Recommendation system for SmartMart-A virtual supermarket

Abstract

The current online shopping system does not provide the actual shopping experience to the users where they scroll through the web pages and select an item to purchase, which becomes monotonous soon and lacks the shopping experience. This paper describes the development of the SmartMart, a 3D shopping mart where the user can navigate around the mart with the help of a virtual cart and interact with the products as they would do in real life. Further, this paper explains the integration of the recommendation system with the SmartMart that provides efficient and seamless customer experience. Here, the customers are recommended with various products based on their previous purchase history. This gives a more personalized touch to each user adding on to their shopping experience. This paper describes the implementation of recommendation system using three types of algorithms namely item-based collaborative filtering, popularity model and user-based collaborative filtering. These algorithms are integrated with Unity SmartMart application and tested with the open source dataset. This dataset is organized in three ways-original data, data with a dummy field and normalized data. Distance metrics such as Pearson co-relation co-efficient, Jaccard similarity, cosine similarity metrics are used to compute the similarity among customers. This paper presents a comparative study of these different similarity metrics for user-based recommendation algorithm based on precision, recall, F-value and accuracy. From results obtained it can be seen that the Pearson coefficient metric gives the highest accuracy value (86.6%) but the F-value, mean precision and mean recall values are very less compared to Jaccard Similarity Metric. Hence, Jaccard similarity metric is preferred over Pearson co-relation co-efficient. The design and development of SmartMart along with the recommendation system adds a new dimension to both the existing online shopping system and provides better customer satisfaction.