Research and Application of Clothing Recommendation System Combining Explicit Data and Implicit Data

Apparel recommendation algorithms are widely used to solve the problem of clothing information overload and satisfying users' personalized wear requirements. The recommendation algorithm usually adopts a single data type. This paper proposes a recommendation algorithm that combines user explicit data with implicit data. Explicit data in this article uses users preference color, style data, implicit data uses user purchase records, browsing record data, calculated by two data types and feature vectors, obtain user-score high-dimensional matrix, PCA dimensionality reduction for high-dimensional matrix to obtain user scoring feature matrix use the optimized Pearson user similarity algorithm to get a list of recommendations after the Top-N sorting algorithm. The experimental results show that the F1 value of Explicit CF is 0.284, and the F1 value of Implicit CF is 0.265. The F1 value of the proposed algorithm proposed in this paper is 0.32