Item Selection Using K-Means and Cosine Similarity

Abstract

In today's digital world, recommender systems (RS) are crucial since they provide tailored suggestions depending on user preferences. In order to get beyond the constraints of RS, this chapter presents a revolutionary machine learning technique that uses cosine similarity, embeddings, and k-means clustering. The difficulties and solutions associated with using k-means clustering in RS are covered in the first part. Various approaches are investigated to provide an all-encompassing perspective on recommendation systems. The next part discusses using cosine similarity and embeddings to improve the quality of recommendations. High-dimensional data is made simpler by embeddings, and similarity is precisely measured using cosine similarity. Transparency is ensured by covering dataset selection, analysis, and solutions in this chapter. The system architecture is covered in the concluding section, emphasizing approaches. This chapter provides information about the development of RS.