Recency Based Product Recommendation System

In daily life, we need different types of products as we use them. So, recommending recent products using an online system is essential for the consumer. In this paper, we present a novel approach to product recommendation leveraging recency-based techniques to enhance recommendation accuracy in dynamic online environments. The core of the methodology involves calculating recency values to prioritize recent interactions, ensuring that the recommended products align more closely with the current interests and demands of users. The interaction matrix, which depicts the user-product interaction data, was broken down using Non-negative Matrix Factorization (NMF) in order to do this. Once the data has undergone extensive preprocessing to guarantee its quality and relevance—including data cleaning, normalization, and filtering—the interaction matrix was produced. The user matrix and the product matrix, which capture latent components that represent underlying user preferences and product attributes, are two lower-dimensional matrices that we developed by NMF. The recommendation algorithm was then modified to incorporate the recency score, which emphasized recent user interactions by altering the conventional scoring methodologies. The top 5 products were suggested for each user based on their recency-adjusted scores, which took into account their most recent preferences and buying activity. Recency-based criteria are found to be effective in product recommendation systems, as proven by the proposed method, which showed an improvement in recommendation relevance, especially in contexts where user preferences change quickly.