S-SUM: A System for Summarizing the Summaries

Query-specific summarization of multiple documents is a useful task in the current day context of the WWW, that is containing huge amount of information. When different summarizers have access to different sets of documents for a query, generating a summary of the summaries produced by the multiple summarizers becomes an interesting and useful task. In this paper, we propose an efficient solution for this problem. Sentences from the individual summaries are used to construct an Integrated Linear Structure (ILS) and are given unique position numbers. All the sentences in the ILS are then assigned weights that reflect the importance of the sentences to the given query. Sentences are selected according to these weights using the Maximal Marginal Relevance (MMR) approach for inclusion into the final summary of summaries. Finally, the sentences in the final summary are sorted based on their position numbers given using ILS. Experimental results show that S-SUM is efficient.