New graph-based text summarization method

The exponential growth of text data on the World Wide Web as well as on databases off line created a critical need for efficient text summarizers that significantly reduce its size while maintaining its integrity. In this paper, we present a new multigraph-based text summarizer method. This method is unique in that it produces a multi-edge-irregular-graph that represents words occurrence in the sentences of the target text. This graph is then converted into a symmetric matrix from which we can produce the ranking of sentences and hence obtain the summarized text using a threshold. To test our method performance, we compared our results with those from the most popular publicly available text summarization software using a corpus of 1000 samples from 6 different applications: health, literature, politics, religion, science and sports. The simulation results show that the proposed method produced better or comparable summaries in all cases. The proposed method is fast and can be implement for real time summarization.