Plurality, Negation, and Quantification:Towards Comprehensive Quantifier Scope Disambiguation

Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics, pages 64–72, Sofia, Bulgaria, August 4-9 2013. c⃝2013 Association for Computational Linguistics Plurality, Negation, and Quantification: Towards Comprehensive Quantifier Scope Disambiguation Mehdi Manshadi, Daniel Gildea, and James Allen University of Rochester 734 Computer Studies Building Rochester, NY 14627 mehdih,gildea,james@cs.rochester.edu Abstract Recent work on statistical quantifier scope disambiguation (QSD) has improved upon earlier work by scoping an arbitrary num- ber and type of noun phrases. No corpus- based method, however, has yet addressed QSD when incorporating the implicit uni- versal of plurals and/or operators such as negation. In this paper we report early, though promising, results for automatic QSD when handling both phenomena. We also present a general model for learning to build partial orders from a set of pair- wise preferences. We give an n log n algo- rithm for finding a guaranteed approxima- tion of the optimal solution, which works very well in practice. Finally, we signifi- cantly improve the performance of the pre- vious model using a rich set of automati- cally generated features. 1 Introduction The sentence there is one faculty member in ev- ery graduate committee is ambiguous with respect to quantifier scoping, since there are at least two possible readings: If one has wide scope, there is a unique faculty member on every committee. If every has wide scope, there can be different fac- ulty members on each committee. Over the past decade there has been some work on statistical quantifier scope disambiguation (QSD) (Higgins and Sadock, 2003; Galen and MacCartney, 2004; Manshadi and Allen, 2011a). However, the extent of the work has been quite limited for several rea- sons. First, in the past two decades, the main focus of the NLP community has been on shallow text processing. As a deep processing task, QSD is not essential for many NLP applications that do not re- quire deep understanding. Second, there has been a lack of comprehensive scope-disambiguated cor- pora, resulting in the lack of work on extensive statistical QSD. Third, QSD has often been con- sidered only in the context of explicit quantifica- tion such as each and every versus some and a/an. These co-occurrences do not happen very often in real-life data. For example, Higgins and Sadock (2003) find fewer than 1000 sentences with two or more explicit quantifiers in the Wall Street journal section of Penn Treebank. Furthermore, for more than 60% of those sentences, the order of the quan- tifiers does not matter, either as a result of the logi- cal equivalence (as in two existentials), or because they do not have any scope interaction. Having said that, with deep language processing receiving more attention in recent years, QSD is becoming a real-life issue.1 At the same time, new scope-disambiguated corpora have become avail- able (Manshadi et al., 2011b). In this paper, we aim at tackling the third issue mentioned above. We push statistical QSD beyond explicit quantifi- cation, and address an interesting, yet practically important, problem in QSD: …

Content truncated to 500 words