Presentation on theme: "The Ontology Construction Problem Ontology construction requires the active engagement of domain experts Existing ontology authoring tools are not tailored."— Presentation transcript:
The Ontology Construction Problem Ontology construction requires the active engagement of domain experts Existing ontology authoring tools are not tailored to the needs of domain experts: require good knowledge engineering skills focus on some aspects (e.g. controlled language input) but do not support the whole ontology construction process suffer from usability problems Supporting Domain Experts to Construct Conceptual Ontologies: The Rabbit to OWL Authoring Approach Leeds University team: Ronald Denaux, Vania Dimitrova, Tony Cohn, Ilaria Corda Ordnance Survey team: Ian Holt, Catherine Dolbear, Glen Hart, John Goodwin Project web site: http://www.comp.leeds.ac.uk/confluence/ The ROO tool can be downloaded from: http://sourceforge.net/projects/confluence The Ordnance Survey Experience Topographic ontologies (Hydrology and Buildings & Places) integrated with OS Master Maps Active involvement of domain experts is paramount Methodology for involving domain experts in the ontology authoring process (Kanga) has been developed and utilised in previous projects Controlled natural language (Rabbit) has been developed in response to a need for domain experts to be able to understand and author ontologies Rabbit to OWL Ontology authoring Assist domain experts with ontology construction by developing a tool (ROO) which provides: intuitive interface for authoring ontologies in the Rabbit controlled language systematic guidance throughout the whole ontology construction process based on the Kanga methodology compliance with existing ontology authoring tools Ensure rigour and effectiveness of the ontology development process, and produce better quality ontologies that can be reused more efficiently Architecture of the ROO Tool ROO is a Protégé plug-in. The GUI extends existing Protégé GUI components. The user guidance is implemented as a rule system based on the JBoss Drools rule engine. The parsing of the Rabbit sentences is based on an extension of GATE Concept entry in ROO. The user can inspect or edit the natural description from a domain source, as well as the Rabbit sentences that define the concept. The list of existing concepts and relations follows the Protégé interface. The resultant ontology is in OWL. Entering Rabbit constructs in ROO. The user is given support with the construction of sentences (errors, warning, and library with Rabbit patterns), as well as guidance what to do next (a task setter in the top right corner presented as a guide dog. Current State and Future Work Currently, we have implemented the main Rabbit constructs and have conducted several small user studies to examine the usability of ROO and identify future improvements. The next stage of the project includes: Larger evaluation study with domain experts at Ordnance Survey; Completion of the set of Rabbit constructs with appropriate support based on an analysis of common user errors; Extend ROO to provide support with ontology verification and to follow best practices of ontology construction.