In recommender systems. the lack of interaction data between users and items tends to lead to the problem of data sparsity and cold starts. Recently. the interest modeling frameworks incorporating multi-modal features are widely used in recommendation algorithms. https://www.jmannino.com/top-offer-Simply-Southern-World-Full-Valentine-T-Shirt-top-buy/