WP 3: Enhancing eLearning with semantic knowledge - PowerPoint PPT Presentation

1 / 25
About This Presentation
Title:

WP 3: Enhancing eLearning with semantic knowledge

Description:

We have participated in Bulgarian and European Projects ... Glossary. LT4eL Ontology: overview. Target Domain: Computer Science for non-computer scientists ... – PowerPoint PPT presentation

Number of Views:21
Avg rating:3.0/5.0
Slides: 26
Provided by: lt4
Category:

less

Transcript and Presenter's Notes

Title: WP 3: Enhancing eLearning with semantic knowledge


1
WP 3 Enhancing eLearning with semantic knowledge
  • Kiril Simov
  • Utrecht, 01.02.2007

2
Plan of the Talk
  • Preliminaries
  • General overview of LT4eL Ontology
  • Synopsis of the Ontology Construction
    Methodologies
  • The construction of LT4eL ontology
  • Conclusions

3
Our Background (Why we?)
  • We have participated in Bulgarian and European
    Projects related to Ontologies, e.g.
    OnToKnowledge
  • We have experience with annotation of Bulgarian
    Treebank with ontological information
  • We have experience on aligning ontologies with
    lexicons

4
WP3 Partners
  • Multilinguality - University of Hamburg
  • Ontology development and annotation - University
    of Malta
  • Key Words providers All partners
  • Pilot members - University of Hamburg, Utrecht
    University, IPP - Bulgarian Academy of Sciences,
    University Al.I.Cuza of Iasi
  • Lexicon providers All partners
  • Annotation of learning objects All partners

5
WP3 Dependencies
  • WP 1 collection and preparation of LOs
  • WP 2 WP 2 results are input to WP3
  • WP 4 Integration of tools
  • WP 5 Evaluation and validation

6
WP3 Goals
  • Creation of Ontology and Ontology System to
    support
  • Classification of learning objects
  • Annotation of LOs with concepts ontology search
  • Multilingual search for learning objects
  • Interlingua between the different languages

7
Documents SCORM
Pseudo-Struct.
Basic XML
CONVERTOR 2
Documents SCORM
Pseudo-Struct
Documents HTML
Glossary
CONVERTOR 1
Metadata (Keywords) Ling. Annot XML
BG
EN
Documents User (PDF, DOC, HTML, SCORM,XML)
REPOSITORY
8
LT4eL Ontology overview
  • Target Domain Computer Science for non-computer
    scientists
  • The creation of ontology is data-driven (based
    on LOs in all languages)
  • Language independent
  • Evaluation
  • Consistency of the definitions
  • OntoClean Methodology (Guarino)
  • Coverage of the domain
  • Annotation of the LOs in all languages

9
Synopsis of the Ontology Construction
Methodologies (1)
  • Uschold and King Methodology
  • Purpose, Building, Evaluation, Documentation
  • GrĂ¼ninger and Fox Methodology
  • Motivation, Competency Questions, Terminology,
    Axioms, Competency Theorems
  • Methontology Methodology
  • Specification, Knowledge Acquisition,
    Conceptualization, Integration, Implementation,
    Evaluation, Documentation
  • SENSUS Methodology
  • Seed terms, Extension pn the basis of WordNet

10
Synopsis of the Ontology Construction
Methodologies (2)
  • Conclusions
  • The life cycle of an ontology is similar to that
    of other software products design phase,
    prototyping, implementation, exploitation,
    support, documentation.
  • The ontology has to reflect the domain which it
    represents and at the same time be
    application-independent ?

11
Synopsis of the Ontology Construction
Methodologies (3)
  • Conclusions (continued)
  • Ontology development is necessarily an iterative
    process
  • An evaluation of the sources of information is
    essential to the development of an ontology

12
The creation of LT4eL ontology
  • The cycle of ontology creation
  • Processing of the keywords in LOs
  • Formalization of the meanings
  • Linking to an upper ontology
  • Addition of new concepts
  • Addition of relations
  • Documentation
  • Lexicons and pilot experiment

13
Processing of the keywords
  • Extraction of the keywords annotated in the LOs
    (8 languages)
  • Translation in English
  • Initial classification of keywords
  • Collecting definitions
  • Definition selection and Sense differentiation

14
Formalization of the Meanings
  • Domain ontology based
  • for each meaning, an appropriate class in the
    domain ontology was created
  • later on, relations among concepts will be added
  • Upper ontology based
  • each concept was mapped to synsets in WordNet 2.0
    and OntoWordNet
  • mapping relations equality and hypernymy

15
Linking to an Upper Ontology
  • DOLCE ontology was selected as an upper ontology
    mapping via OntoWordNet
  • However, mapping to SUMO ontology is also gained
    through Wordnet 2.0
  • As a consequence, later on a comparison might be
    done between these mappings
  • Top-down approach is adopted

16
Addition of New Concepts
  • In order to ensure better coverage we have added
    concepts that are not related to keywords in the
    LOs
  • Sources for these concepts are
  • OntoWordNet
  • Head words of multi-token keywords
  • Entries in terminological lexicons

17
The Concept Distribution
DOLCE
OntoWordNet
CSnCS
18
Addition of Relations
  • In two ways
  • relations that are defined in DOLCE will be
    inherited by the domain ontology
  • domain specific relations will be introduced on
    the basis of analysis of the definitions that we
    have collected for the keywords

19
Documentation
  • The track is kept
  • on the sources of each concept and relation
  • on information about all changes in the
    definition of the concept or the relation

20
Lexicons and pilot experiment
  • Creation of lexicons for all languages of the
    project English, German, Dutch, Portuguese,
    Maltese, Polish, Czech, Romanian, and Bulgarian
  • The lexical entry will contain a synset, a
    recognition grammar, concepts

21
Construction of the Ontology
  • Bottom-Up We started with the keywords in the
    LOs and other available materials like
    glossaries, etc
  • Top-Down constraints The Domain Ontology will be
    an specialization of an Upper Ontology

22
Precision of the Ontology
  • Lexicon
  • Vocabulary with NL definitions
  • Simple Taxonomy
  • Thesaurus
  • Taxonomy plus related-terms
  • Relational Model
  • Unconstrained use of arbitrary relations
  • Fully Axiomatized Theory

We start here
Now we are here, based on WN20 and OWN
Our goal
23
Format of the Lexicon
  • lt!ELEMENT entry (hwg,gram,def,trg,defbag)gt
  • lt!ATTLIST entry
  • id ID REQUIRED
  • pos CDATA IMPLIED
  • status (compotherunchecked) "unchecked"gt
  • lthwggt group of head words for one sense in
    English
  • ltgramgt grammar explanation in free text
  • ltdefgt - canonical definition
  • lttrggt - translation group in some language
  • ltdefbaggt - a set of definitions taken from
    Internet

24
Ontology Annotation
  • Two types of annotations
  • Annotation in the metadata section of LOs
  • LOs will be annotated with concepts that are
    related in the ontology
  • Annotation in the text
  • Identification of the text chunk that will be
    annotated
  • Assigning of all possible senses for the chunk
  • Sense disambiguation
  • Relation annotation

25
Conclusion
  • We have selected a methodology which reflects the
    nature of the ontology for LT4eL
  • We have selected the relevant concepts
  • We have to add relations to the Ontology
  • Ontology System is in design phase
  • We have to construct Lexicon and Annotation
    Grammars
Write a Comment
User Comments (0)
About PowerShow.com