Abstract:
In recent times, social media has received great attention among the research communities towards the domain of sentiment analysis (SA). The proficient design of SA is needed to improve the service and product qualities for the marketing and financial schemes for increasing the company's profit and user satisfaction. Although several SA techniques are available in the literature, it is needed to further enhance the classification results of the user review which helps to comprehend the user reviews, thereby quality of the products can be improved. This study devises an effective SA and classification technique using grasshopper optimization algorithm (GOA) with bidirectional long short term memory (Bi-LSTM), named GOA-BiLSTM. The GOABiLSTM model involves word2vec based feature extraction process to derive a useful set of features. In addition, Bi-LSTM based classifier is applied to determine the optimal class label of the extracted features. Moreover, GOA is utilized for the hyperparameter optimization of the Bi-LSTM model. To ensure the better outcome of the GOA-BiLSTM model, an extensive set of simulations were carried out on four datasets. The simulation outcome verified the superiority of the GOA-BiLSTM model by accomplishing a higher accuracy of 99.57%, 99. 71%, 99. 06%, and 98.98% on the applied Canon, Nokia DVD, and iPod dataset respectively.
Page(s):
423-437
DOI:
DOI not available
Published:
Journal: Journal of Theoretical and Applied Information Technology, Volume: 100, Issue: 2, Year: 2022
Keywords:
deep learning
,
Sentiment analysis
,
Classification
,
BiLSTM
,
Parameter optimization