The Research And Implement Of The Recommendation Algorithm Of User's Preference Based On The APP Of Mobile Terminal

With the development of Internet,people can enjoy various applications on it to meet their requirement,but on the other hand,the data on it increased explosively so that people can't find what they need immediately.Under this circumstance,people begin to research the recommender system to help people self to find their target quickly.But nowadays,the recommendation algorithm has some problem,it becomes more and more plump so that it costs much performance and it can't be used in the real business system.So the recommendation algorithm still has many to improve.In this paper,we study some main stream recommendation algorithms,and propose an advanced algorithm based on Random Walk,we also use it in a recommender system.Because of the requirement of the performance in the real industry,and the situation of many E-commerce has limited recommendation area(which is mostly no more than ten to fifteen),so this paper focus on the point to realize an algorithm that its amount of calculation is less and its accuracy in the top is better(also acceptable in the whole result).According to previous research,the Random Walk has a lower amount of calculation and its accuracy is good,so it's our first choice.This paper also takes user's history information into account,so that we can bias the result based on his context to make an more accuracy result,and it takes the correlation of genres in the item into account,the whole result becomes better.In addition,this paper will use the algorithm proposed in an application which is to recommend movies for users,in this case,we can show the effect of our research.The main content of this paper will be shown as follows:An advanced algorithm.This paper proposed an improved algorithm based on Random Walk and the correlation of item's genres,and it merges this two model with bias factor to make an accuracy and low amount of calculation algorithm.The dataset used in this paper is MovieLens,it divide this dataset into two parts-test set and train set to test the performance of this advanced algorithm,and make a compare with other algorithms.Application of movies recommendation.As the mobile internet has the highest use rate,and Android is the hottest system in mobile,this paper realizes an application based on it.The application include the modules as follows:login,main page(contains hot movies,high score movies and the recommended movies calculated),search,more,movie details and movie play modules.The process from recommend to play is successfully realized in the end.