Ontology-Based Integration of Information---A Survey of Existing Approaches

Wache et al

ontology semantic translation information integration

@conference{wache:ois-2001,
  title={Ontology-Based Integration of Information---A Survey of
         Existing Approaches},
  author={Wache, H. and Voegele, T. and Visser, U. and Stuckenschmidt, H.
          and Schuster, G. and Neumann, H. and H{\"u}bner, S.},
  booktitle={{IJCAI}-01 Workshop: Ontologies and Information Sharing},
  volume={2001},
  pages={108--117},
  year={2001}
}

Problems that may arise in information interoperability (KIM & SEO 1991; KASHYAP & SHETH 1996a):

• Structural (schematic) heterogeneity: Data form is different
• Semantic (data) heterogeneity: Data is different---the intent, content, or interpretation

Three main causes of semantic heterogeneity: (GOH 1997)

• Confounding conflicts are when information seems to have same meaning, but differs in reality
• Scaling conflicts are when different referents are used to measure a value
• Naming conflicts are when labels differ significantly

Single ontology approaches have one single common interlingua into which all information sources are mapped

• Works well when information sources are very similar
• Can be difficult, e.g., if one has different granularity
• Can be easily affected by changes in information sources

Multi-ontology schemes

• Each information source has its own ontology
• Less commitments to centralized scheme
• Need some sort of representation formalism in which to define inter-ontology mappings
  • Not quite the same as an interlingua ontology, but filling the meta version of that role?
• Difficult to define all these ontologies

Hybrid approach

• Each information source has its own ontology
• But they're crafted from a common vocabulary or ontology
  • Primitives
    • Primitive core, conservative outer ontologies?

May use ontologies to describe information, or as query language

• Not clear whether ontology or source query models would be preferred by users