A hierarchical network embedding method based on network partitioning
by Xiankun Zhang; Yunbin Ma; Xuexiong Luo; Junlong Kang
International Journal of Computer Applications in Technology (IJCAT), Vol. 65, No. 1, 2021

Abstract: In order to analyse the real-world information network more effectively, we propose a hierarchical network embedding method based on network partitioning, NPHNE. NPHNE is a nested algorithm that can be combined with existing baseline algorithms to enhance their representation. NPHNE consists of two parts: graph abstracting and embedding propagation. The process of NPHNE is as follows: Firstly, modularity is used to pre-determine the network partition, the purpose is to constrain the maximum number of levels. Then, based on the hybrid collapsing method, a series of abstract graphs with successively smaller scales are constructed. The representation of the coarsest abstract graph is learned by the existing baseline algorithms. Finally, the representation is propagated and refined level by level from the coarsest abstract graph to the original graph. We evaluate NPHNE on multi-label classification tasks on citation network and social network. The experimental results demonstrate that the maximum performance gains over the baseline algorithm by up to 29.1% Macro-F1 score.

Online publication date: Mon, 15-Mar-2021

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Computer Applications in Technology (IJCAT):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com