Research On Hybrid Recommendation Algorithm Based On User Geographic Information

Recommended system to provide users with personalized services,acting on the various sites is growing.Whether it is a movie site or e-commerce site,with the increase in the number of projects,the user's choice is becoming increasingly difficult,the last recommendation algorithm has slowly suited to this era,we need to know more about the user's habits and interests orientation to recommend more appropriate items for the user.At present,Chinese rapid development of mobile devices,Mi,Huawei and other companies can offer affordable,powerful mobile devices for users,such as smart phones,tablet PCs,smart watches.The popularity of these devices allows us to know more about the user's buying habits.Catch location information in the past was a problem,Not only need professional equipment,and precision was unsatisfactory.However,with the popularity of mobile devices,the user is no longer required to obtain position information of the GPS device expensive,also with the development of indoor positioning technologies,subscriber location information when shopping can be more accurate.In this paper,the user's location information as an element added to the user's personalized recommendation system,the habits of the user online and offline habits combine to provide a more accurate recommendation results.This paper introduces the development status and a variety of personalized recommendation algorithm for personalized recommendation algorithm structure as well as the advantages and disadvantages of information,and then introduced the new indoor positioning technology,and horizontal comparison.Recommendation algorithm based on user geographic information is the user's location information and user rating information in this article weighted composite,get to the target user similar neighbor user,and then to provide a recommendation to the target user via neighbor users.Location information of the user,there are two,one is the user's attendance information,the other is the residence time of the user information in the shop.The algorithm in Java development platform,using the public comment 3315 users to 499 stores user ratings and attendance information proved by experiments,after the addition of location information of the user,the user's ratings have significantly reduced the prediction error and the recommended accurate It has increased significantly and recall.Finally,the user interface and database designmicro site,used to obtain the user information in the residence time within the shop.