Semantic Web-Based Small Sample Data Recommendation Algorithm Using Weighted Mutual Information

In today's digital landscape, recommendation systems are vital for personalized content delivery across e-commerce, social media, and online education. This paper presents a Semantic Web-Based Small Sample Data Recommendation Algorithm Using Weighted Mutual Information, addressing significant challenges posed by data sparsity in conventional recommendation methods. Traditional algorithms often struggle with limited user behavior and item data, resulting in poor recommendations. Our approach utilizes a weighted mutual information similarity algorithm with a smoothing coefficient to better capture relationships among user actions, product features, and contextual tags. Experimental results reveal that the proposed method significantly outperforms existing algorithms, achieving a hit rate exceeding 72% and notable improvements in recommendation coverage. This research not only enhances the effectiveness of recommendation systems in sparse data contexts but also offers valuable insights for future advancements in personalized information retrieval.