出版社：Academy & Industry Research Collaboration Center (AIRCC)
摘要：Many of previous research have proven that the usage of rhetorical relations is capable toenhance many applications such as text summarization, question answering and naturallanguage generation. This work proposes an approach that expands the benefit of rhetoricalrelations to address redundancy problem in text summarization. We first examined andredefined the type of rhetorical relations that is useful to retrieve sentences with identicalcontent and performed the identification of those relations using SVMs. By exploiting therhetorical relations exist between sentences, we generate clusters of similar sentences fromdocument sets. Then, cluster-based text summarization is performed using Conditional MarkovRandom Walk Model to measure the saliency scores of candidates summary. We evaluated ourmethod by measuring the cohesion and separation of the clusters and ROUGE score ofgenerated summaries. The experimental result shows that our method performed well whichshows promising potential of applying rhetorical relation in cluster-based text summarization.
关键词：Rhetorical Relations; Text Clustering; Extractive Text Summarization; Support Vector Machine;Probability Model; Markov Random Walk Model
