Life Stage Based Recommendation in E-commerce

The rapid development of e-commerce has greatly changed the lifestyle of people. Nowadays, people are used to buying various kinds of things online, and recommender systems become more and more necessary since users are overwhelmed by a large amount of information. However, that a user's consuming behavior would change with his life stage has not been taken into consideration in most existing recommender systems. In this paper, we find the obvious correlation between temporal evolution and consuming behavior from a large amount of data. Motivated by this, we propose to obtain the relationship between items and user's life stage first. And then based on the relationship model, we can predict user's current stage according to his/her recent consuming behavior. Finally, we can recommend appropriate items to the user according to the prediction result of his/her life stage. The experimental results show that the introduction of temporal evolution of consuming behavior plays a significant part in improving the effectiveness of recommendation.