Sentences Optimum Selection for Multi-Document Summarization

An approach for sentence optimum selection based on sub-topics of multi-documents is proposed. Multi-documents can be clustered into sub-topics after sentence similarity calculation, which can be sorted by scoring. Then sentences from all sub-topics are selected in order to get maximum coverage ratio of effective words. Using this method, the information redundancy of each sub-topic and among sub-topics is reduced. The information coverage ratio of the summarization is better improved. The experiment shows that the result is satisfied.