A review of techniques used in e-commerce recommendation system
Ke Yan · Applied and Computational Engineering · 2023
In the age of the internet, the explosive growth of information makes the e-commerce platform users need more and more requirements for the shopping experience. Recommendation systems are developed to retrieve information efficiently and provide a personalised shopping experience for users. There are three types of algorithms that are most frequently employed in recommendation systems: content-based filtering, collaborative filtering, and hybrid techniques. In this paper, content-based filtering and collaborative filtering are firstly explained in terms of working principles, limitations, and advantages. Then, some hybrid techniques that combine content-based filtering and collaborative filtering are proposed to overcome the problems of these two conventional methods. It is found that while collaborative filtering may struggle with scalability, sparsity, and cold start issues, content-based filtering may have quality issues and a lack of variation. The hybrid techniques are introduced to address the drawbacks of these two algorithms. It is, however, more complex and requires more memory resources to use hybrid techniques. Implementing deep learning in recommendation systems would be a promising area in the future. Through the analysis above, this paper would provide a systematic review of the development of the recommendation system and inform the future study. Students or researchers who are interested in big data can quickly grasp the concepts of techniques used in e-commerce recommendation systems as well as the benefits and drawbacks of different techniques.