Mobile and Web Recommender System for Shopping
Luís Miguel Couto Moreira · Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT) · 2015
Since the advent of digital point-of-sale systems in the early 90’s, there has been an ever increasing interest in using the recorded transactions to obtain valuable information and relationships contained in the data. Today this field is incorporated in the broad research fields of data mining and machine learning. With the transition from traditional brick and mortar retailers to e-commerce, this information has become easier to collect and explore. At the same time, a typical online store contains thousands of products in its catalog, with this huge inventory it is impossible for a customer to known every single product and from there make the best purchasing decisions according to his/hers tastes. Recommendation systems were created to solve this problem. They help customers by providing sensible product recommendations that the customer will appreciate, while providing to the stores potentially higher sales and lower costs related to product marketing. In this thesis, recommendation systems will be explored with an emphasis on aided product replacement and periodically bought products recommendations, a prototype recommendation system that provides these features was developed and the system was tested on real transaction data provided by a large retail chain. The system achieves encouraging results on traditional recommendations (used on aided product replacement) but the results of periodically bought products recommendations need further analysis.