In the current scenario of the COVID-19 pandemic, ensuring vaccination is a top priority. Our project aims to clarify people's attitudes toward vaccination against COVID-19. Various studies have already been conducted to measure mood associated with COVID-19 vaccines, but they all suffer from a major common flaw: poor mood classification accuracy. Our project aims to increase the accuracy of mood classification for COVID-19 vaccines compared to previous studies. In our project, we used ABSA (Aspect-Based Sentiment Analysis) and TF-IDF (Term Frequency-Inverse Document Frequency) models for sentiment classification. We also tested the classification accuracy using five traditional machine learning models: Random Forest, Naive Bayes, Support Vector Machines, Logistic Regression, and Ensemble Classification. For our project, we classify sentiment into three categories called positive, negative, and neutral. This in many ways distinguishes our project from the literature. First, according to the limited research we read, ABSA and TF-IDF were not used together. Second, most of the previous studies used the bag-of-words approach, an outdated model compared to ABSA. Finally, traditional machine learning models such as random forests and ensemble classification have never been used in ABSA and TF-IDF. This project provides higher accuracy than stated in our results as a mood classifier for COVID-19 vaccines, whereas previous studies have less than 67% accuracy.