Improving the Interpretation of Noun Phrases with Cross-linguistic Information

Proceedings of the 45th Annual Meeting of the Association of Computational Linguistics, pages 568–575, Prague, Czech Republic, June 2007. c⃝2007 Association for Computational Linguistics Improving the Interpretation of Noun Phrases with Cross-linguistic Information Roxana Girju University of Illinois at Urbana-Champaign girju@uiuc.edu Abstract This paper addresses the automatic classifi- cation of semantic relations in noun phrases based on cross-linguistic evidence from a set of five Romance languages. A set of novel semantic and contextual English– Romance NP features is derived based on empirical observations on the distribution of the syntax and meaning of noun phrases on two corpora of different genre (Europarl and CLUVI). The features were employed in a Support Vector Machines algorithm which achieved an accuracy of 77.9% (Eu- roparl) and 74.31% (CLUVI), an improve- ment compared with two state-of-the-art models reported in the literature. 1 Introduction Semantic knowledge is very important for any ap- plication that requires a deep understanding of natu- ral language. The automatic acquisition of semantic information in text has become increasingly impor- tant in ontology development, information extrac- tion, question answering, and other advanced natural language processing applications. In this paper we present a model for the auto- matic semantic interpretation of noun phrases (NPs), which is the task of determining the semantic re- lation among the noun constituents. For example, family estate encodes a POSSESSION relation, while dress of silk refers to PART-WHOLE. The problem, while simple to state is hard to solve. The rea- son is that the meaning of these constructions is most of the time ambiguous or implicit. Interpreting NPs correctly requires various types of information from world knowledge to complex context features. Moreover, the extension of this task to other natu- ral languages brings forward new issues and prob- lems. For instance, beer glass translates into tarro de cerveza in Spanish, bicchiere da birra in Italian, verre `a bi`ere in French, and pahar de bere in Roma- nian. Thus, an important research question is how do the syntactic constructions in the target language contribute to the preservation of meaning in context. In this paper we investigate noun phrases based on cross-linguistic evidence and present a domain inde- pendent model for their semantic interpretation. We aim at uncovering the general aspects that govern the semantics of NPs in English based on a set of five Romance languages: Spanish, Italian, French, Portuguese, and Romanian. The focus on Romance languages is well motivated. It is mostly true that English noun phrases translate into constructions of the form N P N in Romance languages where, as we will show below, the P (preposition) varies in ways that correlate with the semantics. Thus Ro- mance languages will give us another source of evi- dence for disambiguating the semantic relations in English NPs. We also present empirical observa- tions on the distribution of the syntax and meaning of noun phrases on two different corpora based on two state-of-the-art classification tag sets: Lauer’s set of 8 prepositions (Lauer, 1995) and our list of 22 semantic relations. …

Content truncated to 500 words