Building and Analyzing

Social Networks
Semantic Web and
Social Networks

Dr. Bhavani Thuraisingham

February 8, 2013
5/7/2018 03:36
23-2

Semantic Web: Chapter 1

0 Reference: P. Mika, Semantic Web and Social Networks,

Springer, 2008: Chapter 1
0 Limitations of the Current Web
0 The Semantic Solution
0 Developments of the Semantic Web
0 Emergence of the Social Web
5/7/2018 03:36
23-3

Limitations of the Current Web

0 Who is Frank van Harmelen?

0 Show me photos of Paris
0 Find new music that I might like
0 Tell me about music players with a capacity of at least 4GB
5/7/2018 03:36
23-4

The Semantic Solution

0 Apply knowledge technologies to fill the knowledge gap

between the human and the machine
- Provide personal information in semantic format
- Attach metadata – e.g., to images
- Provide background knowledge
- Aggregate information
0 Knowledge representation and reasoning
5/7/2018 03:36
23-5

Developments of the Semantic Web

0 Early developments include the WWW, Mosaic, HTML, XML

0 Semantic Web Technology Stack
- RDF, OWL
0 Reasoning with semantic web technologies and the
development of SWRL
0 Query languages and data management - SPARQL
5/7/2018 03:36
23-6

Emergence of the Social Web

0 Web Services
0 Blogs
0 Wikipedia
0 Online social networks
0 Web 2.0 + Semantic Web = Web 3.0
5/7/2018 03:36
23-7

Social Network Analysis: Chapter 2

0 Reference: P. Mika, Semantic Web and Social Networks,

Springer, 2008: Chapter 2
0 What is Network Analysis
0 Development of Social Network
0 Concepts and Measures in Network Analysis
5/7/2018 03:36
23-8

What is Network Analysis

0 Social Network Analysis is the study of social networks

among a set of actors
0 Focus is on the relationships between the actors and not on
the actors themselves
- Some relationships are more important than others
- Some actors are more important than other actors
0 Data collection and analysis
- Collect data and build a graph, analyze the graph
- Manual process consisted of filling questionnaire and
analyzing the data using statistical methods
- Automated methods: extracting nuggets from massive
amounts of data and building relationships
5/7/2018 03:36
23-9

Development of Social Networks

0 Social scientists influenced the field since the 1930s

0 Moreno’s concept of sociogram
- Sociogram visualized as a collection of nodes and links
0 WWW is a collection of nodes and links
- Links in the WWW represents the relations between two
web pages
5/7/2018 03:36
23-10

Concepts and Measures in Network Analysis

0 Networks represented as graphs
- Mathematical representations of a graph could be a
matrix.
- 1 represents a links between nodes Vi and Vj.
- 0 if there no link between Vi and Vj
0 Add weights to links
- Strength between Vi and Vj is 0.9, between Vi and Vk is
0.2
0 Observations
- People are separated by 6 steps
- Most people have about two coauthors while very few
have more than 20 coauthors
5/7/2018 03:36
23-11

Concepts and Measures in Network Analysis

0 Analysis
- Find in-degree and out-degree
- Find the hub
- Find the clusters
0 Questions to answer
- Important people in the network
- Who do people go to often
- Who has many relationships
- Which two have the strongest relationship
5/7/2018 03:36
23-12

Some Examples

0 This unit describes the relationship between Social Networks

and Semantic Web
0 FOAF
0 LINK (Peter Mika, Free University)
0 Extracting social networks from Semantic Web Data
(Tim Finin et al, UMBC, Jennifer Golbeck UMC)
0 Our Work
0 Convergence
5/7/2018 03:36
23-13

Semantic Social Networks

0 The latest breed of social networking services combine social networks
with the sharing of content such as bookmarks, documents, photos,
reviews.
0 The use of of Semantic Web technology facilitated distributed control.
- The friend-of-a-friend (FOAF) project is a first attempt at a formal,
machine processable representation of user profiles and friendship
networks. (Unlike with Friendster and similar sites that have central
control)
- FOAF profiles are created and controlled by the individual user and
shared in a distributed fashion.
- http://www.foaf-project.org.
5/7/2018 03:36
23-14

FOAF
0 The Friend of a Friend (FOAF) project is creating a Web of
machine-readable pages describing people, the links between
them and the things they create and do; it is a contribution to
the linked information system known as the Web.
0 FOAF defines an open, decentralized technology for
connecting social Web sites, and the people they describe.
0 FOAF is part of a shift towards a Web where we can choose
the sites and tools we like, without being cut off from friends
who made different choices.
0 FOAF lets you share and inter-connect information from
diverse sources, move it around, and use it in unexpected
new ways.

