Investigating Entropy for Extractive Document Summarization

Automatic text summarization aims to cut down readers’ time and cognitive effort by reducing the content of a document without compromising on its essence. Ergo, informativeness is prime attribute summary generated an algorithm, selecting sentences that capture essence primary goal extractive summarization. In this paper, we employ Shannon’s entropy sentences. We Non-negative Matrix Factorization (NMF) reveal probability distributions for computing terms, topics, in latent space. present information theoretic interpretation computed entropy, which bedrock proposed E-Summ unsupervised method The algorithm systematically applies principle informative from important topics document. generic fast, hence amenable use documents real time. Furthermore, it domain-, collection-independent agnostic language Benefiting strictly positive NMF factor matrices, transparent explainable too. standard ROUGE toolkit performance evaluation four well known public data-sets. also perform quantitative assessment quality semantic similarity w.r.t original Our investigation reveals though using approach promises efficient, explainable, independent summarization, needs be bolstered match deep neural methods.