Distributional Semantics for Resolving Bridging Mentions
Tim Feuerbach, Martin Johannes Riedl, Chris Biemann · Recent Advances in Natural Language Processing · 2015
We explore the impact of adding distributional knowledge to a state-of-the-art coreference resolution system. By integrating features based on word and context expansions from a distributional thesaurus (DT), automatically mined IS-A relationships and shallow syntactical clues into the Berkeley system (Durrett and Klein, 2013), we are able to increase its F1 score on bridging mentions, i.e. coreferent mentions with non-identical heads, by 8.29 points. Our semantic features improve over the Web-based features of Bansal and Klein (2012). Since bridging mentions are a hard but infrequent class of coreference, this leads to merely small improvements in the overall system.