social network analysis (sna) and its implications for knowledge discovery in informal networks
DESCRIPTION
Social Network Analysis (SNA) and its implications for knowledge discovery in Informal Networks- Talk by Dr Jai Ganesh, SETLabs, Infosys at Search and Social Platforms tutorial, as part of Compute 2009, ACM BangaloreTRANSCRIPT
Social Network Analysis (SNA) and its implications for knowledge discovery in Informal Networks
Dr. Jai GaneshWeb 2.0 Research Lab
2 2
Overview
• Social Networks
• Social Network Analysis (SNA)
• SNA in Web 2.0 scenarios
• Why Invest in SNA
• Examples
– Example 1: Customer Service Operation
– Example 2: Organisational Network Analysis
– Example 3: Criminal Investigation
• Analysing Data
– Tools and Products
• Issues
• Conclusion
4 4
Web 2.0: Overview
• Web 2.0 is about harnessing the potential of the Internet
– In a more collaborative and peer-to-peer manner
– Users communicate and collaborate while at the same time contribute and participate
– Is shaping the way you work and interact with information on the web
– Mindset change towards collaborative participation
– Shifts the focus to the user of the information
– User can search, choose, consume and modify the relevant content
Web 2.0 refers to the adoption of open technologies and architectural frameworks to
facilitate participative computing
8 8
Social Network
• A social network is structure made of nodes (representing people or organizations)
– that are connected together by one or more interdependencies (representing values, ideas,
friendship, financial exchange, or trade)
• Represented as a social graph–based structure often very complex
• A web of trust exists in every social network
– nodes represent members of the web and edges represent the amount of trust among pairs
of acquaintances
• Rapid emergence and acceptance of online social networks
– Computer Mediated Social Spaces (LinkedIn, Orkut, Facebook, SecondLife, Myspace)
– Peer to Peer Networks (Bit Torrent, Napster, KaZaA, Fasttrack, Freenet)
– Agent based systems (Cite-U-Like)
– Online transactions (Amazon, eBay)
1111
Multitude of networks
University networks
ProfessionalNetworks
ResearchNetworks
Product -based Networks
State-wise Networks
Language Networks
GamingNetworksStudent
Networks
Supplier/BuyerNetworks
Lifestyle NetworksEntrepreneurship
networks
Software developer networks
FamilyNetworks
PoliticalNetworks
1212
Dimensions of Social Network formation
Dimensions Scenarios1 Space Physical, Virtual
2 Time Persistent, Campaign based
3 Theme Healthcare, Home, Gaming
4 Product/Commerce Wii, iPhone
5 Demographics State, Income, Race, Language
6 Life Cycle Teens, Adults, Middle Aged, Elderly
7 Customer Profile Single Parent, Single Professional, Separated professional, Retired Professional
9 Software/Tool based PC configurator, Mashups, Widgets
10 Enterprise Small Businesses, Mom & Pop stores
11 Entities Universities, Governments, Research Labs
1414
Social/Organizational Network Analysis
• Social Network Analysis (SNA) relates to mapping, understanding,
analyzing and measuring interactions across a network of people
– Social networks, both formal as well as informal can foster knowledge sharing
among participants
– This has interesting implications on enterprises wanting to leverage social
networks to draw insights and inferences on user preferences as well as user
participation in networks
– Using SNA, analysts can explore questions related to social networks such as
• Who are the members to watch?
• What are they saying?
• Where do they interact?
• Strength of interactions?
• Emergence of sub-groups?
• ----------
1515
Social/Organizational Network Analysis
• Social Network Analysis (SNA) is the mapping and measuring of
relationships and flows between people (Borgatti et al 2002)
• Organizational Network Analysis (ONA) applies SNA to interactions in an
organizational setting
• Focus on the persons involved
– i.e., the WHO question
1717
Key Question
• How do you derive value from Web 2.0 assets?– Direct
• Better Customer/Consumer Experience
• Leading to
– Increased Customer Base
– Increased Sales
– Less Direct• DATA from Web 2.0 assets as an ASSET
– Derived
• Better understanding of the customer
• Learning from the customer
– Customer driven innovation
– Examples: E-bay, Amazon
1818
SNA and Web 2.0
• Peer-to peer
– Peer-to peer network wherein collaboration and sharing are important activities
– Self managed collaboration as opposed to a central node-managed collaboration
– Wikis, blogs, video sharing etc.
• Collective Intelligence
– Lays emphasis on the large scale distributed Intelligence of the participants in
the network over central Intelligence
– User created, modified, updated content
– User tagging, reviews etc.
