Generic Relation Identification: Models and Evaluation

Generic relation identification (GRI) aims to build models of relation-forming entity pairs that can be transferred across domains without modification of model parameters. GRI has high utility in terms of cheap components for applications like summarisation, automated data exploration and initialisation of bootstrapping of relation extraction. A detailed evaluation of GRI is presented for the first time, including explicit tests of portability between newswire and biomedical domains. Experimental results show that a novel approach incorporating dependency parsing is better in terms of recall. And, accuracy is shown to be comparable across domains.