EnerSearch Case Study: VE Prototype D28 3rd Review of OnToKnowledge on Milestone III June 21, 2002 A

1
EnerSearchCase Study VE Prototype (D28)3rd
Review of OnToKnowledge on Milestone III June 21,
2002AmsterdamVictor Iosif
2

EnerSearch and its shareholders

• EnerSearch is a virtual organisation researching
  new IT-based business strategies and customer
  services in deregulated energy markets.
• Shareholders Major EU utilities and IT suppliers
  
• Focus understand opportunities of advanced IT
  for deregulated utility industry
• Goal knowledge creation and distribution for
  Shareholders/Researchers and Others
• KM is therefore key!

3
EnerSearch Current Shareholders

• ABB (Sweden/Switzerland)
• E.On (Germany)
• ECN (Netherlands)
• Electricidade de Portugal
• Iberdrola (Spain)
• IBM (Germany/Sweden)
• Sydkraft (Sweden)

4
Exploitation

• Work with and knowledge transfer to shareholders
  in this area is an ongoing process
• Board meetings, on-site executive workshops,
  projects
• Ontology-based tools will improve the
  effectiveness and efficiency of this key activity
  
• EnS website is major resource (has even brought
  in new shareholders) and will be significantly
  improved according to OTK lessons learned

5
Motivation for Using the Ontology Based
Tools

• The tools are developed in the project
• Users can start simple query consisting of only
  small number of search terms in order to get a
  picture on what kind on information that is in
  the database
• It present the information according to the
  inherent
• structuring that is offered by the ontology
• The search object is surrounded by related
  objects
• Generalisation and specialisation are offered to
  the user

6
Users Tools

• Three different interest groups
• Shareholders from four different companies (15)
• Researchers from Ronneby Software Center (15)
• Outsiders ( 15-20 )
• Three different tools
• Free text search tool gt EnerSEARCHer
• Ontology based search tool gt QuizRDF
• Ontology based navig. tool gt Spectacle

7
Experiment Procedure

• Compare the EnerSEARCHer with the performance of
  subjects using the QuizRDF Spectacle tool
• Each subject is confronted with different
  questions related to the EnerSearch web
• Equal number of questions for each tool to be
  tested in different task order - testing all
  three tools

8
Experiment Design

• Hypotheses
• H1 Users will be able to complete search task in
  less time
• H2 Users will make fewer mistakes..
• H3 The reduction in completion and number of
  mistakes will be more noticeable for less
  experienced users.
• H4 The reduction in time will also be more
  noticeable for users lacking detailed knowledge
  of the knowledge based system implementation.
• H5 The ontology based search tool will be viewed
  as more useful for a broad range of domains and
  knowledge acquisition scenarios by different type
  of persons than using free text search.
• H6 The effort developing and maintaining the
  ontology and information structure will not
  significantly exceed the effort to develop and
  maintaining the free text approach.

9
Place in On-To-Knowledge
OntoShare
User
RQL
QuizRDF
Spectacle
Knowledge Engineer
OntoEdit
OIL-Core
OMM
BOR
Sesame
OIL-Core ontology repository
Annotated Data Repository
RDF
RDF
pers05
731
par05
car
tel
about
OntoExtract
OntoWrapper
Data Repository (external)
10
Place in On-To-Knowledge
11
Why Decision Styles?

• Make the experiment more informative by also
  looking at cognitive styles of persons
• Measuring cognitive styles is done by
  administering standard test to subjects
• Find out whether there are correlations between
  cognitive styles and semantic web tool use (see
  test hypotheses)
• Theory used in the field of organisational
  behaviour (Driver)

12
Style Dimension I Information Use

• Information Use
• Amount of information actually used in the
  process of making decisions and solving problems.
• Maximizer
• High
• Level Satisficer
• Of Good enough
• Understanding/ solution
• Value of info
• Low
• Low Amount of information used
  High

13
Style Dimension II Solution Focus

• Solution Focus
• Zeroing-in on one clear-cut solution or course
  of action vs. identifying multiple solutions
  and/or alternate courses of action.
• Uni-Focused Style
• Multi-Focused Style

Data
Data
14
What Data To Collect

• Semi-structured interviews (pre- and post trial
  interviews)
• Time to complete the whole search task
• Automatic log of changes
• Automatic log of errors in the search task
• Automatic log of commands executed
• Detailed notes of the actions performed by the
  subject/on-site observations
• Questionnaires to be filled in by the user

15
Some Examples of Test Questions

• How does load management affect pricing?
• How is a virtual organisation formed?
• What styles of decision making can be
  distinguished?
• In what project has Claes Badenscheneider been
  active? (name misspelt with purpose)
• To whom should a registration for a ISPLC 2001 be
  sent? (email link to Yvonne. H)
• Where can you find information on virtual
  organisations?
• Name a knowledge management methodology?