Twitter sentiment analysis on online food services based on elephant herd optimization with hybrid deep learning technique

Twitter is a social media stage, making it a valuable resource for learning about people's opinions, feelings, and thoughts. For this reason, experts came up with methods to analyse the tone of tweets and determine whether they were favourable or negative. This article aims to assist businesses, and especially app-based meal delivery businesses, in conducting competitive research on social broadcasting and transforming social broadcasting data into data production for decision-makers. In this analysis, we compared Swiggy, Zomato, and UberEats. Customers' tweets about all these brands are obtained using R-Studio, and a deep learning-based sentiment examination approach is functional on the retrieved tweets. The pseudo-inverse learning autoencoder is able to provide feature extraction in the form of an analytic solution after pre-processing, without resorting to many iterations. In this research, we suggest framework for combining the Convolutional Neural Network (CNN) and Bi-directional Long Short Term Memory (Bi-LSTM) models. ConvBiLSTM is used, which is a word embedding model that uses numerical values to represent tweets. The CNN layer takes the feature implanting as input and outputs lower features. In this instance, elephant herd optimization is used to fine-tune the Bi-LSTM weights. Among the three firms, the results indicate that Zomato got the most positive feedback (29%), followed by Swiggy (26%), and UberEats (25%). Zomato also had fewer bad reviews than Swiggy and UberEats, with only 11% of users having a poor experience. In addition, tweets were evaluated for unfavourable views against all three meal delivery services, and suggestions for improvement were offered.