A Degree-Based Method to Solve Cold-Start Problem in Network-Based Recommendation
Recommender systems have become increasingly essential in fields where mass personalization is highly valued. In this paper, we propose a model based on the analysis of the similarity between the new item and the object that the users have selected to sol
- PDF / 1,884,558 Bytes
- 8 Pages / 439.37 x 666.142 pts Page_size
- 38 Downloads / 193 Views
Abstract Recommender systems have become increasingly essential in fields where mass personalization is highly valued. In this paper, we propose a model based on the analysis of the similarity between the new item and the object that the users have selected to solve cold-start problem in network-based recommendation. In order to improve the accuracy of the model, we take the degree of the items that have been collected by the user into consideration. The experiments with MovieLens data set indicate substantial improvements of this model in overcoming the cold-start problem in network-based recommendation. Keywords Recommender systems Item degree Cold-start
Network-based filtering
Similarity
Introduction With the rapid development of the computer network and other mass media, the amount of information that we can get is growing exponentially [1]. We are facing so much information that we have to spend a lot of time and effort to obtain the Y. Liu F. Jia (&) School of Electronic and Information Engineering, Beijing Jiaotong University, Beijing 100044, China e-mail: [email protected] Y. Liu e-mail: [email protected] Y. Liu F. Jia Key Laboratory of Communication and Information Systems, Beijing Municipal Commission of Education, Beijing Jiaotong University, Beijing 100044, China W. Cao China Information Technology Security Evaluation Center, Beijing, China e-mail: [email protected]
Y.-M. Huang et al. (eds.), Advanced Technologies, Embedded and Multimedia for Human-centric Computing, Lecture Notes in Electrical Engineering 260, DOI: 10.1007/978-94-007-7262-5_102, Ó Springer Science+Business Media Dordrecht 2014
897
898
Y. Liu et al.
information that are most appropriate for us. Recommendation system is regarded as the most promising way to solve information overload problem. Recommendation systems recommend items of interest to users based on users’ own explicit and implicit preferences, the preferences of other users, the attributes of users, and the attributes of items [2]. For example, a book recommender might integrate explicit ratings data (e.g., Tom rates Introduction to Algorithms a 3 out of 5), implicit data (e.g., Tom purchased Algorithms in a Nutshell), user demographic information (e.g., Tom is male), and book content information (e.g., Introduction to Algorithms is marketed as a computer-related book) to make recommendations to specific users. As recommendation system has enormous significance to the development of economy and society, a wide variety of recommender algorithms have been proposed, such as collaborative filtering algorithm [3, 4], content-based filtering algorithm [5], spectral analysis, principle component analysis, networkbased algorithm [6–9], and so on. In recent years, the network-based recommender algorithm, which was proposed by Tao Zhou in Ref. [4], has been extensively investigated over the past several years and it tends to be one of the most successful technologies for recommendation system. Some physical dynamics, including heat conduction process and mass
Data Loading...
