Entity Linking via Joint Encoding of Types, Descriptions, and Context

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 2681–2690 Copenhagen, Denmark, September 7–11, 2017. c⃝2017 Association for Computational Linguistics Entity Linking via Joint Encoding of Types, Descriptions, and Context Nitish Gupta∗ University of Pennsylvania Philadelphia, PA nitishg@seas.upenn.edu Sameer Singh University of California Irvine, CA sameer@uci.edu Dan Roth∗ University of Pennsylvania Philadelphia, PA danroth@seas.upenn.edu Abstract For accurate entity linking, we need to cap- ture various information aspects of an en- tity, such as its description in a KB, con- texts in which it is mentioned, and struc- tured knowledge. Additionally, a linking system should work on texts from different domains without requiring domain-specific training data or hand-engineered features. In this work we present a neural, mod- ular entity linking system that learns a unified dense representation for each en- tity using multiple sources of information, such as its description, contexts around its mentions, and its fine-grained types. We show that the resulting entity link- ing system is effective at combining these sources, and performs competitively, some- times out-performing current state-of-the- art systems across datasets, without requir- ing any domain-specific training data or hand-engineered features. We also show that our model can effectively “embed” en- tities that are new to the KB, and is able to link its mentions accurately. 1 Introduction Entity linking, the task of identifying the real-world entity a mention in text refers to, provides the abil- ity to ground text to existing knowledge bases, and thus supports multiple natural language understand- ing, and knowledge acquisition tasks. A key challenge for successful entity linking is the need to capture semantic and background information at various levels of granularity. For example, to resolve the mention “India” in “India plays a match in England today” to the correct en- tity, India cricket team, one needs to use ∗Work performed while these authors were at UIUC. mention-level context to identify that the sentence refers to a sports team (using plays and match), use document-level context to identify the sport, and information about the entity to realize that India cricket team is a sports team and the string “India” may refer to it. The problem has been studied extensively by employing a variety of machine learning, and inference methods, in- cluding a pipeline of deterministic modules (Ling et al., 2015), simple classifiers (Cucerzan, 2007; Ratinov et al., 2011), graphical models (Durrett and Klein, 2014), classifiers augmented with ILP infer- ence (Cheng and Roth, 2013), and more recently, neural approaches (He et al., 2013; Sun et al., 2015; Francis-Landau et al., 2016). We present a neural approach to linking1 that learns a dense unified representation of entities by encoding the semantic and background informa- tion from multiple sources – encyclopedic entity descriptions, entity-type information, and the con- texts the entity occurs in – thus capturing differ- ent aspects of the “meaning” of an entity. Hence, we overcome the shortcomings of several existing models that do not capture all these aspects. For example, methods, such as Vinculum (Ling et al., …

Content truncated to 500 words