Ontology Development ..the Lean way Lieke Verhelst MSc Linked Data Factory
Ontology Development
..the Lean way
Lieke Verhelst MSc
Linked Data Factory
2
www.linkeddatafactory.comCC-BY-SA
Lessons learned in collaborative semantic modelling
Start Finish
AgendaT h e s t a g e
Lean = value for the customer
3
www.linkeddatafactory.comCC-BY-SA
Use casesS e m a n t i c t e c h n o l o g y
Identify Compare
Match
Integrate
Relate
Transform
Find
Understand (human)
Evaluate Logic
(machine)
4
www.linkeddatafactory.comCC-BY-SA
Integrate, transform
Model complexityU s e C a s e R e q u i r e m e n t s
Identify
Understand (human)
Match, compare
5
www.linkeddatafactory.comCC-BY-SA
Very Simple ModelM o d e l C o m p l e x i t y
UID-1
Book @en
“A printed work..” @en
UID-2
Dog @en
“Animal with...” @en
6
www.linkeddatafactory.comCC-BY-SA
UID-1
Book @en
“A printed work..” @en
Thesaurus ModelM o d e l C o m p l e x i t y
UID-2
Fiction @en
UID-3
Nonfiction @en
UID-A
Novel @en
UID-163
Magazine @en
“.. often illustrated ..” @en
BROADER/NARROWER
MATCHRELATED
7
www.linkeddatafactory.comCC-BY-SA
UID-71
Ontology ModelM o d e l C o m p l e x i t y
UID-12
Topic @en
UID-B
Publisher @en
UID-A
Novel @en
UID-2
Author @en
Audience @en
hasTopic
UID-1
Book @en
has Author min 1
hasPublisher
hasAuthor
hasAudience
8
www.linkeddatafactory.comCC-BY-SA
IdentifyUnderstand
(human)Relate Compare/Match Transform All
URI strategy URI strategy …… …… …… …… …… …… Skills
Meaningful, meaningless
Meaningful, meaningless …… …… …… …… …… …… Tools
Labels (preferred, alternative) …… …… …… Restrictions Environments
Language Relation definition Type Propagation Datatype definition Resources
Definition Relation direction Equality SPARQL Construct Users
Relation semantics Inference Rules Targets
Technical DesignU s e C a s e R e q u i r e m e n t s
9
www.linkeddatafactory.comCC-BY-SA
Integrate, transform
Model strategyU s e C a s e R e q u i r e m e n t s
Identify
Understand (human)
Match, compare, infer
BRIDGE THIS GAP!
10
www.linkeddatafactory.comCC-BY-SA
Identify
Tasks
Design URI strategy
Decide on identifiers
(meaningful,
meaningless)
Register and
document
Phase 1
Describe
Tasks
Name it
Translate it
Define it
Phase 2
Relate
Tasks
Relate it
Relate it
Relate it
Phase 3
Design
Tasks
Refine relations
Redefine hierarchy
Bridge SKOS and OWL
Patterns
Phase 4
Code
Tasks
Refine relations
Define datatypes
Code conversions
Write rules
Phase 5
Incremental designM o d e l S t r a t e g y
MOST EFFORT (SME’s)!
11
www.linkeddatafactory.comCC-BY-SA
01Identifiers
List of
identifiers and
e.g. link to picture
Simple API
02Names
Items have definite
pref., alt. names and
translations
DeliveryI n c r e m e n t a l D e s i g n
03Items defined and
related in
thesaurus model (SKOS)
Relations
04Separate OWL model with
more specific relations
and hierarchy.
OWL-SKOS mapping Instances
05OWL model with
datatypes, restrictions
and rulesTransformations
12
www.linkeddatafactory.comCC-BY-SA
Tools
• Off the shelf available
• Open source available
• Tool selection by experienced staff
Change
• Change management process
• Change of model
• Change in data
• When ready for use?
Use
• Learning by using
• Incremental improvements
• User feedback
Skills
• Targeted training
• Separation of concern
• Modeling guidelines available (e.g. ISO 25964 - thesaurus)
Communicate
• Change management
• Methods for access
• User feedback
Practical issuesM o d e l S t r a t e g y