Modified firefly algorithm and different approaches for sentiment analysis

Abstract

The firefly algorithm is the tool of swarm intelligence in almost all areas of optimization and engineering practice. Firefly algorithm and its variants have been used to solve diverse problems of real-world and motivate new researchers and algorithm developers to use this simple algorithm for problem-solving. The population of the initial solution, firefly light intensity rank selection and position update are important parameters of the firefly algorithm. The optimization result will vary in the variation of type of function selection by the researchers and developers. Therefore, here an effort has been made to propose a hybrid firefly algorithm where a population of initial solution generation done by the differential evolution algorithm (DEA) and selection of light intensity and change in position of firefly position will be done by position updating techniques of PSO. The hybrid firefly algorithm has been tested on five benchmark data set with nine benchmark algorithms and experiment result showed that firefly outperformed the other used algorithm. The aim of this study is also to motivate researchers to apply this algorithm for sentiment analysis also.