Airline Twitter Data-Based Sentiment Analysis Using Improved Deep Neural Network

Abstract

Client feedback plays a major role in  extemporising the Quality of service and facilities in Airline Industries which are facilitated to the clients. Sentimental Analysis is greatly employed in airline industry exploiting customary feedback approaches which encompasses consumer loyalty surveys and structures. For sentiment analysis, tweets of clients from twitter are considered in this approach. Using this airline twitter data,the sentiment analysis is performed using Improved Deep Neural Networks (IDNN) in this paper. Initially, pre-processing approaches are applied to twitter data. The SentiWordnet is chiefly utilized for extraction of tweets positive and negative scores obtained from the pre-processed data. The IDNN classifier is primarily suggested for considering these extracted features as inputs. Pity Beetle Optimization Algorithm (PBOA) is another key element for optimizing the weight parameters  involved in the hidden layers and thereby achieving the improved performance. The classification of the input tweets  are given as positive, negative and neutral sentiment are obtained  through training and testing. Various performance metrics such as accuracy, precision, recall, and F-measure are evaluated for analysis of the suggested IDNN classifier and validated through simulation results