Latent Structures for Coreference Resolution

Latent Structures for Coreference Resolution Sebastian Martschat and Michael Strube Heidelberg Institute for Theoretical Studies gGmbH Schloss-Wolfsbrunnenweg 35, 69118 Heidelberg, Germany (sebastian.martschat|michael.strube)@h-its.org Abstract Machine learning approaches to coreference resolution vary greatly in the modeling of the problem: while early approaches operated on the mention pair level, current research fo- cuses on ranking architectures and antecedent trees. We propose a unified representation of different approaches to coreference reso- lution in terms of the structure they operate on. We represent several coreference reso- lution approaches proposed in the literature in our framework and evaluate their perfor- mance. Finally, we conduct a systematic anal- ysis of the output of these approaches, high- lighting differences and similarities. 1 Introduction Coreference resolution is the task of determining which mentions in a text are used to refer to the same real-world entity. The era of statistical natural lan- guage processing saw the shift from rule-based ap- proaches (Hobbs, 1976; Lappin and Leass, 1994) to increasingly sophisticated machine learning models. While early approaches cast the problem as binary classification of mention pairs (Soon et al., 2001), recent approaches make use of complex structures to represent coreference relations (Yu and Joachims, 2009; Fernandes et al., 2014). The aim of this paper is to devise a framework for coreference resolution that leads to a unified rep- resentation of different approaches to coreference resolution in terms of the structure they operate on. Previous work in other areas of natural lan- guage processing such as parsing (Klein and Man- ning, 2001) and machine translation (Lopez, 2009) has shown that providing unified representations of approaches to a problem deepens its understanding and can also lead to empirical improvements. By im- plementing popular approaches in this framework, we can highlight structural differences and similar- ities between them. Furthermore, this establishes a setting to systematically analyze the contribution of the underlying structure to performance, while fix- ing parameters such as preprocessing and features. In particular, we analyze approaches to corefer- ence resolution and point out that they mainly dif- fer in the structures they operate on. We then note that these structures are not annotated in the train- ing data (Section 2). Motivated by this observation, we develop a machine learning framework for struc- tured prediction with latent variables for coreference resolution (Section 3). We formalize the mention pair model (Soon et al., 2001; Ng and Cardie, 2002), mention ranking architectures (Denis and Baldridge, 2008; Chang et al., 2012) and antecedent trees (Fer- nandes et al., 2014) in our framework and high- light key differences and similarities (Section 4). Fi- nally, we present an extensive comparison and anal- ysis of the implemented approaches, both quantita- tive and qualitative (Sections 5 and 6). Our analy- sis shows that a mention ranking architecture with latent antecedents performs best, mainly due to its ability to structurally model determining anaphoric- ity. Finally, we briefly describe how entity-centric approaches fit into our framework (Section 7). An open source toolkit which implements the ma- chine learning framework and the approaches …

Content truncated to 500 words