Research on Singular Value Decomposition Recommendation Algorithm Based on Data Filling

2.1 Collaborative Filtering Recommendation Method

CF algorithm is divided into the model-based and the neighbor-based. The model-based CF refers to the SVD method (Al-Sabaawi, A. M. A., Karacan, H., & Yenice, Y. E., 2021). The neighbor-based refers to generating a recommendation list for a user based on the preferences of nearby users. Various similarity measurement techniques are used in the CF algorithm to calculate the similarity between items and between users. Most of these methods use co-scoring to calculate the similarity. One of the similarity measurement methods is Tanimoto similarity, which ignores the absolute value of the score and the average score of the user (Zhang, Qin. et al., 2019), and uses the ratio of the intersection and the union of the number of scores to measure the similarity. The neighbor-based CF predicts the score of the target object through the scores of other objects similar to the target object (Yuan, X. F. et al., 2019), which is a commonly used method for predicting missing values. The neighbor-based CF first calculates the similar users of the target user through the traditional similarity calculation formula, and then performs the score prediction of missing values.