What is an ontology? An inductive explanation.
Term list:
-
word list
- list of used terms
- stemming?
-
e.g.
- back in book index
- folksonomy
-
index
- prefered label
- alternative label
- deprecated label
-
glossary
- related
-
thesaurus
- broader term
- narrower term
- conceptGroup
-
taxonomy
- class - subclass
- class vs. instance
- classes are defined by features (feature: charakteristisches Merkmal)
- intension vs. extension
- class tree or concept lattice (Begriffsverband)
- "light weight ontolgy"
-
ontology
-
classes are defined by predefined rule types
- class axioms
- classification rules
-
classes are defined by predefined rule types
-
rule based ontology
- first order logic
- fully rule based system
- arbitrary class definitions
Each term is a specialization of it's predecessor.
- In fact we have a built a taxonomy.
- According to this term list an ontology is a sophisticated taxonomy.
classification agenda
- top down
- for each subclass classify if all features are given
- extension shrinks
Meta terms
- terms (classes, sets) are defined by features (intension)
- features define class tree
- examples (extension)
- intension vs. extension
from native format to XML to semantic
-
free format
- CSV
-
XML
- data are parsed according to a uniform syntax: you don't need to write a parser any more!
- underlying meta data model: XML tree, i.e. nested tagged content
- XML Schema allows for validating syntax of data
- language XSLT allows for transforming syntax, i.e. XML trees into XML trees
- schema check = analyse syntax of data = check coherence with grammar
- understanding of the data is open / 2b defined / undefined: bug or feature?
-
semantic
- data are interpreted according to an uniform knowledge representation: you don't need to write an inferencing engine any more!
- underlying data model: directed labeled graph
- schema part of an ontology allows for validating semantic (model conformity) of data
-
description logic
allows for
- classifying individuals
- checking soundness of a classification system
- rule languages allow for model transformations, i.w. translations of data models into data models
- semantic check = analyze understanding of data = check for coherence with a set of logical rules
- ontology schema allows for checking logical soundness (Widerspruchsfreiheit)