Sharif University of
Technology,
5/7/2018 03:36
23-15

FOAF Example
0 <foaf:Person rdf:about="#me“
xmlns:foaf="http://xmlns.com/foaf/0.1/">

<foaf:name>Dan Brickley</foaf:name>

<foaf:mbox_sha1sum>241021fb0e6289f92815fc210f9e9137262c252e<
/foaf:mbox_sha1sum>

<foaf:homepage rdf:resource="http://danbri.org/" /> <foaf:img

rdf:resource="/images/me.jpg" />

</foaf:Person>
5/7/2018 03:36
23-16

Semantic Social Networks

Semantic Web researchers and their connections across

the globe.
5/7/2018 03:36
23-17

Semantic Social Networks

Social
Network
of a
Semantic
Web
Researcher
5/7/2018 03:36
23-18

FLINK (Peter Mika, Free University)

0 Flink, the system developed at Free University 9The Netherlands) is one
of the early semantic social networks that exploits FOAF for the
purposes of social intelligence.
- social intelligence, is consdiered to be the semantics-based
integration and analysis of social knowledge extracted from
electronic sources under diverse ownership or control. In our case,
these sourcesFrom
0 Flink extracts knowledge about the social networks of the community
and consolidates what is learned using a common semantic
representation, namely the FOAF
5/7/2018 03:36
23-19

FLINK Architecture

Architecture
Of Flink
5/7/2018 03:36
23-20

FLINK Architecture
0 The architecture of Flink can be divided in three layers concerned with metadata
acquisition, storage and visualization
0 Acquisition layer of the system concerns the acquisition of metadata. (e.g., HTML pages
from the web, FOAF profiles from the Semantic Web, public collections of emails and
bibliographic data)
0 The web mining component of Flink employs a co-occurrence analysis technique The
web mining component also performs the additional task of finding topic interests, i.e.
associating researchers with certain areas of research.
0 The middle layer is responsible for storing and enhancing metadata through reasoning.
0 Inference is another major task of the middle layer. Sesame (we can also use JENA)
applies the RDF closure rules to the data at upload time. This feature can be extended
by defining domain-specific inference rules in Sesame’s custom rule language.
0 The third layer, is the browing and visualization layer,. The user interface of Flink is a
pure Java web application based on the Model-View-Controller (MVC) paradigm.
5/7/2018 03:36
23-21

Social Network Analysis on Semantic Web Data

0 Social network analysis tasks for Flink augments the web mining
task with finding which people belong to which groups (called
GROUP DETECTION)
0 The association and links between people including what is the
relationship between John and James? Are they just friends or do
they have a romantic relationship? Do they often travel together?
0 Semantic web reasoning tools (e.g., based on OWL, RDF and SWRL)
may be used to reason and extract the nuggets.
5/7/2018 03:36
23-22

Group Detection
0 A large community often breaks up to a set of closely knit groups of
individuals, woven together more loosely by the occasional
interaction across groups.
Based on this theory, SNA offers a number of clustering algorithms for
identifying communities based on network data. Alternatively, the
subgroups may be identified by the researcher using additional
attribute data on the
Peter Mika’s research uses an interactive clustering software
provided as a sample with the JUNG Java toolkit for SNA. This
software allows the user to cluster a network using an edge-
betweenness clusterand visualize the results.
As an example, a group of researchers from the AIFB Institute of the
University of Karlsruhe quickly emerge as a single cluster of the
network.
5/7/2018 03:36
23-23

Linking Social Networks with FOAF

0 One of the core goals of the Semantic Web is to store data in distributed
locations, and use ontologies and reasoning to aggregate it.
0 Social networking is a large movement on the web, and social networking
data using the Friend of a Friend (FOAF) vocabulary makes up a
significant portion of all data on the Semantic Web.
0 Many traditional web-based social networks share their members’
information in FOAF format.
0 While this is by far the largest source of FOAF online, there is no
information about whether the social network models from each network
overlap to create a larger unified social network model, or whether they
are simply isolated components.
0 Researchers at the U of MD have studied the intersection of FOAF data
found in many online social networks. Using the semantics of the FOAF
ontology and applying Semantic Web reasoning techniques, they show
that a significant percentage of profiles can be merged from multiple
networks.
5/7/2018 03:36
23-24

Extracting Social Networks

0 Extracting social network from noisy, real world data is a
challenging task, even if the information is already encoded in RDF
using well defined ontologies.
0 The process consists of three steps: discovering instances of
foaf:Person, merging information about unique individuals, and
linking person through various social relation properties such as
foaf:knows.
5/7/2018 03:36
23-25

Extracting Social Networks (Tim Finin)

0 A critical problem is determining whether two foaf:Person instances
denote the same person. The semantics of FOAF vocabulary
suggests several heuristics to answer this question:
- • named URI. Non-anonymous individuals using the same URI
denote the same person.
- • Inverse-functional properties. Inverse functional properties
such as foaf:mbox and foaf:homepage identify unique
individuals. Other properties, such as foaf:name and foaf:nick,
while not strictly inverse functional, can be used in practice in
conjunction with other properties like foaf:phone to identify
individuals with high probability.
- Semantic equality. When two or more values of an inverse
functional property co-exist in the same individual’s description,
they are semantically equivalent as identifying the same
individual.
0 \
5/7/2018 03:36
23-26
Our Work for IARPA Under KDD Program
(2007-2010)

