Optimized Attention-Driven Bidirectional Convolutional Neural Network: Recurrent Neural Network for Facebook Sentiment Classification

Abstract

This paper devises an optimization-based technique for sentiment analysis using the set of reviews. The major processes involved for the developed sentiment analysis approach are tokenization and sentiment classification. Initially, the input reviews are considered from the database and are subjected to the tokenization process. The tokenization process is performed using Bidirectional Encoder Representations from Transformer (BERT) where the input review data is partitioned into individual words, named as tokens. Finally, sentiment classification is carried out using Attention-based Bidirectional CNN-RNN Deep Model (ABCDM), which is trained by proposed Chimp Deer Hunting Optimization (CDHO) approach. Accordingly, the proposed CDHO algorithm is newly designed by incorporating Chimp Optimization Algorithm (ChOA) and Deer Hunting Optimization Algorithm (DHOA). The proposed CDHO-based ABCDM provided enhanced performance with highest precision of 93.5%, recall of 94.5% and F-measure of 94%.