K Nearest Neighbour Collaborative Filtering for Expertise Recommendation Systems

With digitization the number of internet users are increasing and a wide variety of businesses based on the internet have emerged and modern ecommerce is one of them. In modern ecommerce there are a large number of products and a huge number of users accessing those to fulfill their needs. Searching for products is a time consuming task and a user may not have knowledge about every category of products listed. The recommendation system is important in the digital space and e-commerce because it suggests content to users based on their preferences. Modern recommendation systems based on machine learning models can more precisely recommend items to a user which they actually like. Collaborative filtering is one of them. It works by recommending items to a user by finding similar users preferences based on the predicting ratings of unknown items. In this work we have applied the k nearest neighbor (KNN) collaborative filtering algorithm on the Amazon Kindle Store Book review dataset which is the largest online e-book store on the internet. We have also proposed a modified version of this algorithm by considering expert users on the system to generate more precious recommendations for a user. Finally, We have evaluated our models using RMSE, MAE, hit rate and coverage and achieved outstanding results compared to the baseline algorithm.