Book classification is a critical task in libraries and bookstores to enhance the accessibility and management of collections. Traditional classification methods often suffer from inefficiencies and inconsistencies due to human subjectivity. This study explores machine learning techniques as an alternative, focusing on Support Vector Machines (SVM) and Long Short Term Memory (LSTM) models for book genre classification. SVM, particularly with the Radial Basis Function (RBF) kernel, demonstrates superior performance, achieving the highest accuracy of 68.61% and an F1-score of 69.59%. In contrast, LSTM models, enhanced through K-Fold Cross Validation, show competitive results, with an accuracy of 65.59% and an F1-score of 67.07%. These findings emphasize the strengths of SVM in precision-critical tasks and the ability of LSTM to capture sequential text patterns, making both models viable solutions for efficient and reliable classification. This research not only addresses the challenges in book classification but also highlights the broader applicability of SVM as a robust method across multidisciplinary fields.