Advertisement Recommendation System Based On Context Aware In Mobile Environment

In recent years,with the rapid development of mobile intelligent terminal equipment,mobile internet industry has been booming.Mobile Internet advertising has attracted more and more attention,and led to many new business models.However,there are the problem of information overload in the mobile internet advertisement,because the user's data and the advertisement data is so huge.In the face of a large number of users,if the advertisers can not targeting the users accurately,only relying on a huge number of advertising,will cause the waste of the delivery cost;at the same time,it is difficult for users to find their own true needs of advertisement in mass data.By developing the personalized advertisement recommendation system for advertisers and users,the problem of information overload can be solved..This paper focuses on the mobile Internet advertisement recommendation.The main innovation is as follows:(1)This paper proposed an advertisement recommendation method based on user's browsing behavior.According to the feature of mobile internet advertisement market selling user's tags,the proposed method collects the content of web page browsed by the users for a period of time and gathers the user's advertisement clicks to establish the similarity model.Further,the rule of memory forgetting is used to optimize the user similarity model.According to the model personalized advertisements is recommended to the users.The experiments on a set of real data are conduct,and the time of model construction and the F-measure value of the results of advertisement recommendation are obtained.The results show that the proposed method has a good effect.(2)This paper proposed a new advertisement recommendation method based on Bayesian probability model.This method fuses user context information in mobile environment and integrates Bayesian probabilistic model into advertisement recommendations;meanwhile,in view of the large amount of user context information and complex types of information,this paper uses the information gain to pruning attributes to eliminate redundant information and improve the efficiency of Bayesian Model.The experiments on a set of real data are conducted and the time of model construction and the F-measure value of the results of advertisement recommendation are obtained.The results show that the proposed method has a better performance in time and recommendation effect.