Ontoprise: B3 - Semantic B2B Broker

1
Ontoprise B3 - Semantic B2B Broker
CSCI 8350, Spring 2002,UGA
Bernhard Schueler

• whitepaper review

2
Overview

• Aim of Ontoprise
• Problem interoperability
• Previous approaches
• Using ontologies
• Using RDF
• Remarks

3
Aim of Ontoprise

• The aim of Ontoprise is to provide ontology-based
  B2B-applications.
• Those applications should make the semantics of
  business information machine-processable.

4
Example Product Catalogues
The sellers product catalogue is transformed
into an electronic data format. Programmers
write transformation programs to generate
different formats and views required by
buyers. This process requires human resources
and time.
5
Example computer dealer

• Seller and buyer have catalogues, which are
  structured differently.
• Sellers catalogue a printer has an attribute
  type with values laser, ink or matrix.
• Buyers cataloguethere is a subclass laser
  printer.

6
Problem Mapping

• How can the data formats automatically be mapped
  between the catalogues?
• Mapping is necessary to create and update the
  buyers catalogue.
• It is expensive to program those mappings.
• Solutionautomate the creation of mappings,
  which can automate creation and update of the
  catalogue.

7
Previous approaches - EDIFACT

• EDIFACT
• provides a standardized electronic format to
  exchange data between sellers and buyers,
• allows the specification of standards for
  application areas.
• The standards cover the semantics of different
  application areas.

8
Previous approaches XML

• XML-specifications provide a semi-structured
  description of products, orders, .
• DTDs or XML-schema specifications define the
  structure of the data.

9
Advantages of XML over EDIFACT

• Readable for non-experts,
• Integrated with other document/data exchange
  standards,
• DTDs, XML-schema enable generation of tools like
  form editors,
• Designed for the web.

10
XML Description, Example

• ltarticlegtltarticleidgta_5634lt/articleidgtltcategorygt
  printerlt/categorygtlttypegtlaserlt/typegt
• lt/articlegt
• ltarticlegtltarticleidgta_5634lt/articleidgtltcategorygt
  laserprinterlt/categorygt
• lt/articlegt

11
Deficiencies of XML 1

• Implicit agreement on meaning of tags is
  necessary.
• Validation of XML using DTDs is only syntactic.
  The labeled tree structure is checked.
• There cant be semantic constraints, e.g. The
  price of a mouse may only be 1/100 of the price
  of a computer system.
• There is no inheritance.

12
Deficiencies of XML 2

• TransformationsThey are purely syntactical,
  using XSLT and DTDs.Therefore they would have to
  bee hand-coded.
• QueriesXQL and XML-QL allow to query the
  document structure.Give me all printers for
  which cartridges are available that last for more
  than 5000 pages.Can be done. But its very
  cumbersome and overloaded with syntactic aspects.

13
Ontologies

• Ontologies define a domain and provide a shared
  vocabulary.
• leight-weight ontologies model concepts with
  properties and relationships, which are embedded
  in an is-a-hierarchy.They are comparable to
  database schemata.
• heavy-weight ontologies also contain rules and
  constraints.

14
Ontologies in B3

• Ontology

Ontology
OntoMap
DTD Maker
OntoLift
DTD
DTD Custom
DTD XML documents
DTD XML documents
15
Ontologies in B3 - with RDF

• Ontology

Ontology
OntoMap
RDF documents
RDF documents
16
The B3 B2B Broker Basic tools

• OntoEdit,
• OntoFact,generates forms for attributes and
  relations of instances (of concepts of an
  ontology),
• OntoAnnotate,allows the user to add meta
  information to HTML, Word, PDF.

17
The B3 B2B Broker Advanced tools

• OntoLift, provides means to identify concepts
  and relationships from an XML-DTD or XML-schema,
• DTDMaker,
• OntoMap, semi-automated creation of mapping
  rules,
• OntoCompile,creates a set of XSLT
  transformations.

18
Remarks

• By the time the paper was published not all of
  the tools had been implemented.
• By now, data would rather be stored using RDF
  than pure XML.
• A means to create forms, views and web content
  form RDF might be more interesting.