Research On Collaborative Filtering Algorithm And Application On Icon Recommendation

The era of image reading has brought about the problem of information overload of icons.The recommendation system can effectively alleviate the problem of information overload and provide users with personalized recommendation services.The recommendation algorithm is the soul of the recommendation system,and the collaborative filtering algorithm is one of the most effective and classic recommendation algorithms.There are still user-item data sparse problem,cold start problem,and user interest change problem.Aiming at the above problems existing in the collaborative filtering algorithm and the need for the application of icon recommendation,this thesis studies the collaborative filtering algorithm and its application in the icon recommendation system.The research contents are as follows:1.Aiming at the problem of sparse users’ data and the failure to consider the change of user interests over time,an improved collaborative filtering algorithm based on user interest diffusion and temporal correlation(UIDFT)is proposed.The algorithm calculates the direct similarity and diffusion similarity of users’ interests,and obtains the comprehensive similarity of users’ interests through parameter adjustment,which effectively alleviates the problem of users’ data sparseness,at the same time,an improved time correlation function is introduced to calculate the similarity of users’ interests,improves the quality of recommendations.2.Aiming at the problem of sparse data and cold start of icon items,combined with the extraction technology of icon features,this thesis proposes a collaborative filtering recommendation algorithm based on item image features and item scoring time correlation(IFTCF).The similarity between the icons can be utilized to effectively alleviate the cold start problem of the icons.At the same time,the algorithm introduces the improved item time correlation function into the item similarity calculation for better icon recommendation.3.The "Hequ" icon recommendation system is designed and implemented.As the application scenario of UIDFT and IFTCF algorithm,the system provides users with personalized icon recommendation services,so that users can find interesting icons more quickly.