Context-Based Persian Multi-document Summarization (Global View)

Abstract

Multi-document summarization is the automatic extraction of information from multiple documents of the same topic. This paper proposes a new method, using LSA, for extracting the global context of a topic and removes sentence redundancy using SRL and WordNet semantic similarity for Persian language. In the previous approaches, the focus was on the sentence features (local view) as the main and basic unit of text. In this paper, the sentences are selected based on the main context hidden in the all documents of a topic. The experimental results show that our proposed method outperforms other Persian multi-document systems.