Trust, but Verify! Better Entity Linking through Automatic Verification

Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers, pages 828–838, Valencia, Spain, April 3-7, 2017. c⃝2017 Association for Computational Linguistics Trust, but Verify! Better Entity Linking through Automatic Verification Benjamin Heinzerling∗ AIPHES Heidelberg Institute for Theoretical Studies benjamin.heinzerling@h-its.org Michael Strube Heidelberg Institute for Theoretical Studies michael.strube@h-its.org Chin-Yew Lin Microsoft Research cyl@microsoft.com Abstract We introduce automatic verification as a post-processing step for entity linking (EL). The proposed method trusts EL sys- tem results collectively, by assuming en- tity mentions are mostly linked correctly, in order to create a semantic profile of the given text using geospatial and temporal information, as well as fine-grained entity types. This profile is then used to auto- matically verify each linked mention indi- vidually, i.e., to predict whether it has been linked correctly or not. Verification allows leveraging a rich set of global and pairwise features that would be prohibitively expen- sive for EL systems employing global in- ference. Evaluation shows consistent im- provements across datasets and systems. In particular, when applied to state-of-the- art systems, our method yields an abso- lute improvement in linking performance of up to 1.7 F1 on AIDA/CoNLL’03 and up to 2.4 F1 on the English TAC KBP 2015 TEDL dataset. 1 Introduction Entity linking (EL) is the task of automatically linking mentions of entities such as persons, loca- tions, or organizations to their corresponding entry in a knowledge base (KB). The task is generally approached by generating a set of candidate enti- ties1 for a given mention and then ranking those candidates. Approaches differ in whether they rank a mention’s candidates independently of the candidates of other mentions (“local inference”) or ∗The majority of this work was done during an internship at Microsoft Research Asia. 1We use entity to refer to both real-word entities and to their corresponding entries in the KB. whether they rank all candidates of all mentions si- multaneously by incorporating a global coherence measure into the optimization goal (“global infer- ence”). While linguistically well-founded in the con- cept of lexical cohesion (Halliday and Hasan, 1976), global inference approaches (Kulkarni et al., 2009; Hoffart et al., 2011a) do not scale well with number of mentions and number of candidate entities. In contrast, local approaches do not suffer from scalability issues, since they only optimize the similarity between mention con- text and candidate KB entry text (Bunescu and Pas¸ca, 2006; Cucerzan, 2007), usually also includ- ing a popularity prior2 (Milne and Witten, 2008; Spitkovsky and Chang, 2012). Recent local ap- proaches achieve state-of-the-art results by using convolutional neural networks to capture similar- ity at multiple context sizes (Francis-Landau et al., 2016), but, by definition, fail to take global coher- ence into account. To avoid the trade-off between the efficiency of local inference on the one hand and the coherence benefits of global inference on the other, we pro- pose a two-stage approach: In the first stage, can- didate entities are ranked by a fast, local inference- based …

Content truncated to 500 words