Capturing Semantic Similarity for Entity Linking with Convolutional Neural Networks

Proceedings of NAACL-HLT 2016, pages 1256–1261, San Diego, California, June 12-17, 2016. c⃝2016 Association for Computational Linguistics Capturing Semantic Similarity for Entity Linking with Convolutional Neural Networks Matthew Francis-Landau, Greg Durrett and Dan Klein Computer Science Division University of California, Berkeley {mfl,gdurrett,klein}@cs.berkeley.edu Abstract A key challenge in entity linking is making ef- fective use of contextual information to dis- ambiguate mentions that might refer to differ- ent entities in different contexts. We present a model that uses convolutional neural net- works to capture semantic correspondence be- tween a mention’s context and a proposed tar- get entity. These convolutional networks oper- ate at multiple granularities to exploit various kinds of topic information, and their rich pa- rameterization gives them the capacity to learn which n-grams characterize different topics. We combine these networks with a sparse lin- ear model to achieve state-of-the-art perfor- mance on multiple entity linking datasets, out- performing the prior systems of Durrett and Klein (2014) and Nguyen et al. (2014).1 1 Introduction One of the major challenges of entity linking is re- solving contextually polysemous mentions. For ex- ample, Germany may refer to a nation, to that na- tion’s government, or even to a soccer team. Past approaches to such cases have often focused on col- lective entity linking: nearby mentions in a docu- ment might be expected to link to topically-similar entities, which can give us clues about the identity of the mention currently being resolved (Ratinov et al., 2011; Hoffart et al., 2011; He et al., 2013; Cheng and Roth, 2013; Durrett and Klein, 2014). But an even simpler approach is to use context information from just the words in the source document itself to make sure the entity is being resolved sensibly in context. In past work, these approaches have typi- cally relied on heuristics such as tf-idf (Ratinov et 1Source available at github.com/matthewfl/nlp-entity-convnet al., 2011), but such heuristics are hard to calibrate and they capture structure in a coarser way than learning-based methods. In this work, we model semantic similarity be- tween a mention’s source document context and its potential entity targets using convolutional neural networks (CNNs). CNNs have been shown to be ef- fective for sentence classification tasks (Kalchbren- ner et al., 2014; Kim, 2014; Iyyer et al., 2015) and for capturing similarity in models for entity linking (Sun et al., 2015) and other related tasks (Dong et al., 2015; Shen et al., 2014), so we expect them to be effective at isolating the relevant topic semantics for entity linking. We show that convolutions over mul- tiple granularities of the input document are useful for providing different notions of semantic context. Finally, we show how to integrate these networks with a preexisting entity linking system (Durrett and Klein, 2014). Through a combination of these two distinct methods into a single system that leverages their complementary strengths, we achieve state-of- the-art performance across several datasets. 2 Model Our model focuses on two core ideas: first, that topic semantics at different granularities in a document …

Content truncated to 500 words