Recommender systems are crucial in modern software, aiding users by suggesting items aligned with their preferences within a large amount of items. This paper proposes an item-based e-book recommendation model utilizing a Weighted KNN algorithm. The data set is subjected to cleaning, pre-processing and segmentation using cross-validation into 3 folds, and an item similarity matrix is created that is used in model evaluation and item recommendation. Weighted KNN is presented as an alternative to KNN-Means, KNN-Baseline, and KNN-ZScore for rating prediction. In this model, weights were adopted for the elements proportional to the number of evaluations for them, as elements that have a large number of evaluations have higher weights in addition to the similarity value used in the basic nearest neighbor algorithm. Testing on the book-crossing dataset reveals that the proposed model achieves the lowest error rates, particularly with the Pearson-Baseline methodology, showing a mean absolute error of 1.252 and a root mean square error of 1.650. With the cosine metric, the errors are 1.271 and 1.693 respectively. This study provides an overview of these methods and highlights the potential improvements of the Item-based Collaborative Filtering methodology through further research.