Cross-Lingual Induction and Transfer of Verb Classes Based on Word Vector Space Specialisation

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 2546–2558 Copenhagen, Denmark, September 7–11, 2017. c⃝2017 Association for Computational Linguistics Cross-Lingual Induction and Transfer of Verb Classes Based on Word Vector Space Specialisation Ivan Vuli´c1 , Nikola Mrkši´c2 and Anna Korhonen1 1 Language Technology Lab, University of Cambridge, UK 2 Dialogue Systems Group, University of Cambridge, UK {iv250,nm480,alk23}@cam.ac.uk Abstract Existing approaches to automatic VerbNet- style verb classification are heavily de- pendent on feature engineering and there- fore limited to languages with mature NLP pipelines. In this work, we propose a novel cross-lingual transfer method for inducing VerbNets for multiple languages. To the best of our knowledge, this is the first study which demonstrates how the architectures for learning word embeddings can be ap- plied to this challenging syntactic-semantic task. Our method uses cross-lingual trans- lation pairs to tie each of the six target lan- guages into a bilingual vector space with English, jointly specialising the representa- tions to encode the relational information from English VerbNet. A standard cluster- ing algorithm is then run on top of the VerbNet-specialised representations, using vector dimensions as features for learning verb classes. Our results show that the pro- posed cross-lingual transfer approach sets new state-of-the-art verb classification per- formance across all six target languages explored in this work. 1 Introduction Playing a key role in conveying the meaning of a sentence, verbs are famously complex. They dis- play a wide range of syntactic-semantic behaviour, expressing the semantics of an event as well as rela- tional information among its participants (Jackend- off, 1972; Gruber, 1976; Levin, 1993, inter alia). Lexical resources which capture the variabil- ity of verbs are instrumental for many Natural Language Processing (NLP) applications. One of the richest verb resources currently available for English is VerbNet (Kipper et al., 2000; Kipper, 2005).1 Based on the work of Levin (1993), this largely hand-crafted taxonomy organises verbs into classes on the basis of their shared syntactic- semantic behaviour. Providing a useful level of gen- eralisation for many NLP tasks, VerbNet has been used to support semantic role labelling (Swier and Stevenson, 2004; Giuglea and Moschitti, 2006), se- mantic parsing (Shi and Mihalcea, 2005), word sense disambiguation (Brown et al., 2011), dis- course parsing (Subba and Di Eugenio, 2009), in- formation extraction (Mausam et al., 2012), text mining applications (Lippincott et al., 2013; Rimell et al., 2013), research into human language acqui- sition (Korhonen, 2010), and other tasks. This benefit for English NLP has motivated the development of VerbNets for languages such as Spanish and Catalan (Aparicio et al., 2008), Czech (Pala and Horák, 2008), and Mandarin (Liu and Chiang, 2008). However, end-to-end manual re- source development using Levin’s methodology is extremely time consuming, even when supported by translations of English VerbNet classes to other languages (Sun et al., 2010; Scarton et al., 2014). Approaches which aim to learn verb classes au- tomatically offer an attractive alternative. How- ever, existing methods rely on carefully engi- neered features that are extracted using …

Content truncated to 500 words