A Hybrid Recommendation Architecture for Nigerian Online Stores

International audience Online retailing, a business activity in e-commerce, is the process of selling physical goods on the internet while online shopping is the process of buying and selling goods and services over the internet. Item recommendations, however, is a business strategy in e-commerce employed in promoting goods sold on an online store. Majority of the online stores in Nigeria have their shopping systems implemented similar to online shopping systems used Developed Countries. Much focus have been placed on the provision of non-personalized recommendations in Nigerian online stores, as ratings information needed for personalized recommendations is sparse. These systems are mostly a hybrid of content-based and collaborative filtering approaches to recommendation generation. The use of content-based, collaborative and demographic hybrids have not been fully explored and implemented. This paper, however, proposes a hybrid item recommendation architecture that combines the content-based, collaborative and demographic filtering approaches in a mixed hybridization strategy for provision of recommendations on online stores. The architecture provides collaborative recommendations using the vector similarity and adjusted cosine similarity measure. The proposed system will go a long way in providing adequate item recommendation in Nigerian stores.