With the rapid development of the internet and the increase in the use of mobile devices, the use of social networks has increased in recent years. The fact that people want to share their personal ideas, opinions and suggestions with other people and to learn about other people's opinions and suggestions on a subject has made social media an important information store. Sentiment analysis is one of the most effective methods in the studies to be done with the data to be obtained from social networks. Sentiment analysis is based on understanding the emotion in all or part of a text with computer-based techniques. In this study, tweets collected from the Turkish twitter social network regarding distance learning during the pandemic period were divided into two classes as positive and negative. Datasets were created using these labeled data. For machine learning methods, experiments were carried out using Term Frequency-Inverse Document Frequency (TF-IDF) method. In the study, machine learning methods such as Random Forest, Linear Regression, Support Vector Machines, k-NN, Decision Trees, Gaussian Naive Bayes were compared with each other. Decision Trees give the highest performance with an average accuracy of 98% compared to other machine learning algorithms. In addition, Gaussian Naive Bayes showed the lowest performance with 67% accuracy. In this study, it was determined that machine learning techniques can be used to better understand how students feel in the transition to distance learning during the pandemic process.
Anahtar Kelimeler: Sentiment Analysis, Machine Learning, Twitter