Research And Application On Collaborative Filtering Recommendation Algorithm In Mobile Intelligent Recommendation

By studying the target user's interest preference,Collaborative Filtering helps to find the nearest neighbors who have similar interests with them,recommend the nearest neighbors' content of interest which is widely used insocial network and electronic commerce.However,when traditional Collaborative Filtering Algorithm is applied to mobile intelligent recommendation,the recommendation faces new challenges of real-time and location-aware,etc.and this really reduces the efficiency and accuracy of mobile recommendation.To solve this problem,this paper brings context in Collaborative Filtering,and then proposes Collaborative Filtering Algorithms based on fusion context,design and implement a mobile food recommendation system.The main contents of this thesis are as follows:1.Aiming at challenges that Collaborative Filtering Algorithm cannot satisfy the requirements in mobile intelligent recommendation,we introduce context to Collaborative Filtering,furtherly divide it into strict context and loose context.At the same time we put forward the context similarity and portfolio similarity calculation method,collaborative filtering recommendation algorithm based on fusion of context was proposed,and the time complexity is analyzed.2.According to the characteristic of the algorithm,we design it into online and offline phase.Calculate item context similarity in data set at offline phase,to obtain loose context of the easing context under the context characteristics;Calculate the portfolio similarity at the same time,to obtain a association recommend list of item.Online stage according to the user's context,click or browse items and the results obtained by the offline phase processing,rapid access to recommend results.3.In order to verify the feasibility and effectiveness of the proposed algorithm,in the Node.Js development environment to fusion of context based on collaborative filtering recommendation algorithm,design and implement a mobile recommendation system.The recommendation system mainly has recommended module,search module and evaluation module,user context collection module and other modules.Using the mean absolute error(MAE),the root mean square error(RMSE),and other metrics,and the recommendation algorithm based on user similarity and context pre filtering recommendation algorithm were compared.The experimental results show that the proposed collaborative filtering recommendation algorithm on the contrast indicators are better than the other two classic recommendation algorithm,has a higher efficiency and user satisfaction.