Abstarct: The COVID-19 pandemic has significantly affected global health and social dynamics. This epidemic led to increased activity on social media. Users around the world express their emotions and concerns on platforms like Twitter. Analyzing these emotional reactions can provide valuable insights into the feelings and general behavior of individuals during crises. In this study, we focused on identifying emotions from tweets written by users during the pandemic. Studies on emotion extraction from text have been conducted using several methods: rule-baxsed methods, machine learning methods, and hybrid methods. In this study, we employed the first two methods to extract emotions. The most frequent words in tweets were utilized as part of the rule-baxsed method, while a cross-encoder was used to extract features, and Support Vector Machines (SVM) were applied to classify these features. To achieve this goal, we applied the cross-encoder model and support vector machines (SVM) to identify the emotions expressed in these tweets related to COVID-19. The cross-encoder was used to extract meaningful embeddings at the sentence level from the tweet text. Subsequently, SVM was used for classification baxsed on the extracted features. The results indicate that this approach achieves good performance in identifying emotions from users' tweets. It can also provide valuable information to health officials, policymakers, and researchers dealing with the social impact of this epidemic, allowing them to adopt more effective strategies to address the emotional and psychological needs of individuals in challenging times. We achieved accuracies of 0.48 and 0.46 for the first and second peaks, respectively, in our dataset.