Event-based summarization using critical temporal event term chain

Abstract

In this paper, we investigate whether temporal relations among event terms can help improve event-based summarization and text cohesion of final summaries. By connecting event terms with happens-before relations, we build a temporal event term graph for source documents. The event terms in the critical temporal event term chain identified from the maximal weakly connected component are used to evaluate the sentences in source documents. The most significant sentences are included in final summaries. Experiments conducted on the DUC 2001 corpus show that event-based summarization using the critical temporal event term chain is able to organize final summaries in a more coherent way and make improvement over the well-known tf idf-based and PageRank-based summarization approaches. © 2009 Springer Berlin Heidelberg.