Reed drastically, people can make connections and share information with friends easier and timely. Many researchers have noticed this subtle change and devoted to recommend products through WOM (word-of-mouth) on micro-blogs. However seldom of them consider that the popular topics and interests appeared in micro-blogs are changed frequently over time since the information is spread rapidly. In this paper, a new method to recommend products based on the changing preferences of individuals is proposed. Several experiments are done to verify the evolution of personal preference changes and to compare the recommendations made by top popular subjects, other domain experts and the personal preferences. The experimental results show the proposed method can recommend products more effectively.