Hypernyms under Siege: Linguistically-motivated Artillery for Hypernymy Detection

Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers, pages 65–75, Valencia, Spain, April 3-7, 2017. c⃝2017 Association for Computational Linguistics Hypernyms under Siege: Linguistically-motivated Artillery for Hypernymy Detection Vered Shwartz1, Enrico Santus2,3 and Dominik Schlechtweg4 1Bar-Ilan University, Ramat-Gan, Israel 2Singapore University of Technology and Design, Singapore 3The Hong Kong Polytechnic University, Hong Kong 4University of Stuttgart, Stuttgart, Germany {vered1986,esantus}@gmail.com, dominik.schlechtweg@gmx.de Abstract The fundamental role of hypernymy in NLP has motivated the development of many methods for the automatic identi- fication of this relation, most of which rely on word distribution. We investigate an extensive number of such unsupervised measures, using several distributional se- mantic models that differ by context type and feature weighting. We analyze the per- formance of the different methods based on their linguistic motivation. Comparison to the state-of-the-art supervised methods shows that while supervised methods gen- erally outperform the unsupervised ones, the former are sensitive to the distribution of training instances, hurting their relia- bility. Being based on general linguistic hypotheses and independent from training data, unsupervised measures are more ro- bust, and therefore are still useful artillery for hypernymy detection. 1 Introduction In the last two decades, the NLP community has invested a consistent effort in developing auto- mated methods to recognize hypernymy. Such ef- fort is motivated by the role this semantic relation plays in a large number of tasks, such as taxonomy creation (Snow et al., 2006; Navigli et al., 2011) and recognizing textual entailment (Dagan et al., 2013). The task has appeared to be, however, a challenging one, and the numerous approaches proposed to tackle it have often shown limitations. Early corpus-based methods have exploited pat- terns that may indicate hypernymy (e.g. “animals such as dogs”) (Hearst, 1992; Snow et al., 2005), but the recall limitation of this approach, requir- ing both words to co-occur in a sentence, mo- tivated the development of methods that rely on adaptations of the distributional hypothesis (Har- ris, 1954). The first distributional approaches were unsu- pervised, assigning a score for each (x, y) word- pair, which is expected to be higher for hyper- nym pairs than for negative instances. Evalua- tion is performed using ranking metrics inherited from information retrieval, such as Average Pre- cision (AP) and Mean Average Precision (MAP). Each measure exploits a certain linguistic hypoth- esis such as the distributional inclusion hypothesis (Weeds and Weir, 2003; Kotlerman et al., 2010) and the distributional informativeness hypothesis (Santus et al., 2014; Rimell, 2014). In the last couple of years, the focus of the re- search community shifted to supervised distribu- tional methods, in which each (x, y) word-pair is represented by a combination of x and y’s word vectors (e.g. concatenation or difference), and a classifier is trained on these resulting vectors to predict hypernymy (Baroni et al., 2012; Roller et al., 2014; Weeds et al., 2014). While the origi- nal methods were based on count-based vectors, in recent years they have been used with …

Content truncated to 500 words