Resolving Discourse-Deictic Pronouns: A Two-Stage Approach to Do It

Proceedings of the Fourth Joint Conference on Lexical and Computational Semantics (*SEM 2015), pages 299–308, Denver, Colorado, June 4–5, 2015. Resolving Discourse-Deictic Pronouns: A Two-Stage Approach to Do It Sujay Kumar Jauhar Carnegie Mellon University Pittsburgh, PA 15213, USA sjauhar@cs.cmu.edu Raul D. Guerra University of Maryland College Park, MD 20742, USA rguerra@cs.umd.edu Edgar Gonzàlez Google Research Mountain View, CA 94043, USA edgargip@google.com Marta Recasens Google Research Mountain View, CA 94043, USA recasens@google.com Abstract Discourse deixis is a linguistic phenomenon in which pronouns have verbal or clausal, rather than nominal, antecedents. Studies have esti- mated that between 5% and 10% of pronouns in non-conversational data are discourse deic- tic. However, current coreference resolution systems ignore this phenomenon. This paper presents an automatic system for the detec- tion and resolution of discourse-deictic pro- nouns. We introduce a two-step approach that first recognizes instances of discourse-deictic pronouns, and then resolves them to their ver- bal antecedent. Both components rely on lin- guistically motivated features. We evaluate the components in isolation and in combina- tion with two state-of-the-art coreference re- solvers. Results show that our system out- performs several baselines, including the only comparable discourse deixis system, and leads to small but statistically significant improve- ments over the full coreference resolution sys- tems. An error analysis lays bare the need for a less strict evaluation of this task. 1 Introduction Coreference resolution is a central problem in Nat- ural Language Processing with a broad range of ap- plications such as summarization (Steinberger et al., 2007), textual entailment (Mirkin et al., 2010), in- formation extraction (McCarthy and Lehnert, 1995), and dialogue systems (Strube and Müller, 2003). Traditionally, the resolution of noun phrases (NPs) has been the focus of coreference research (Ng, 2010). However, NPs are not the only participants in coreference, since verbal or clausal mentions can also take part in coreference relations. For example, consider: (1) The United States says it may invite Israeli and Palestinian negotiators to Washington. (2) Without planning it in advance, they chose to settle here. In (1), the antecedent of the pronoun is an NP, while in (2) the antecedent1 is a clause2 (Webber, 1988). Current state-of-the-art coreference resolution sys- tems (Lee et al., 2011; Fernandes et al., 2012; Dur- rett and Klein, 2014; Björkelund and Kuhn, 2014) focus on the former and ignore the latter cases. Corpus studies across several languages (Eckert and Strube, 2000; Botley, 2006; Recasens, 2008) have estimated that between 5% and 10% of pro- nouns in non-conversational data, and up to 20% in conversational, have verbal antecedents. A corefer- ence system that is able to handle discourse deixis will thus be more accurate, and benefit downstream applications. In this paper we present an automatic system that processes discourse-deictic pronouns. We resolve the three pronouns it, this and that, which can appear in linguistic contexts that reflect the phenomenon il- lustrated in (2). Our system has a modular archi- tecture consisting of two independent stages: clas- sification and resolution. The first stage classifies a pronoun …

Content truncated to 500 words