Research on the collaborative filtering recommendation algorithm in ubiquitous computing

It is very difficult for primary users to make up new policies by themselves. To deal with such situation, in this paper a fundamental framework is proposed to fully describe the generation process of policies in pervasive computing applications. Furthermore, the collaborative filtering algorithms based on cosine vector are utilized to calculate characteristic similarity and classic similarity to aggregate the user identity similarity. The machine learning algorithm is adopted to generate the policies which will be recommended to the users. By utilizing the recommended policies, the users can finish the system policies setting process in a more quick and accurate way.