A Multilingual Evaluation Dataset for Monolingual Word Sense Alignment

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 3232–3242 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 3232 A Multilingual Evaluation Dataset for Monolingual Word Sense Alignment Sina Ahmadi*, John P. McCrae*, Sanni Nimb1, Fahad Khan3, Monica Monachini3, Bolette S. Pedersen8, Thierry Declerck2,12, Tanja Wissik2, Andrea Bellandi3, Irene Pisani4, Thomas Troelsg˚ard1, Sussi Olsen8, Simon Krek5, Veronika Lipp6, Tam´as V´aradi6, L´aszl´o Simon6, Andr´as Gy˝orffy6, Carole Tiberius9, Tanneke Schoonheim9, Yifat Ben Moshe10, Maya Rudich10, Raya Abu Ahmad10, Dorielle Lonke10, Kira Kovalenko11, Margit Langemets13, Jelena Kallas13, Oksana Dereza7, Theodorus Fransen7, David Cillessen7, David Lindemann14, Mikel Alonso14, Ana Salgado15 Jos´e Luis Sancho16, Rafael-J. Ure˜na-Ruiz16, Jordi Porta Zamorano16, Kiril Simov17, Petya Osenova17, Zara Kancheva17, Ivaylo Radev17, Ranka Stankovi´c18, Andrej Perdih19, Dejan Gabrovˇsek19 *Insight Centre for Data Analytics, National University of Ireland, Galway {sina.ahmadi,john.mccrae}@insight-centre.org (other affiliations in Appendix A) Abstract Aligning senses across resources and languages is a challenging task with beneficial applications in the field of natural language process- ing and electronic lexicography. In this paper, we describe our efforts in manually aligning monolingual dictionaries. The alignment is carried out at sense-level for various resources in 15 languages. Moreover, senses are annotated with possible semantic relationships such as broadness, narrowness, relatedness, and equivalence. In comparison to previous datasets for this task, this dataset covers a wide range of languages and resources and focuses on the more challenging task of linking general-purpose language. We believe that our data will pave the way for further advances in alignment and evaluation of word senses by creating new solutions, particularly those notoriously requiring data such as neural networks. Our resources are publicly available at https://github.com/elexis-eu/MWSA. Keywords: lexical semantic resources, sense alignment, lexicography, language resource 1. Introduction Lexical semantic resources (LSRs) are knowledge reposi- tories that provide the vocabulary of a language in a de- scriptive and structured way. One of the famous examples of LSRs are dictionaries. Dictionaries form an important foundation of numerous natural language processing (NLP) tasks, including word sense disambiguation, machine trans- lation, question answering and automatic summarization. However, the task of combining dictionaries from different sources is difficult, especially for the case of mapping the senses of entries, which often differ significantly in gran- ularity and coverage. Approaches so far have mostly only been evaluated on named entities and quite specific domain language. In order to support a shared task at the GLOB- ALEX workshop1, we have developed a new baseline that covers 15 languages and will provide a new baseline for the task of monolingual word sense alignment. Different dictionaries and related resources such as word- nets and encyclopedia have significant differences in struc- ture and heterogeneity in content, which makes aligning information across resources and languages a challenging task. Word sense alignment (WSA) is a more specific task of linking dictionary content at sense level which has been proved to be beneficial in various NLP tasks, such as word- sense disambiguation (Navigli and Ponzetto, 2012), seman- tic role labeling (Palmer, 2009) and information …

Content truncated to 500 words