Sentiment Analysis of Twitter Reviews using SVM, ANN and CNN Models

One of the most critical factors that an E-commerce considers when it comes to enhancing its customer experience is the availability of feedback. Through this study, we present an approach which can analyse the tweet characters to maximize the experience of the customers. The tweet features were mined using the word embedding and the n-gram approach. A classification model capable of identifying tweets in either positive or negative categories was proposed by SVM and ANN architectures. A CNN-based model was also developed to classify the tweets. The results of the study revealed that the model that was most accurate was the one that was used for analyzing the tweets. The results of the study revealed that the CNN-based model was more accurate than the SVM and ANN models when it came to analyzing the tweets. The results of the study also showed that the model was able to identify interesting associations between the various categories of tweets.