Dynamic graph summarization: a tensor decomposition approach

• Type: Article in International Scientific Journal
• Date: 2018

Authors

Fernandes, S (Author) Other; Fanaee T, H (Author) Other; João Gama (Author) FEP

Journal

Title: Data Mining and Knowledge Discovery

Vol. 32

Pages: 1397-1420

ISSN: 1384-5810

Publisher: Springer Nature

Indexing

ISI Web of Knowledge - 10 Citations

Scopus - 11 Citations

Other information

Authenticus ID: P-00P-31F

DOI: 10.1007/s10618-018-0583-9

Abstract (EN): Due to the scale and complexity of todays' social networks, it becomes infeasible to mine them with traditional approaches. A possible solution to reduce such scale and complexity is to produce a compact (lossy) version of the network that represents its major properties. This task is known as graph summarization, which is the subject of this research. Our focus is on time-evolving graphs, a more complex scenario where the dynamics of the network also should be taken into account. We address this problem using tensor decomposition, which enables us to capture the multi-way structure of the time-evolving network. This property is unique and is impossible to obtain with other approaches such as matrix factorization. Experimental evaluation on five real world networks implies promising results demonstrating that tensor decomposition is quite useful for summarizing dynamic networks.

Language: English

Type (Professor's evaluation): Scientific

No. of pages: 24