1919
Amazon Recommendations
• Keeps track of browsing history, past purchases, your ratings as well as
purchase by other users
• Include four types of ‘personalized’ recommendations– Social recommendation (What Do Customers Ultimately Buy After Viewing This Item?)
– Item recommendation (New for You)
– Package recommendation (Frequently Bought Together)
– ‘Others like you’ recommendation (Customers who bought …. also bought)
• Extensive customer reviews which include– 1- 5 star ratings
– Favorable vs. Critical reviews
– Detailed review comments
– Your rating of the review comments (Help other customers find the most helpful reviews )
– Comments on the review themselves
2121
Why invest in SNA
• User/customer generated information could provide key insights
which will aid decision making
• Insights into new products/services
• Informal listening board
• Influence customer decision making
• Social computing becoming popular
• Increasing role of communities
2323
What is the data required?
• Online Individual Identity
– Assumptions
• Real identity may be unavailable
• Contact channel is available
• Multiple personalities/avatars
possible
– Peer Evaluations
• Rating or “Respect” measures
• Message Data
– Sender
– Recipient (individual, group or online
location)
– Content is text (for now…)
• Message threads more valuable
– Ability to relate one message to another
– Chronology of messages
• Online conversations
– Captured as log files
• Defined User Roles
– Enable online community to create user
roles
– Map identity to user roles
• Uniform Time Stamps
– Chronology of all actions in the
community
2424
Why focus on the individual?
• Analyze past history of inputs
– Internal measure(s) of quality
– Community perspective(s) of quality
• Watch more closely their future inputs
– Presuming that
• Highly respected or individuals with high quality levels will provide higher
quality inputs or insights in future
• Interact directly with those individuals
– Make them part of the “internal” team
• Understand interactions between individuals in the network
2525
How to go about understanding the data?
• Unit of analysis
– “Message”
• Content sent from an
individual sender to a
recipient (individual or group)
– Message threads
• Identify concepts
– Categorizing messages
– Relate concepts and
individuals
• Identify individuals related to
concepts
– User Role
– User Status
• Links between individuals
– Sub-groups
• Links between concepts
– Locations on the network
2626
How to go about understanding the data? Contd…
• Link concept to source of the concept
• Determine reliability of
– Concept
– Source of the concept
– Through peer evaluation
• Discover issues of interest to the community
– As opposed to asking what we think is interesting
• Dynamic Analysis
– What has changed since the last time we looked?
2727
Tools and Products: Diagramming and Analysis
• Online Tools/Products– BuddyGraph/Social Network
Fragments (Experimental tool)
– Visible Path (Email)
– Metasight KMS (Email)
– ActiveNet/Illumio (Email +
Documents)
– ContentExchange
(Classification of user
generated content)
• Traditional SNA Tools– UCINet 6
– MOST + SNA
– Pajek (Diagramming tool)
• Others– CustomerConversation
– ZoomInfo
2828
Other Techniques
• Collaborative Filtering
– Recommendation Engines
• Text mining
– Identify concepts and key words
• Web usage mining
– Usage patterns
– Identify what an individual is reading
• Process Mining
– Identify what sequence of activities take place
3030
Effective Use of the Network
• 4 dimensions for effective use of a network (Cross, Parker and
Borgatti, 2002)
– Knowledge
• Knowing what someone knows
– Access
• Gaining timely access to that person
– Engagement
• Creating viable knowledge through cognitive engagement
– Safety
• Learning from a safe relationship
3131
Application Areas
• Customer Facing (External)
– “Customer Intelligent Enterprise”
• Employee Facing (Internal)
– Break down internal silos
– Increase points of contact
• Hybrid (Customers and Employees)
– Facilitate interaction
– Direct connection to customers with insight and ideas
3232
Processes and Avenues
• Create/provide online venues for interaction
• Identify key network members
• Proactive contact with key members
• Facilitate interaction
– Connect key members to internal units
– Seed conversations (?)
• Facilitate listening/learning
– Feedback vs. listening
3434
What about...
• Data Sources
– Ownership
– Access
• Boundaries
– Of the firm
– Of the network
• Privacy and Other Legal
Constraints
– Global network
– Local restrictions
• Processing Data
– Pre-processing Bias
– Formatting and storing Data
• Questions:
– When do I know I have
something interesting?
– When do I know that
something is no longer
interesting?
3636
Conclusion
• Web 2.0 environments
– Rich source of data
• Huge potential to tap the insights of the consumer base
• Organizational Network Analysis
– Focus on the Individual/Community
– Identify likely sources of interesting data
– Watch for what they say in future
• Application Areas: Listening to
– Consumers
– Employees