Research on Singular Value Decomposition Recommendation Algorithm Based on Data Filling

Abstract

In the era of big data, the problem of information overload has become increasingly prominent. Recommendation systems are widely studied due to the problem. Due to the sparseness of data, the recommendation effect is not always ideal. To alleviate the problem of data sparsity, a singular value decomposition recommendation algorithm based on data filling is proposed. First, an improved Tanimoto similarity coefficient calculation method is proposed to calculate the similarity, and effective interpolation data is generated for the singular value decomposition model according to the proposed prediction formula. The experimental results show that when using the same dataset MovieLens100K, compared with several commonly used recommendation algorithms, the improved algorithm improves the prediction accuracy of the model, In the best case, RMSE is 10.1% lower than KNNBasic, 7.8% lower than Slope One algorithm, 6.9% lower than SVD algorithm, and 4.8% lower than SVD++ algorithm, verifying that this method can improve the recommendation quality.