Authors : Abdul Ahad, Suresh Babu Yalavarthi and Ali Hussain
Abstract: Sentiment analysis comes under study within natural language processing. It helps in finding the sentiment or opinion hidden within a text. This research focuses on finding sentiments for twitter data as it is more challenging due to its unstructured nature, limited size, use of slangs, misspells abbreviations, etc. Most of the researchers dealt with various machine learning approaches of sentiment analysis and compare their results but using various machine learning approaches in combination have been underexplored in the literature. This research has found that various machine learning approaches in a hybrid manner gives better result as compared to using these approaches in isolation. Moreover, as the Tweets are very raw in nature, this research makes use of various preprocessing steps, so that, we get useful data for input in machine learning classifiers. This research basically focuses on two machine learning algorithms K-Nearest Neighbours (KNN) and Support Vector Machines (SVM) in a hybrid manner. The analytical observation is obtained in terms of classification accuracy and F-measure for each sentiment class and their average. The evaluation analysis shows that the proposed hybrid approach is better both in terms of accuracy and F-measure as compared to individual classifiers.
Abdul Ahad, Suresh Babu Yalavarthi and Ali Hussain, 2018. Multi-Level Tweets Classification and Mining using Machine Learning Approach. Journal of Engineering and Applied Sciences, 13: 3907-3915.