Comprehesion of an RDF dataset presents some problems for potential users of the data since a conceptual model is lacking. A single vocabulary is not sufficient to describe how multiple reused vocabularies are used in a given dataset, describe the intended rather than actual usage and are often absent.

This experimental browser explores the possibility of induction of a conceptual model from the raw triples and its use in accessing RDF data.

This project is also a demonstration of XQuery application development and will be released as part of the XQuery wikibook project.

• For a given SPARQL endpoint, a type/property/relationship model is induced and used to guide RDF interface the browser.
• The induced model is viewable as a standard Entity-Relationship diagram, showing type cardinalities and relationship multiplicities.
• Inverse relationships are induced and removed to simplify the diagram
• The model may be augmented with rules (expressed in XQuery) to compute derived resource properties and resource set properties
• Some degree of faceted browsing is supported - only conjunctive resource constraints at present.

• Sampling is required in places for types and relationships with high cardinality, so model may be incomplete.
• Namespaces for vocabs and resources are not yet induced
• Abstract type induction is needed for vocabularies which use rdf:type for simple categorisation (education, os)

There is a long list of work and aspirations:

• Navigation is confusing even to the developer!
• Improved faceting, searching and exporting.
• Constraint input via the model diagram
• Collaborative editing of the model to add descriptive information and notifications of errors and omissions
• Export of an owl vocabulary
• Reconcilliation with available owl statements

• The SPARQL interface assumes the availability of aggregate functions and has only be run against Talis plstform endpoints.

• XQuery on the eXist XML db platform
• Chris Wallace 12 Nov 2009

This work is made possble by the work of the eXist db team, the folk at Talis and the support of the Bristol Institute of Technology at UWE Bristol.