Generationary or “How We Went beyond Word Sense Inventories and Learned to Gloss”

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 7207–7221, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 7207 Generationary or: “How We Went beyond Word Sense Inventories and Learned to Gloss” Michele Bevilacqua∗1, Marco Maru∗1,2, and Roberto Navigli1 1Sapienza NLP Group Department of Computer Science, Sapienza University of Rome 2Department of Literature and Modern Cultures, Sapienza University of Rome firstname.lastname@uniroma1.it Abstract Mainstream computational lexical semantics embraces the assumption that word senses can be represented as discrete items of a prede- fined inventory. In this paper we show this needs not be the case, and propose a unified model that is able to produce contextually ap- propriate definitions. In our model, Genera- tionary, we employ a novel span-based encod- ing scheme which we use to fine-tune an En- glish pre-trained Encoder-Decoder system to generate glosses. We show that, even though we drop the need of choosing from a prede- fined sense inventory, our model can be em- ployed effectively: not only does Generation- ary outperform previous approaches in the gen- erative task of Definition Modeling in many settings, but it also matches or surpasses the state of the art in discriminative tasks such as Word Sense Disambiguation and Word-in- Context. Finally, we show that Generationary benefits from training on data from multiple inventories, with strong gains on various zero- shot benchmarks, including a novel dataset of definitions for free adjective-noun phrases. The software and reproduction materials are available at http://generationary.org. 1 Introduction Virtually all modern approaches to Word Sense Disambiguation (WSD), i.e. the task of automat- ically mapping a word in context to its meaning (Navigli, 2009), use predetermined word senses from a machine lexicon, both in supervised (Huang et al., 2019; Bevilacqua and Navigli, 2020; Scar- lini et al., 2020b) and in knowledge-based settings (Tripodi and Navigli, 2019; Scarlini et al., 2020a; Scozzafava et al., 2020). Nevertheless, researchers in Natural Language Processing (NLP), lexical se- mantics, and lexicography, have long been warning the community about the cognitively inaccurate ∗These authors contributed equally. nature of discrete sense boundaries (Rosch and Mervis, 1975; Kilgarriff, 1997; Tyler and Evans, 2001). As Kilgarriff (2007) argued, different lan- guage users have different understandings of words. This fact explains why inter-annotator agreement (ITA) estimates on WSD annotation tasks have never exceeded the figure of 80% (Edmonds and Kilgarriff, 2002; Navigli et al., 2007; Palmer et al., 2007). Moreover, this casts doubt upon the reliabil- ity of human-made inventories and “gold standard” evaluation datasets (Ramsey, 2017). Having no indisputable way of determining where one sense of a word ends and another begins, together with the fact that little consensus about how to represent word meaning has hitherto existed (Pustejovsky, 1991; Hanks, 2000; Nosofsky, 2011), are issues lying at the core of what makes WSD hard (Jack- son, 2019). Moreover, while English inventories of senses and corpora are widely available the same cannot be said for other languages (Scarlini et al., 2019; Barba et al., 2020; Pasini, 2020), and this lim- its the …

Content truncated to 500 words