Turkish sentiment analysis for open and distance education systems

Abstract

Students’ opinions are the most essential source to enhance the quality of education and educational services in Open and Distance education (ODE) Systems. How to access and analyze students’ real opinions is a problem for ODE institutions. The purpose of the present study is to conduct a sentiment analysis (SA) on the collected Turkish tweets about an ODE system to monitor students’ opinions and sentiments about the system. Firstly, the related 63699 tweets about the ODE system are gathered and analyzed. Later, pre- processing is applied to the dataset. Sentence-based SA is performed with the data provided. The dataset is vectorized using two vector space models to test four classifiers which are Support Vector Machines, K-Nearest Neighbor, Logistic Regression (LR), and Artificial Neural Networks. F-score values obtained with these classifiers are evaluated, and the results are discussed. LR classifier gives the best F-score values with %75 for each vector space model. Through the SA results, students’ dissatisfaction, appreciation, and concerns will be learned quickly by the university administration to develop strategies that will increase the quality of education and educational services.