Unsupervised Entity Linking with Abstract Meaning Representation

Human Language Technologies: The 2015 Annual Conference of the North American Chapter of the ACL, pages 1130–1139, Denver, Colorado, May 31 – June 5, 2015. c⃝2015 Association for Computational Linguistics Unsupervised Entity Linking with Abstract Meaning Representation Xiaoman Pan1, Taylor Cassidy2, Ulf Hermjakob3, Heng Ji1, Kevin Knight3 1Computer Science Department, Rensselaer Polytechnic Institute {panx2,jih}@rpi.edu 2IBM Research & Army Research Laboratory taylor.cassidy.civ@mail.mil 3Information Sciences Institute, University of Southern California {ulf,knight}@isi.edu Abstract Most successful Entity Linking (EL) meth- ods aim to link mentions to their referent en- tities in a structured Knowledge Base (KB) by comparing their respective contexts, of- ten using similarity measures. While the KB structure is given, current methods have suf- fered from impoverished information repre- sentations on the mention side. In this paper, we demonstrate the effectiveness of Abstract Meaning Representation (AMR) (Banarescu et al., 2013) to select high quality sets of en- tity “collaborators” to feed a simple similar- ity measure (Jaccard) to link entity mentions. Experimental results show that AMR captures contextual properties discriminative enough to make linking decisions, without the need for EL training data, and that system with AMR parsing output outperforms hand labeled tradi- tional semantic roles as context representation for EL. Finally, we show promising prelimi- nary results for using AMR to select sets of “coherent” entity mentions for collective en- tity linking 1. 1 Introduction The Entity Linking (EL) task (Ji et al., 2010; Ji et al., 2011; Ji et al., 2014) aims at automati- cally linking each named entity mention appearing in a source text document to its unique referent in a target knowledge base (KB). For example, con- sider the following sentence posted to a discussion forum during the 2012 U.S. presidential election: 1The web service of this EL system is at: blender02.cs.rpi.edu:3300 and some related AMR tools are at: github.com/panx27/amr-reader “Where would McCain be without Sarah?”. An Entity Linker should link the entity mentions “Mc- Cain” and “Sarah” to the entities John McCain and Sarah Palin, respectively, which serve as unique identifiers for the real people. A typical EL system works as follows. Given a mention m (a string in a source document), the top N most likely entity referents from the KB are enu- merated based on prior knowledge about which en- tities are most likely referred to using m. The can- didate entities are re-ranked to ultimately link each mention to the top entity in its candidate list. Re- ranking consists of two key elements: context rep- resentation and context comparison. For a given mention, candidate entities are re-ranked based on a comparison of information obtained from the con- text of m with known structured and/or unstructured information associated with the top N KB entities, which can be considered the “context” of the KB en- tity2. The basic intuition is that the entity referents of m and related mentions should be similarly con- nected in the KB. However, there might be many entity mentions in the context of a target entity mention that could po- tentially …

Content truncated to 500 words