Title: Hybrid optimisation enabled deep learning for sentiment rating prediction towards text summarisation and question answering system

Authors: Parsi Kalpana; Pelluri V. Sudha

Addresses: Department of Computer Science and Engineering, Vasavi College of Engineering, Hyderabad, India ' Department of Computer Science and Engineering, University College of Engineering, Osmania University, Hyderabad, India

Abstract: This work presents a novel question answering (QA) system using deep learning (DL) networks for generating answers to the questions based on review data. Here, QA is performed based on sentiment prediction, and extractive summarisation, wherein hierarchical deep learning for text (HDLTex) is used for sentiment prediction, and deep convolutional neural network (DeepCNN) is employed for generating an extractive summary. Further, a novel optimisation technique named the child drawing African vulture optimisation algorithm (CDAVOA) is devised for training HDLTex and DeepCNN. Further, QA is carried out using the RMDL based on the predicted sentiment and the extractive summary generated. Additionally, the RMDL_QA is examined for its competence based on quantitative measures, like accuracy, F-measure, recall, and precision, and is found to attain values of 0.907, 0.942, 0.958, and 0.926, respectively.

Keywords: random multimodel deep learning; aspect term extraction; hierarchical deep learning; BERT tokenisation; deep convolutional neural network; DeepCNN; hierarchical deep learning for text; HDLTex.

DOI: 10.1504/IJAHUC.2023.135111

International Journal of Ad Hoc and Ubiquitous Computing, 2023 Vol.44 No.4, pp.240 - 257

Received: 06 Apr 2023
Accepted: 21 Jun 2023
Published online: 30 Nov 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article