PRODUCT RECOMMENDATION SYSTEM USING COLLABORATIVE FILTERING TECHNIQUE

Abstract

Today, e-commerce websites host and offer a huge number of products with many different types and characteristics. E-commerce websites always want to develop the number of customers, diversify the types of products to meet the shopping needs of many types of customers, so the number of products and types of products displayed in the store increases by date and will limit the ability to communicate customers to choose products, customers have to browse through many links, filter a lot of information to find the desired product. So how to support customers in choosing shopping products? Specifically, what products should be suggested next to the products that have been reviewed or selected by the customer in the shopping cart? How many products should be recommended to the customer? The recommendation system was formed and developed not outside the purpose of limiting these weaknesses of e-commerce. The core part of the project is a collaborative filtering algorithm written in python language, so it is easy to embed into other sales website systems written in many different languages and techniques. The algorithm works by going through user reviews for the product, thereby finding other users similar to the current user and predicting the current user's rating for other products from those same users. The product recommendation system has been completed and deployed on the sales website. After testing, the system gives reliable suggestions. In the future, the system will be upgraded in terms of algorithms to improve predictability and can work well on platforms with large volumes of data.