1
Web Mining Techniques for
On-line Social Network Analysis
I-Hsien Ting
Department of Information Management National University of Kaohsiung, Taiwan
iting@nuk.edu.tw
Outline
1. Introduction
2. Literature Review
Social Networks Analysis Web Mining
3. Web Mining Technique for On-line Social Networks Analysis
4. The Process to use Web Mining Technique for On-line Social Networks Analysis
3
Introduction
A Social Network is usually formed and
constructed
by daily and continuous communication Between people and includes different
relationships
• Such as the position, betweeness an clossness among individuals and groups.
Social Network Analysis is essential for understanding these relationships.
Introduction
On-line Social Networking has become a very popular application in the age of web 2.0
Allowing users to communicate, interact and share in the WWW
Some Popular On-line Social Networking Websites
Facebook MySpace Flickr Blogs……
Huge Resources of Interpersonal Communications, Relationships and Behaviours
5
Social Network Analysis
The history of SNA is older than everybody who is here
More than 100 years (Cooley 1909, Durkheim 1893)
The main task of SNA is usually about how to extract social networks from different communication
resources
Web, email communication, Internet relay chats,
telephone communications, organization and business events, etc.
Social Network Analysis
An example: E-mail Communication
the relationship between email senders and
receivers can be transformed
by measuring the frequency of email communication
taking the communication behavior (such as reply, forward, etc.) into account
7
Social Network Analysis
Visualization
A hot topic and suitable technique in this area
Network structure, nodes distribution, links
Social Network Analysis
Other measurements Centrality Degree • Betweenness • Closeness Clustering Coefficient Density Path Length Reachability Structural Hole9
Web Mining
The main analysis targets of on-line SNA is from the WWW
Web mining is claimed the most suitable technique
Three different types of web mining
Web content miningWeb usage mining
Web Mining Techniques
Classification & Clustering
30% of users browse product/food during the hours 8:00-10:00 PM
Association Rules
the people who view web page index.htm and also view product.htm the support=50% and the confidence=60%
Visualization
illustrating the structure of hypertexts and links in a website or the linking relationship between websites
11
The Three Web Mining Types for
On-line SNA
Web content mining
categorize or classify documents on an on-line social networking website, especially articles on blogs or text forums
Article categorization is usually the first task for many on-line social networks analyses or
The Three Web Mining Types for
On-line SNA
Web usage mining
useful for the analysis of social networks extraction
measuring centrality degree or closeness An example:
13
The Three Web Mining Types for
On-line SNA
Web structure mining
path length, reachability or to find structural holes, which are very basic and traditional social
networks analyses.
Uses graphs and visualized means to represent the data about social networks, enabling the analyst to easily understand and analyze social networks
Web Mining Techniques for On-line
SNA
Clustering
can be used with a large social network in identifying more groups and clusters
it can provide more detailed information than just using visualization
• including the closeness of a group, detailed information of the members in a group and the relationship
15
Web Mining Techniques for On-line
SNA
Association Rules
Can help to discover the hidden relationships
between nodes in a social network or even cross networks
• Person A knows person B and also knows person C, the support is 0.9 and the confidence is 0.5
• The person who read person A’s blog article and also read person B’s blog article, the support is 0.9 and the confidence is 0.5
17
Discussion and Future Research
Challenges
Data Sampling
• it becomes a difficult task to select suitable samples representative of the real social networks
how to combine different types of web mining techniques for a particular piece of on-line SNA a fine tuning of the process that proposed in the
Thank You!
Any Questions?
I-Hsien Ting
Department of Information Management National University of Kaohsiung, Taiwan