A hybrid user-item-based collaborative filtering model for e-commerce recommendations

Abstract

The COVID-19 pandemic deepened understanding of e-commerce as an extremely promising sphere. Nowadays, even small businesses are widely using e-shops and e-markets. Thus, small and medium-sized e-commerce companies need powerful, flexible recommender systems, which do not require significant computing and financial resources. Currently, the main vector of such systems developing is targeted at hybridization, i.e. a combination of well-known effective methods. The paper proposes a hybrid model (bagging) to achieve high-quality e-commerce recommendations, which are based on the effective combination of collaborative filtering techniques. The model consists of the following components: User-based collaborative filtering (UBCF, classical variant and method involving text comments to compute the rating matrix), and Item-based collaborative filtering (IBCF) using the attributes of the predicted objects to calculate their similarity. The proposed model can become a methodological basis to introduce the recommender system for the medium-scale e-commerce platforms that are not able to afford deep learning. The recommender system based on this model does not require real-time updates and is easy for website integration. Besides, Root Mean Squared Error (RMSE) of the proposed method is significantly lower than in IBCF and UBCF models. Due to the significant improvement of recommendations accuracy, e-commerce companies will have a chance to positively affect customer loyalty without considerable investment.