Authors:
Lixin Fu
and
Jing Deng
Affiliation:
University of North Carolina at Greensboro, United States
Keyword(s):
Shortest Paths, Error Rate, Social Networks.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Data Mining
;
Databases and Information Systems Integration
;
Enterprise Information Systems
;
Sensor Networks
;
Signal Processing
;
Society, e-Business and e-Government
;
Soft Computing
;
Software Agents and Internet Computing
;
Web 2.0 and Social Networking Controls
;
Web Information Systems and Technologies
Abstract:
Scalable and efficient algorithms are needed to compute shortest paths between any pair of vertices in large
social graphs. In this work, we propose a novel ROBE scheme to estimate the shortest distances. ROBE is
based on a hub serving as the skeleton of the large graph. In order to stretch the hub into every corner in the
network, we first choose representative nodes with highest degrees that are at least two hops away from
each other. Then bridge nodes are selected to connect the representative nodes. Extension nodes are also
added to the hub to ensure that the originally connected parts in the large graph are not separated in the hub
graph. To improve performance, we compress the hub through chain collapsing, tentacle retracting, and
clique compression techniques. A query evaluation algorithm based on the compressed hub is given. We
compare our approach with other state-of-the-art techniques and evaluate their performance with respect to
miss rate, error rate, as well as construct
ion time through extensive simulations. ROBE is demonstrated to
be two orders faster and has more accurate estimations than two recent algorithms, allowing it to scale very
well in large social graphs.
(More)