The rapid growth of digital content, particularly in the form of electronic books, has presented new challenges for users in discovering reading materials that match their preferences. The problem becomes more complex due to the everchanging nature of user interests, which may evolve as users are influenced by trends, varying personal goals, or informational demands. Several previous studies have proposed approaches such as attribute-based filtering, personality-aware recommendations, and context-aware systems to improve recommendation quality. However, most of these methods do not explicitly model the temporal dimension, limiting their ability to anticipate changes in user preferences. This study proposes the utilization of the TimeSVD++ algorithm in a book recommender system as a solution to this issue. TimeSVD++ is a matrix factorization-based method that incorporates user and item temporal bias, as well as the contribution of implicit feedback. Evaluation was performed on the Amazon Book Reviews dataset, incorporating interaction weighting based on daily interaction distribution and normalized review scores. Experimental results show that TimeSVD++ outperforms the baseline SVD model, achieving a Precision@10 of 0.7035, Recall@10 of 0.8502, MAP of 0.8662, MRR of 0.9818, and NDCG@10 of 0.9940. These findings demonstrate that integrating temporal aspects leads to measurable improvements in the accuracy and relevance of book recommender systems when compared to the baseline SVD model.