An Advanced Recommendation System by Combining Popularity-Based and User-Based Collaborative Filtering Using Machine Learning

Finding the similarity is one of the critical rules of a recommender system. Popularity-based filtering offers to generalize recommendations to every user based on its popularity. Recommendation system or recommender engine is one of the most discussed topics in machine learning for business. There are three types of the commonly used recommender system. They are Popularity-based, Content-based, and Collaborative filtering-based. In content-based, it recommends a similar item based on a particular item, where collaborative filtering recommends an item to a person based on other equivalent persons’ interests. This research work has developed an algorithm that combines Popularity-based and User-based collaborative filtering using machine learning. In this paperwork, we applied textual clustering to keep the similar item precisely. Simultaneously, we tried to optimize our recommendation search time complexity. To evaluate the approaches, we used the Deskdrop dataset [6], and in the final stage, we used some distance matrix-like Pearson correlation to find similar users. The main lacking in general user-based collaborative filtering is its laziness. We have reduced the time complexity by applying textual clustering and file hashing to implement the data through machine learning.