Ontology Maturing: a Collaborative Web 2.0 Approach to Ontology Engineering

1
Ontology Maturinga Collaborative Web 2.0
Approach to Ontology Engineering
Simone Braun, Andreas Schmidt, Valentin
Zacharias, Gabor Nagypal, Andreas Walter

• FZI Research Center for Information Technologies
• Dept. Information Process Engineering
• Karlsruhe, GERMANY
• braunaschmidtzach,walter_at_fzi.de
  nagypal_at_disy.net
• http//www.fzi.de/ipe

Presented by Ilya Zaihrayeu at KnowDive seminar
on 28.05.2008
2
Motivation

• Ontologies are shared understandings of a
  particular domain
• Two perspectives to ontology engineering exist
• Ontologies are developed by ontology engineering
  experts
• Expensive
• Ontologies express shared understanding of
  experts, not users
• Can still be error-prone if modeling experts are
  not (also) domain experts
• Ontologies are developed by experienced users who
  become ontology modeling experts right away
• Unrealistic that users learn ontology modeling
  fast
• Traditional ontology modeling methodologies are
  treated as an overhead to their work process
  time lag between the emergence of concepts and
  their inclusion in ontologies are far too big for
  ontologies to be useful

3
Motivation, contd
The spaghetti-problem

Ontology food
Tasks use ontology for describing new web-pages
or images

• Challenges
• Ontologies never correct, constantly changing
• Integration of new ideas that emerge in users
  workflow
• Users familiar with simple tagging Requires
  lightweight tools

Our approach A work-integrated process for
collaborative ontology maturing
In real world Spaghetti is not just
spaghetti Ask your preferred Italian cook
(domain expert) or a gastronome (user)
4
Outline

• The ontology maturing model
• Important observations about ontologies
• Motivation and triggers for maturing activities
• The ontology maturing process
• Our tools for collaborative ontology maturing
• ImagenotionsImage based and collaborative
  construction of ontologies
• SOBOLEOSocial Bookmarking and Lightweight
  Engineering of Ontologies

5
Important observations about ontologies

• Ontology building - a learning process
• Individuals deepen their understanding
• of the real world
• of the vocabulary that describe the world
• emerging ideas impossible to integrate directly
  
• Barriers formality and complexity in usage
• Social bookmarking systems (e.g.
  del.icio.us)informal, lightweight, easy-to-use,
  and easy to understand.
• Requirement ontology editing tools should be
  easy to use
• Ontologies continuously evolve in work processes
• Ontology building process of continuous
  evolution
• Concept ideas emerge in daily work
• Maturing of concepts through collaborative
  interaction
• gt Ontology building integration in the daily
  work process

6
Motivation and triggers for ontology editing

• Ontologies users expectation's
• Search results retrieve appropriate content,
  Enhance their work performance
• Description of contents publish for the
  community (e.g. bookmarks, images)
• Motivation and triggers for maturing activities
• Unsatisfied user expectations
• Incomplete search results
• Misspelling, synonyms, multilingualism
• Homonyms bad search results
• Description of own contents
• Missing concepts
• Mismatch of abstraction level
• Missing relations between concepts

7
The ontology maturing process
All users

• Emergence of ideas
• Introduction of new concept ideas
• Representation with tags
• Consolidation in communities
• Reuse of tags from others
• Create concepts
• Add synonyms, separate homonyms
• Multilingualism tags in different languages
• Formalization
• Relations between concepts Hierarchies, ad-hoc
• Axiomatization
• Adding background knowledge

Experienced users / domain experts
Domain experts/knowledge engineers
8
The ontology maturing model

missing concepts (mis-)spelling
synonyms multilingualism homonyms
Missing guidance
mismatch of abstraction level

• New sub concepts
• Thin spaghetti
• Thick spaghetti

spaghetti alla chitara
spaghetti vs. spagheti
ziti
9
Our tools for collaborative ontology maturing

• Image based and collaborative construction of
  ontologies
• Current application domain semantic description
  and search for images
• By Andreas Walter and Gabor Nagypal
• SOBOLEO
• Social Bookmarking and Lightweight Engineering of
  Ontologies

10
Semantic description and search for images

• Tagging is standard for image descriptions
• Problems
• Convert tags to concepts
• Domain experts are not knowledge engineers
• Contents constantly change
• Solution Imagenotion
• An imagenotion graphically represents a semantic
  notion
• associate tags with an imagenotion
• Label titles, synonyms - in different languages
• It is possible to create relations between
  imagenotions
• Hierarchies, ad-hoc
• Different imagenotions with same label represent
  homonyms

11
Imagenotion
Image based and collaborative construction of
ontologies
Dog canis
Retriever
Labrador
Lab
Ball
Homonyms notebook
12
The imagenotion methodology

• Step 1 Emergence of new ideas
• Introduction of new concept ideas
• Representation with tags

13
Imagenotion methodology

• Step 2 Consolidation in communities
• Add synonyms, Multilingualism tags in different
  languages

14
Imagenotion methodology

• Step 2 Consolidation in communities, contd
• Split homonyms into two distinct imagenotions

15
Imagenotion methodology

• Step 3 Formalization - relations

16
Imagenotion methodology
Describing images with imagenotions

17
Imagenotion future work

• Improvement and extension of the tool
• Imagenotion is the official tool for ontology
  editing in the EU-funded project IMAGINATION
• Export functionalities
• SKOS
• OWL-DL
• Evaluation of imagenotion in different setups
• Expert only only experts may generate
  imagenotions
• Collaborative everyone may manipulate
  imagenotions
• Collaborative version
• Integration of voting to reach a collaborative
  agreement of ontologies

18
Our tools for collaborative ontology maturing

• Imagenotion
• Image based and collaborative construction of
  ontologies
• SOBOLEO
• Social Bookmarking and Lightweight Engineering of
  Ontologies
• Current application domain Tagging of web
  resources
• By Simone Braun and Valentin Zacharias

19
Overview

• SOBOLEO Social Bookmarking and Lightweight
  Engineering of Ontologies
• SOBOLEO is
• a tool for editing ontologies
• a tool for collecting and annotating web
  resources as bookmarks
• collaboratively usable
• web based accessible

Im Wissensnetz
20
Procedure

• User finds a new web resource she wants to add to
  the shared index
• Annotating with concepts from the ontology or
  arbitrary tags
• Gathering arbitrary tags as prototypical
  concepts for later consolidation and placement
• Structuring concepts with taxonomic relations
  according to the SKOS Core Vocabulary

Im Wissensnetz
21
Adding annotations

• Collecting and annotating the current web
  page from within the browser with a popup window
• As bookmarklet
• Including full text content of the web page
• New tags are automatically added to the ontology
  aufgenommen

Im Wissensnetz
22
Editing annotations
23
Searching shared index
1 Labels
2 Annotation
3 Ontology
4 Full Text
5 Refinement
6 Relaxation
Im Wissensnetz
24
Conclusions and future work

• Our model of ontology maturing and tools allow
• Integrated ontology development with daily work
  processes
• Four different steps of ontology maturing
• The continuously evolvement in a maturing process
• Future Work
• Evaluations of our tools (already done in a
  subsequent paper)
• Refinement of our model based on evaluation
  results
• Thanks? Do you have questions?