Desai, Mittul
(2016)
COMBINED: COmmunity Mining By Inducing Node & Edge Data.
Masters thesis, Indian Institute of Technology Hyderabad.
Abstract
Community Detection is the process of identifying a group of nodes in a graph that are distinguish-
able in some context. Two sources of information have been studied in detail by the community
namely: Edge Structure and Node Attributes. Most work only deals with one of the above meth-
ods. However there is some work in the recent years that use them to complement each other. In
this paper, we aim to add a new dimension to the problem namely,
edge attributes
. In addition to
using the aforementioned methods we add edge attributes to detect communities. Edge attributes
uncover micro-communities that might not be easily retrievable using Node attributes and Edge
Structure. Especially in social networks, edge attributes might constitute a large part of the in-
formation content. This information along with contextual information of the edge helps uncover
previously undisclosed communities. It also helps uncover disjoint and overlapping communities.
Our approach uncovers the related attributes that form the community.
Actions (login required)
 |
View Item |
Altmetric
Altmetric