P-CSREC: A New Approach for Personalized Cloud Service Recommendation

Zhang, Chengwen, Li, Zengcheng, Li, Tang, Han, Yunan, Wei, Cuicui, Cheng, Yongqiang and Peng, Yonghong (2018) P-CSREC: A New Approach for Personalized Cloud Service Recommendation. IEEE Access, 6. pp. 35946-35956. ISSN 2169-3536

[img]
Preview
PDF
08387824.pdf - Published Version
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (14MB) | Preview

Abstract

It is becoming a challenging issue for users to choose a satisfied service to fit their need due to
the rapid growing number of cloud services and the vast amount of service type varieties. This paper proposes
an effective cloud service recommendation approach, named personalized cloud service recommendation
(P-CSREC), based on the characterization of heterogeneous information network, the use of association rule
mining, and the modeling and clustering of user interests. First, a similarity measure is defined to improve the
average similarity (AvgSim) measure by the inclusion of the subjective evaluation of users’ interests. Based
on the improved AvgSim, a new model for measuring the user interest is established. Second, the traditional
K-Harmonic Means (KHM) clustering algorithm is improved by means of involving multi meta-paths to
avoid the convergence of local optimum. Then, a frequent pattern growth (FP-Growth) association rules
algorithm is proposed to address the issue and the limitation of traditional association rule algorithms to offer
personalization in recommendation. A new method to define a support value of nodes is developed using
the weight of user’s score. In addition, a multi-level FP-Tree is defined based on the multi-level association
rules theory to extract the relationship in higher level. Finally, a combined user interest with the improved
KHM clustering algorithm and the improved FP-Growth algorithm is provided to improve accuracy of cloud
services recommendation to target users. The experimental results demonstrated the effectiveness of the
proposed approach in improving the computational efficiency and recommendation accuracy

Item Type: Article
Subjects: Computing > Data Science
Computing > Artificial Intelligence
Divisions: Faculty of Technology
Depositing User: Yonghong Peng
Date Deposited: 20 Nov 2018 14:50
Last Modified: 28 Nov 2018 14:03
URI: http://sure.sunderland.ac.uk/id/eprint/10178

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year