Research On Stock Personalized Recommendation Method

With the continuous development of China’s stock market，the number of theparticipants in the stock market has a explosive growth momentum. Judging fromthe structure of investors at the present stage, most are small and medium-sizedinvestors with weak professional knowledge. In order to correctly guide the valueorientation to of the investors and reduce the waste of resources about blindinvestment, it is needed to provide practical investment suggestions for small andmedium-sized investors. At present, there are mainly two kinds of methods aboutstock recommendation.They are online stock recommendation method based onstock comments and stock price forecasting model based on mathematical analysis.The two methods have their respective advantages.However, their defects are quiteobvious. The former method can not meet the personalized needs of investors aboutstock recommendation. The application process of latter method is quite complex.Therefore, the method is difficult to be mastered by investors. Based on the abovereasons, research on stock personalized recommendation method become s animportant topic at present.In recent years, electronic commerce is booming. Personalizedrecommendation method is widely used in the field of electronic commerce tocomplete the recommendation for target users.The stock can be viewed as a kind ofspecial commodity. Thus, using the core idea of personalized recommendationmethod to construct the stock personalized recommendation model is a feasibleidea.Based on the domestic and foreign research about recommendation method,this article extracted the core ideas of commodity personalized recommendationmethods.Through the improvement and innovation of the original recommendationmethods, this paper constructed the stock personalized recommendation model. Themodeling approach consists of two steps: firstly, constructing stock feature indexsystem and using fuzzy clustering method to carry out the investor classification,secondly, using graph theory and maximum information preservation te chnology toimprove the original recommendation methods, establishing the stock personalizedrecommendation model basing on fuzzy clustering method. After completing themodel, this paper used simulation method to establish the investor-stock ratingdatabase. Then, we applied programming method to realize the process of stockpersonalized recommendation. Finally, we compared this method with therecommendation approach based on collaborative filtering and randomrecommendation method in terms of the recommendation accuracy andrecommendation errors. We concluded that the stock personalized recommendation method had higher recommendation accuracy and it is a kind of high-performancereal-time online stock recommendation method.