A Joint Model for Entity Analysis: Coreference, Typing, and Linking

A Joint Model for Entity Analysis: Coreference, Typing, and Linking Greg Durrett and Dan Klein Computer Science Division University of California, Berkeley {gdurrett,klein}@cs.berkeley.edu Abstract We present a joint model of three core tasks in the entity analysis stack: coreference res- olution (within-document clustering), named entity recognition (coarse semantic typing), and entity linking (matching to Wikipedia en- tities). Our model is formally a structured con- ditional random field. Unary factors encode local features from strong baselines for each task. We then add binary and ternary factors to capture cross-task interactions, such as the constraint that coreferent mentions have the same semantic type. On the ACE 2005 and OntoNotes datasets, we achieve state-of-the- art results for all three tasks. Moreover, joint modeling improves performance on each task over strong independent baselines.1 1 Introduction How do we characterize the collection of entities present in a document? Two broad threads exist in the literature. The first is coreference resolution (Soon et al., 2001; Ng, 2010; Pradhan et al., 2011), which identifies clusters of mentions in a document referring to the same entity. This process gives us ac- cess to useful information about the referents of pro- nouns and nominal expressions, but because clusters are local to each document, it is often hard to situate document entities in a broader context. A separate line of work has considered the problem of entity linking or “Wikification” (Cucerzan, 2007; Milne and Witten, 2008; Ji and Grishman, 2011), where mentions are linked to entries in a given knowledge 1System available at http://nlp.cs.berkeley.edu base. This is useful for grounding proper entities, but in the absence of coreference gives an incom- plete picture of document content itself, in that nom- inal expressions and pronouns are left unresolved. In this paper, we describe a joint model of corefer- ence, entity linking, and semantic typing (named en- tity recognition) using a structured conditional ran- dom field. Variables in the model capture deci- sions about antecedence, semantic type, and entity links for each mention. Unary factors on these vari- ables incorporate features that are commonly em- ployed when solving each task in isolation. Bi- nary and higher-order factors capture interactions between pairs of tasks. For entity linking and NER, factors capture a mapping between NER’s seman- tic types and Wikipedia’s semantics as described by infoboxes, categories, and article text. Coreference interacts with the other tasks in a more complex way, via factors that softly encourage consistency of semantic types and entity links across coreference arcs, similar to the method of Durrett et al. (2013). Figure 1 shows an example of the effects such fac- tors can capture. The non-locality of coreference factors make exact inference intractable, but we find that belief propagation is a suitable approximation technique and performs well. Our joint modeling of these three tasks is moti- vated by their heavy interdependencies, which have been noted in previous work (discussed more in Section 7). Entity linking has been employed for coreference resolution (Ponzetto and Strube, 2006; Rahman and Ng, 2011; Ratinov …

Content truncated to 500 words