摘要：This paper introduces a novel approach of automatic multi-document summarization based on text segmentation. The approach acquires concepts with a How-Net oriented tool and calculates the importance degree of sentences by means of employing the improved DotPlotting model and establishing a sentence-based vector space model (VSM). A conclusion is made according to the importance degree and similarity of sentences. Experimental results show that the performance on ROUGE of the approach put forward hereby is effective and significant.
