Movie Recommendation Engine Based on Cosine Similarity and KNN

Abstract

Recommendation systems are becoming increasingly important as a means of managing information overload due to the proliferation of large volumes of material on the internet. In this paper, the hybrid method that uses the cosine similarity technique has been employed to find the similarity between the user preferences and the Netflix movie dataset. K-Nearest Neighbor method was emerged to identify films that closely correspond with the user's interests. This strategy aims to increase customer acceptability and utilization of the Netflix service by offering individualized recommendations to consumers based on their preferences. Although the examined papers presented performance for the model, they did not provide precise information about accuracy. The experimental results were evaluated using key metrics such as Accuracy, Precision at k, Mean Score, and Cross-Validation Scores. The empirical investigation demonstrates that the suggested approach gives customers precise recommendations based on their preferences where the accuracy of the proposed system scores around 80%.