Annotation

1
Annotation Collaborative Work A Corporate
Approach
Alexandre Passant LaLICC / EDF RD Philippe
Laublet LaLICC François-Xavier Testard-Vaillant
EDF RD
2
Introduction

• Industrial context
• Integrating external information and common
  problems
• Objectives to reach
• Knowledge management architecture
• Retrieve and offer value-added information
• Allow people to react
• Build a corporate memory
• Future work

3
Industrial Context

• EDF Research Development
• More than 2000 engineers and researchers
• Various research domains, as renewable energies
• Economic and technologic development monitoring
• Lots of incoming information
• Technologies, patents, actors

4
Gathering and spreading external information

• A team of analysts (former librarians) per
  research domain
• Manual extraction from web sources
• Online newspapers
• Google queries
• Patents
• Select and resume pieces of information
• Send report via e-mail to engineers and
  researchers

5
Problems

• Analysts
• Lots of incoming information to browse
• Loss of time by manually selecting relevant
  information
• Various domains
• Maintain and organise the diffusion scope
• Engineers / researchers and company
• E-mail is the only way to react to news
• No public archive
• What about a new engineer joining the domain ?

6
Objectives

• Analysts
• More efficience in information retrieval
• Offering value-added information
• Engineer / researcher
• React to the information
• Debate about information
• Share knowledge
• Company
• Up-to-date corporate memory
• Easy access to information

7
How to reach our goals ?

• Texts and annotations
• Information as text metadata references
• Engineer / researcher annotations as  the expert
  voice 
• Social tools
• Incitate collaborative work
• Contribute to collaborative work of communities
  of practice
• Interlinks between web-based tools
• Semantic Web technologies
• A common description formalism
• Ontologies to drive knowledge description
• Ontology based annotations

8
Knowledge management architecture
9
Step 1 Efficiently retrieve and offer
value-added information

• RSS Aggregation Platform
• Advantages
• One central tool for analysts
• A public archive for engineers
• Plus
• A text-analyzer to remove unrelevant items
• An HTML 2 RSS converter
• Value-added information
• ?Annotations

10
Annotations into the aggregator

• Formal automated annotations
• Text-analyzer tags incoming items, eg  A new
  wave farm has been build in England coasts  gt
   Tidal_Energies   United_Kingdom 
• The aggregator also includes librarian
  annotations publication date, author
• Formal manual annotations
• Analysts give a pertinence mark to the content.
  Engineers will be able to subscribe news getting
  a minimal rank.
• Informal manual annotations
• Analysts can add some plain text informal
  annotations to the text

11
Step 2 Allow researchers to react to infomation
and to build a corporate memory

• Researchers want to
• Subscribe some domain-feeds
• Capitalize information
• React to pieces of news and debate
• Annotate documents
• Company needs
• An up-to-date corporate memory
• ?Social tools
• A personal RSS subscription service
• Blogs to debate or react to information
• A wiki as a corporate memory

12
From aggregation to subscription

• Feeds by category
• Internal / external
• Refresh frequency

13
Allow researchers to react to incoming information

• A private workspace to publish their own news
• Private / Public / Group access
• Everyone is allowed to comment others entries or
  to start a new topic regarding the first one
• Formal annotations DublinCore, keywords
• ? Weblog platform

14
A corporate memory

• Our needs
• Engineers and researchers can create pages
• Annotate pages
• Encourage and facilitate collaborative writing of
  articles
• Users can edit / create / annotate pages
• ? A wiki
• Wikis and blogs
• Blog for a personnal use, wiki to express the
  corporate view
• Debate on the blog, consensus on the wiki
• Relationship between these tools

15
Wiki annotation example
16
Adding structuration to wikis

• Problems with classical wikis in a corporate
  context
• Blank page syndrom
• Loss of efficiency in information retrieval
• ?Need to add some kind of structuration to our
  wiki
• Categories or tags
• Metadata
• Structured XML content
• Semantic metadata and ontologies

17
Our structured wiki prototype
18
Structured wiki and annotations

• Structured content
• Metadata extraction as Title, Company, Technology
  
• Plain text zones
• Free content
• Informal notes
• Librarians metadatas
• Author
• Date
• Topic

19
Knowledge management architecture with tools
20
Step 3 Easy access to the whole company
information flow

• Need a standard representation format
• RDF(S) ? OWL ?
• One common representation for all our tools
• One ontology per domain as a common vocabulary
• Global ontologies as FOAF or SIOC
• Inferencial capabilities

21
An domain ontology to facilitate knowledge sharing
22
Thank you for your attention
Questions / Discussion