Automated sentiment analysis in social media using Harris Hawks optimisation and deep learning techniques

Abstract

Sentiment analysis (SA) is a technique used in natural language processing (NLP), which determines the sentiment or emotion expressed in a piece of text. SA technique is frequently implemented on text datasets to assist in understanding customer needs, product sentiment in customer feedback, and businesses monitoring brands. Accordingly, the deep learning (DL) model is evolved as a promising technique that has been widely employed and attained considerable outcomes. Because DL models automatically extract features from the dataset, it can be claimed that intermediate representation extracted from the network is utilized as a relevant feature. Therefore, this article focuses on the design of automated sentiment analysis in social media using Harris Hawks Optimization with deep learning (ASASM-HHODL) model. The presented ASASM-HHODL model primarily processes the raw social media text into a useful format. Besides, the proposed ASASM-HHODL model uses fastText-based word embedding and skip-gram to examine the decrease of language processing dependency on data-pre-processing. Moreover, the attention-based bidirectional long-short-term model (ABiLSTM) is exploited for the classification of sentiments. Furthermore, the primary hyperparameters of the ABiLSTM model are adjusted by the use of the HHO algorithm to improve the classification performance. The performance validation of the proposed ASASM-HHODL model is carried out using a benchmark dataset and the results highlighted the promising performance over recent state of art approaches.