, for whom various approaches to simplifying text have been pursued, often focusing on lexical (Tweissi, 1998) but also sentence-level simplification (Liu and Mat- sumoto, 2016). Other notable groups that have been specifically targeted in text simplification research include dyslexics (Rello et al., 2013), and the aphasic (Carroll et al., 1998), for whom particularly long words and sentences, but also certain surface forms such as …

Content truncated to 500 words