The role of coreference resolution in extractive summarization

In graph based extractive summarization the extraction of sentences can be determined by the importance of the words it contains and the association between the sentences. This method considers the word separately. It does not provide association link handling between the words. This paper addresses to what extent the integration of coreference resolution into the summarization process can improve the performance of extractive summarization. The simple entity based coreference resolution based on three set of features is proposed in this paper. Evaluation of Reuters dataset on baseline graph based summarization algorithm shows that the integration of coreference resolution algorithm can have a positive impact on extracting the informative sentences.