Anchor Word Extraction Enhancement using Coreference Resolution for Narrative texts

Text summarization is an important area in Natural Language Processing studies. It has been shown that good representative words extracted from a story, called anchor words, give an insight of the story. A human can link those words to get a high level idea of the story. However, in order to get the gist of a story, determining important words of a story is not enough. A mechanism capable of determining words which are more important than others is essential to connect the words to get a meaningful idea. In this paper, an algorithm has been proposed where Coreference Resolution is used to determine the anchor words. It has been shown through experiments that the proposed method improves the anchor Word detection task tremendously.