Main Article Content
In the last few years text summarization has gained widespread attention across industries, especially in media and publications, research, business intelligence etc where it helps exploiting large documents to generate a new one with summarized inferences without losing the aspects. However, majority of the conventional approaches either employs extractive summarization or the abstractive summarization for single-document settings. On contrary, above stated application environments demand abstractive summarization in multiple document settings. Though, amongst the major efforts developed so far the attention based neural network methods have performed potentially; however their efficacy under multiple-documents setting and aspect-sensitive summarization has remained confined. Considering it as motive, in this research a novel and robust Improved Attention Layer assisted Recurrent Convolutional Neural Network (IA-RCNN) model is developed for Abstractive Text Summarization in multiple document settings. Unlike conventional efforts we have employed state-of-art techniques such as Sequence-to-Sequence (S2S) paradigm where the inclusion of RCNN, which is modified as Recurrent Neural Network (RNN) encoder technique for text summarization. Our proposed abstractive text summarization model encompasses semantic feature extraction, dependency parsing, semantic role labeling, semantic information etc where it exploits the structural, syntactic, and semantic information of the input text data to generate the summary. Unlike conventional Attention based Summarization, in our proposed model at first performs Clustering and Sentence Merging, which is followed by Transition-based Abstract Meaning Representation (TAMR) parsing, whose output is encoded by means of an improved Tree-LSTM RCNN model, which eventually generates single summarized sentence as output. The overall proposed model is tested with multiple text documents where simulation results affirm satisfactory performance for real-time applications.
Upon receipt of accepted manuscripts, authors will be invited to complete a copyright license to publish the paper. At least the corresponding author must send the copyright form signed for publication. It is a condition of publication that authors grant an exclusive licence to the the INFOCOMP Journal of Computer Science. This ensures that requests from third parties to reproduce articles are handled efficiently and consistently and will also allow the article to be as widely disseminated as possible. In assigning the copyright license, authors may use their own material in other publications and ensure that the INFOCOMP Journal of Computer Science is acknowledged as the original publication place.