A Discrete Differential Evolution Algorithm for Extractive Text Summarization

With the exponential growth of textual data on the web, automatic text summarization has become a key task in many natural language processing applications. Many works have been proposed to solve it, but the problem is still far from being totally solved. In this work, we propose a new extractive text summarizer that uses a discrete differential evolution algorithm. It looks for the best set of sentences to be put in the summary, in addition to the best order to be respected. To evaluate the performances of our approach, we used as benchmarks the DUC2001 and DUC2002 collections and opted for the ROUGE evaluation toolkit. The proposed approach is compared against eleven state-of-the-art methods. The obtained results show that our approach is very competitive. It outperforms many state-of-the-art methods; it ranks first out of 12 methods.