Implementation of Hybridized Meta-Heuristic Model for Feature Selection in Sentiment Analysis
Alok Kumar Jena, K. Murali Gopal, Abinash Tripathy, Sachikanta Dash · Nanotechnology Perceptions · 2024
With the prevalence of social media and its ease of use, scrutinizing social media sentiment can provide effective direction to topics and products. Sentiment analysis is important for understanding the opinions of individuals published on platforms such as social media and product review blogs. Instead, sentiment analysis (SA) is gaining popularity and becoming a buzzword among researchers. Supervised machine learning is popular for its ability to simulate human behavior as an artificial intelligence. Various classification strategies have been developed for this purpose. ANN-based data classification has played an important role in classification. On the other hand, issues such as overfitting, pairwise classification, and parameter regularization affect classifier performance. To overcome this, a group of algorithms known as metaheuristic algorithms iteratively update candidate solutions and identify the best possible solution by maximizing the objective function. Genetic Algorithm (GA) and Firefly Algorithm (FA), which are used to optimize SVM and ANN parameters in this article, outperform SVM, ANN, and ANN. The IMDB film reviews the datasets created for the exper- iment. In order to properly analyze emotions, this study applied trajectory-based and population-based optimization methods and conventional machine learning methods, and found the relatively best results by comparing all the results.