B Applications to the Geospatial Domain

Other
L
Services
A Entity Extraction, Relationship Extraction Technology
C e.g., Insertion into
K
Ontology Matching and Alignment Security, HP’s Jena RDF
Integrity Data Manager
B
Rule based reasoning, Data mining FUTURE
O Contributions
WORK
O RDF Graph Store Management to BLACKBOOK
K

Storage, Transactions, Query, Integration

RDF Graph Store RDF Graph Store RDF Graph Store

5/7/2018 03:36

Military Stabilization and 23-27

Reconstruction Operations (SARO) EASY

Build schools SENSITIVE

Build wells Counter narcotics

Support Elections DIFFICULT
Build clinics
Support DDR Influence warlords
Mitigate conflict
Distribute medical supplies Support ANA Foster sustainable economy
TE ANP
and
P DA Promote stable democracy
U
Improve Human Building Self -Sufficiency
Rights
Build agriculture ‘ Enduring Security’
systems
Improve
Governance
Implement Job ‘ Rule of Law’
programs

‘ Reconstruction and Development

’

Figure 1-1. Stabilization and Reconstruction Operations

Duplicated from (Guttieri, 2007a)
5/7/2018 03:36

Military Stabilization and 23-28

Reconstruction Operations (SARO)

0 Four concurrent tasks have to carry out in parallel (NDU Study). They are the following:
0 (i) Security: Ensure that those who attempt to destroy the emergence of a new society
are suppressed. This will include identifying that are the trouble makers or terrorists and
destroy their capabilities.
0 (ii) Law and order: Military and police skills are combined to ensure that there are no
malicious efforts to disturb peace.
0 (iii) Repair infrastructure: Utilize the expertise of engineers and geographers both from
allied countries and local people and build the infrastructure.
0 (iv) Establish an interim government effectively: Understand the cultures of the local
people, their religious beliefs and their political connections and establish a government.
0 Dr. Karen Guttieri states that Human Terrain is a crucial aspect and we need hyperlinks
to People, Places, Things and Events to answer questions such as * Which people are
where? Where are their centers and boundaries? Who are their leaders Who is who in
the zoo? What are their issues and needs? What is the news and reporting? Essentially
the human domain associations builds relationships between the who, what, where,
when and why (5W)
5/7/2018 03:36
23-29

SARO Lifecycle (SAROL)

0 SAROL consists of three major phases
- (1) information and relationship discovery and acquisition,
- (2) information and relationship modeling and integration and
- (3) information and relationship exploitation.
0 During the discovery and acquisition phase commanders and key people will
discover the information and relationships based on those advertised as well
as those obtained through inference.
0 During the modeling and integration phase the information and the
relationship have to be modeled, additional information and relationships
inferred as well as the information and relationships integrated.
0 During the exploitation phase the commanders and those with authority will
exploit the information, make decisions and take effective actions
5/7/2018 03:36
23-30

Temporal Geo-Social Semantic Web

SPARQL for TGS-SW
TGS
-SW
-SW
Ontologies (OWL)
Ontologies (OWL)forfor
TSG-SW
TSG
-SW
-SW

TRDF+GRDF+SNRDF

TML+GML+SNML
TML+GML+SNML

URI, UNICODE

Figure 3-3. Temporal Geosocial

Semantic Web Technologies
5/7/2018 03:36
23-31

Incentives for Social Communication

0 Incentive based communication is a major component of the SARO system.
0 We are working on building mechanisms to give incentives to
individuals/organizations for information sharing and communication.
0 Once such mechanisms are built, we can use concepts from the theory of
contracts( by Laffont and Martifort) to determine appropriate rewards such as
ranking or, in the case of certain partners, monetary benefits.
0 Currently, we are exploring how to leverage secure distributed audit logs to
rank individual organizations between trustworthy partners.
0 To handle situations where it is not possible to carry out auditing, we are
developing game theoretic strategies for extracting information from the
partners. The impact of behavioral approaches to sharing are also currently
considered.
0 Finally we are conducting studies based on economic theories and integrate
relevant results into incentivized assured information sharing as well as
collaboration/communication.
5/7/2018 03:36
23-32

Convergence

0 Semantic web data includes databases, files, web logs, blogs,

emails, etc.
0 Data mining applied to semantic web data together with the
reasoning capabilities of semantic web result in social
networks
0 Data mining applied to social networks extract the nuggets
0 Nuggets together with additional semantic web data such as
ontologies result in knowledge
0 Knowledge utilized to improve the effectiveness of an
organization
5/7/2018 03:36
23-33

Convergence

Data Semantic Web

Management/ Data/Reasoning
Data Mining/ XML, RDF, OWL
Data Analytics e.g., databases
Blogs, email

Social
Networks/
Analysis

Knowledge/
Knowledge
Management
5/7/2018 03:36
23-34

Vision

0 Improved technologies for data representation

- Data will include structured and unstructured databases,
emails, blogs, files, relationships, video, images, audio,
tags, links, - - - - -
0 Improved tools for reasoning
0 Improved tools for data mining/data analytics
0 Improved tools for social network extraction
0 Improved tools for knowledge extraction
0 Improved tools for knowledge management
0 We call the above Information Analytics