An adaptive clustering model that integrates expert rules and N-gram statistics for coreference resolution

摘要: We present an adaptive clustering model for coreference resolution in which the expert rules of a state art deterministic system are used as features over pairs clusters. A significant advantage new approach is that can be easily added to system. demonstrate this by incorporating semantic compatibility neutral pronouns computed from web n-gram statistics. Experimental results show combination with overall performance improvement and substantial 5% F1 measure target pronouns.