Application of Distributed Seagull Optimization Improved Algorithm in Sentiment Tendency Prediction

Abstract

Emotion analysis is of great practical significance in the aspects of network control, public opinion monitoring and public sentiment guidance. In order to obtain better accuracy of emotion classification and analyze users’ emotion tendency more accurately, a distributed model of emotion classification with improved Seagull optimization algorithm (SOA) is proposed. The improvement of Cauchy variation and uniform distribution of SOA (CC-SOA) is to solve the problems of slow convergence speed, easy to fall into local optimal and poor accuracy of SOA. The uniform population distribution strategy can increase the diversity of the population and enhance the search ability of the local optimal solution of the algorithm. Cauchy variation is helpful to jump out of the local optimal solution and finally reach the global optimal solution. Due to the large amount of data to be processed and the long training time, single machine mode processing could not meet the actual requirements. Combining logistic regression with CC-SOA, a new model LG-SOA is proposed. Finally, LG-CCSOA is distributed processing on Spark platform, Distributed computing is carried out on different nodes, and the final running time is greatly reduced. After testing with benchmark function, the simulation results show that the CC-SOA has higher convergence accuracy and faster convergence speed. It has higher prediction accuracy for both small and large data sets, and the Spark platform improves the running efficiency of the algorithm.