下載/瀏覽Download

Download Report

Transcript 下載/瀏覽Download

Computing in social
networks with relationship
algebra
Javed I. Khan, Sajid S. Shaikh
Internetworking and Media Communications Research Laboratories,
Department of Computer Science,
Kent State University
Journal of Network and Computer Applications
31 (2008) 862–878
Received 25 October 2006; received in revised form 26 February 2007; accepted 10 April 2007
Department of Computer Science and Information Engineering,Southem Taiwan University of Technology,M98G0215 鄭軒宗
Catalogue
 Introduction
 Relationship algebra
 Examples of social networks
 Schema graph of a publication network
- Application: reviewer selection
 Conclusions
Computing in social
networks with relationship
algebra
Introduction
Social network studies of the
past and present
Social networks
Introduce
Network
Members
User Base
Orkut
22,000,000
Designed specifically for friends and
family
LinkedIn
6,000,000
Designed for professionals and
adults
MySpace
54,000,000
Used mostly for fun and blogging
Sporzoo
2,000,000
Real Estate Investors and
Professionals
SelectedMinds
1,000,000
Corporate social networking
5
Cross websites relationship
Computing in social
networks with relationship
algebra
Relationship algebra
Relationship algebra
We define a set of operations on relation matrices. If A is a
relationship matrix then, we define the following semantic
operators:
Computing in social
networks with relationship
algebra
Examples of social networks
Schema graph of a publication
network
The major entities involved in the publication network are the
authors, the organizations, the paper, the journal, the reviewers,
the editors and the topic area.
Application: reviewer selection
The network can be used to select a set of authors who can be
on the reviewing committee of a paper such that they meet
certain restrictions. The reviewer selection can be expressed by
a set of constraints. Below is an example set:
1. The reviewer should not be a coauthor of the paper.
2. He should not be a coworker of the author.
3. The reviewer should not have submitted a paper in
the same journal or conference.
4. He should be well acquainted with the subject area
being discussed in the paper.
Application: reviewer selection
The aim of the example is to find a reviewer set for
papers P5 and P6 from among the four authors available.
Reviewer selection - Step 1
The first step is to determine the authors and coauthors for P5
and P6.
Reviewer selection - Step 2
Step2 we determine, which authors have submitted papers in
the same journal. In order to determine this, we first need to
establish a relationship matrix depicting the relationship
between the authors and the journals.
Note: (SC) Secure Communication, (WC) Wireless Communication
Reviewer selection - Step 3
The coworker matrix McoWorker is computed by multiplying
the matrix representing the authors-organization relationships
MA-org A and its transpose.
Reviewer selection - Step 4
Step4 we determine the non conflict of interest matrix
MnonConflict by subtracting each of the coAuthor, coJournal and
coWorker matrices from the matrix depicting the relationship
between all the authors belonging to the same topic area Mall.
Reviewer selection - Step 5
The reviewer set matrix Mreviewer is calculated by multiplying
the MnonConflict and the [Ma-p] T matrices. Applying the row
extraction set operation r on the reviewer set matrix gives us the
reviewer set for papers P5 and P6.
Conclusions
Over the last couple of years, a number of social networking
websites have become the cornerstones for social interactions.
The information being shared by individuals on these websites
is becoming richer by the day. In this paper, we have presented
a framework,which can be applied to these social network and
the information thus obtained can be used to construct a whole
new range of applications.