Machine Learning based Sentiment Analysis using Graph Based Approach
Monali Bordoloi, Saroj Kr. Biswas · 2019
In the recent times, deriving high quality information from the ever increasing amount of text data is highly desirable in many application domains. Analysis and summarization of online review data is one such domain which demands an effective sentiment analysis technique for optimal decision making. This paper proposes an efficient sentiment analysis model by using a popular graph based method and some standard machine learning techniques in order to produce an overall sentiment using a limited and important set of keywords. A graph based technique using degree centrality measure is used specifically for extracting only the important and relevant keywords. The proposed model produces significant results which are validated with four different datasets and three different classifiers.