Abstract
Bridging anaphora is a challenging task that has similarities with more familiar NLP tasks like coreference and relation extraction. Bridging includes part-whole, set-subset, and other relations that can be inferred from some mention previously in the discourse. Multi-level Bootstrapping is an approach that has been used for named entity recognition and relation extraction. I use bootstrapping to explore possible patterns that identify word pairs that are candidates for bridging relations. Using the patterns extracted from bootstrapping, I generate a distributional semantic resource. I implement a baseline bridging resolution system and report the performance with the addition of the bootstrapped lexical resource.