Cross-Lingual Distributional Profiles of Concepts for Measuring Semantic Distance

Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning, pp. 571–580, Prague, June 2007. c⃝2007 Association for Computational Linguistics Cross-lingual Distributional Profiles of Concepts for Measuring Semantic Distance Saif Mohammad† Iryna Gurevych∗ Graeme Hirst† Torsten Zesch∗ †Dept. of Computer Science University of Toronto Toronto, Canada {smm,gh}@cs.toronto.edu ∗Ubiquitous Knowledge Processing Group Darmstadt University of Technology Darmstadt, Germany {gurevych,zesch}@tk.informatik.tu-darmstadt.de Abstract We present the idea of estimating seman- tic distance in one, possibly resource-poor, language using a knowledge source in an- other, possibly resource-rich, language. We do so by creating cross-lingual distributional profiles of concepts, using a bilingual lexi- con and a bootstrapping algorithm, but with- out the use of any sense-annotated data or word-aligned corpora. The cross-lingual measures of semantic distance are evaluated on two tasks: (1) estimating semantic dis- tance between words and ranking the word pairs according to semantic distance, and (2) solving Reader’s Digest ‘Word Power’ problems. In task (1), cross-lingual mea- sures are superior to conventional monolin- gual measures based on a wordnet. In task (2), cross-lingual measures are able to solve more problems correctly, and despite scores being affected by many tied answers, their overall performance is again better than the best monolingual measures. 1 Introduction Accurately estimating the semantic distance be- tween concepts or between words in context has per- vasive applications in computational linguistics, in- cluding machine translation, information retrieval, speech recognition, spelling correction, and text cat- egorization (see Budanitsky and Hirst (2006) for dis- cussion), and it is becoming clear that basing such measures on a combination of corpus statistics with a knowledge source, such as a dictionary, published thesaurus, or WordNet, can result in higher accu- racies (Mohammad and Hirst, 2006b). This is be- cause such knowledge sources capture semantic in- formation about concepts and, to some extent, world knowledge. They also act as sense inventories for the words in a language. However, applying algorithms for semantic dis- tance to most languages is hindered by the lack of linguistic resources. In this paper, we propose a new method that allows us to compute semantic dis- tance in a possibly resource-poor language by seam- lessly combining its text with a knowledge source in a different, preferably resource-rich, language. We demonstrate the approach by combining German text with an English thesaurus to create English– German distributional profiles of concepts, which in turn will be used to measure the semantic distance between German words. Two classes of methods have been used in deter- mining semantic distance. Semantic measures of concept-distance, such as those of Jiang and Con- rath (1997) and Resnik (1995), rely on the structure of a knowledge source, such as WordNet, to deter- mine the distance between two concepts defined in it (see Budanitsky and Hirst (2006) for a survey). Distributional measures of word-distance1, such as cosine and α-skew divergence (Lee, 2001), deem 1Many distributional approaches represent the sets of con- texts of the target words as points in multidimensional co- occurrence space or as …

Content truncated to 500 words