Figure 7.2 Different forms of knowledge
- Figure 7.2 shows that knowledge can come in a variety of forms which are structured, semi-structured or unstructured.
- The knowledge needs to be organised by gathering the knowledge & grouped, index or categories it in some way.
- If each person organise the same knowledge, we might come up with wide variations depending on our understanding & perspective of the subject.
- To prevent the situation, ontologies were developed.
- Ontologies : so called 'knowledge map' where a vocabulary of terms and relationships are conceptualized to represent the knowledge.
Figure 7.3 Ontology & texonomies
- Ontology: An overall conceptualization.
- An ontology may have non-taxonomic conceptual relationships.
- Taxonomy: Scientifically based scheme of classification.
- Knowledge texonomies generate hierarchical classification of terms that are structures to show relationships between terms.
- Building ontologies manually:
1) Identify purpose & scope.
2) Build the ontology via a three-step process:-
(a) ontology capture
(b) ontology coding
(c) integrate existing ontologies
3) Evaluate ontologies.
4) Document ontologies.
5) Provide guidelines for previous phases.
- Concepts are extracted from raw data using variety mature techniques:
- 'Word sense disambiguation'
- 'Tokeniser'
- 'Pattern matching'
- Semi-automated generation with machine learning
3.1 INTEGRATING ONTOLOGIES
Figure 7.4 Ontology integration techniques
- The current approaches for integrating a number of ontologies:-
- Reusing available ontologies linking different domains.
- Aligning ontologies by establishing links between them through some form of translation function using agent technology.
- Merging ontologies to create a single ontology.
- Integrating ontologies through clustering on the basis of similarities.
No comments:
Post a Comment