Semantic Wiki In The EnterpriseJosef HolySocial Network DesignerSun Microsystems
WebSemantic Wiki In The EnterpriseJosef HolySocial Network DesignerSun Microsystems
Introducing Sun Context
Everyone and Everything Participating on the Network
Sun – Introduction• 30 000+ employees• Network is the computer• Everybody and everything participating
on the Network• HW, SW + Services – we have the whole
stack• Main asset: IP
We Are Open • The biggest Open Source contributor• Not just SW, but also HW (OpenSparc.net)• Utilization of community model is a must• Communities spanning over the actual
corporate boundaries• Strong notion of bottom-up innovation
We Use 'The 2.0ols'• Basic infrastructure for our communities:> Blogs - http://blogs.sun.com
– CEO blog: http://blogs.sun.com/jonathan> Wikis
– Internal: literally hundreds of internal wikis– External: http://wiki.sun.com
> Forums– http://java.sun.com
> “Code-forge”– http://www.kenai.com
> Social networks– Twitter, Facebook, LinkedIn...etc.
KIWI Scope: Customer Engineering Community (CEC)• 9000+ people geographically and
organizationally dispersed• Design, Deployment and Post-sale
support• Sub-groups relate to products, services
and practices• Running on Wiki since years > Project SunSpace> Low barrier for participation
What is SunSpace ?• Wiki-centric collaboration platform• Low barrier for participation• Built on top of Confluence Wiki> http://www.atlassian.com/confluence
• CEC's main community infrastructure> Wiki> File Store> Tags> Ratings, Comments> Social Graph> Value System
• Wiki-centric collaboration platform• Built on top of Confluence Wiki> http://www.atlassian.com/confluence
• CEC's main community infrastructure> Wiki> File Store> Tags> Ratings, Comments> Social Graph> Value System
What is SunSpace ?
25'000 users 400+ communities 10X growth in 6 month Secure Enterprise Wiki Architecture, Methodology, Value System
Who am I : MyProfile
What I need: Communities
Mashup: Personalized Search
KM Challenges• Low barrier for participation brings:> Missing notion of relevance and quality> Relevant knowledge retrieval and discovery> Expert discovery> Shared concept model maintenance
KM Challenges• Missing notion of relevance and quality• Relevant knowledge retrieval and
discovery• Expert discovery• Shared concept model maintenance
Missing Notion Of Relevance and Quality• What is the best white paper about topic
X?• Who is the main contributor in the
Community Y?• What is a current hot topic?• Who is an expert on topic Z?
Community Equity Introduction
Measuring Quality In Open Communities
Deriving value from user's activities
ValueSystem
Social Cloud Activities
Deriving value from user's activities - eBay
Ranking eBay Sell/Buy
Send/Pay/Rate
Top buyers/Top sellers
Deriving value from user's activities
DiggSystem
Digg Post/Read
Digg it!
Top stories/Top Diggers
Deriving value from user's activities – Community Equity
CEQContent-Centric Community
Create/Modify
Rate/Comment/...
Top Content/Top Authors/...
Information Equity Content Activities People
createmodifydownloadviewratecommenttag
Personal Equity my Community Equity Activities Me
ContributioncreatemodifytagParticipationviewratecommentreuse
People
Community EquityCONTRIBUTION
EQUITYAttachments
WikiBlogs
IP
SKILLS EQUITY
EnterpriseUser
Skills RatingTraining
PARTICIPATIONEQUITY
RateComment
Re-useTags
ROLE EQUITY
BusinessProjectsFormal
Informal
CQ+PQ+SQ+RQ
PersonalEquity
SunSpace is a start
CEQActivities
• Wiki-centric social collaboration platform• Traditional Web technologies• Very basic metadata handling (flat tags)
Now let's move further
CEQActivities
• Wiki-centric social collaboration platform• Built on top of semantic web
technologies• Much better metadata handling (triples)> Flat tags → Taxonomy → … → Ontology
KM Challenges• Missing notion of relevance and quality• Relevant knowledge retrieval and
discovery• Expert discovery• Shared concept model maintenance
Relevant knowledge retrieval and discovery• Current PULL techniques are not
sufficient> Full-text search
• A need to PUSH relevant content to users• A need for more structure –> better
metadata handling
Semantic Wiki – Social Semantic Platform
• Source of structure – interaction with existing tools and services• Structure created both by people and
machines• Linking to outside resources → new
relations• Social filtering – content with no social
actions is less relevant
Semantic Wiki - Source Of Structure
Corporate boundaries
Semantic Wiki - Integration Layer
Default Entry Page – Activity Feed• Activities related to content
relevant/interesting for the user> Semantic Graph → Reasoning →
Personalization• PUSH in a natural, non-intrusive way> “Check out this new white paper”> “Hey, new version of Java was just released.”> “Could you maybe help this guy?”
• Empowers participation> Beneficial for the Equity measures
Activity Feed Example: Facebook
KM Challenges• Missing notion of relevance and quality• Relevant knowledge retrieval and
discovery• Expert discovery• Shared concept model maintenance
Expert Discovery• Most of the knowledge is tacit, in people's
heads> → Finding the right people is more important
than finding documents• Who knows what?• Who is an expert on topic X?• Who could help me with this problem?• Who should I learn from?
Introducing Tag Equity:
Structure + Value (Tags) (CEQ)
Tag Equity (TEQ) – Content View
20 10 15 5
Docs tagged with “Java”
10
Aggregated TEQ value of tag “Java”50
Tag Equity (TEQ) – People View
20 10 15 5
People whoproduceddocs with
tag “Java” 10
Aggregated “Java” Expertise
50
TEQ → Concept Equity (SemEQ)• How does the Equity Value spread in the
semantic graph?• Example: SKOS – cumulatively:
Java
Software
broader+
50
70
20
broader+
TEQ → Concept Equity (SemEQ)• But what to do with other types of
relations?
Java
Software
broader+
50
70
20
broader+
built on???
KM Challenges• Missing notion of relevance and quality• Relevant knowledge retrieval and
discovery• Expert discovery• Shared concept model maintenance
Shared Concept Model Maintenance• How to manage folksonomies?> How to bring more structure into the (flat)
tag-space?> How to handle new tags invented by users?> Merge, delete, synonyms...etc.> How to give tags a more formal meaning?> How to relate tags to each other?
• Concept model should evolve with the domain> Open Communities need flexibility
Top-down Approach• “Let it be taken care of by experts who
know everything”> Dictatorship> Failed vision of the SemWeb from 2000
(“World-wide ontology”)> Too static → slow → inefficient
Bottom-up Approach• “Let the community do it in a wiki way!”> Anarchy – chaotic> Some level of control needed > There is no free market in the end> Even wikipedia has its “gardeners”
Solution: Open Concept Model Management (OCM)• Top-down + Bottom-up > Democracy – Elites (“elected” via CEQ)
managing the “folksonomy chaos”• Managing Concept Model in a Wiki way• Expertise-based access rights to the CM> Domains maintained by competent experts
• Sustainable concept model evolution> Opened to expertise emerging from bottom
Summary – Semantic Wiki In The Enterprise
• Semantic Wiki as a core of dynamic intranet> And of Semantic Web
• Source of Structure in Unstructured Environment• Need for a Value System> Merged with the semantic graph
Thanks for your attention!Josef [email protected]/holyj
Semantic Wiki Function(s)• Wiki - Document authoring tool> Users create structured content (documents)> They are shaping the knowledge base on the
instance level• Ontology manager> Creating new classes, defining properties... > Knowledge models need to dynamically
evolve• Source of structure, Integration
layer> Sustainable approach to data integration