Research Information System
International Academic Conference on Management, Economics and Marketing, Vienna, Austria, 7 - 08 July 2023, pp.78-85, (Full Text)
In this paper, we provide an in-depth analysis of various recommendation system algorithms used for movie datasets. With the surge in digital media consumption, personalized recommendations have become critical in enhancing user experience and engagement. We begin by reviewing traditional collaborative filtering techniques, content-based filtering, and hybrid methods, highlighting their strengths and weaknesses in dealing with scalability, sparsity, and cold start problems. In particular, we perform a comparative study of three recently proposed algorithms on a popular movie dataset, MovieLens. For the comparison purpose, we use two versions of this dataset: One with 943 users, the other one with 6040 users. We evaluate their performances based on the accuracy of the recommendations done by the model. The recommendation system studied in this article is shown to work with both small and large datasets. The numerical results of different models can be used in recommendation systems in smart city applications that are expected to work on datasets of different sizes.