The OWL language [1] is being widely used to de¯ne ontologies in the Web. This language provides three increasingly expressive sublanguages, namely OWL Lite, OWL-DL and OWL Full. The OWL-DL XML based syntax together with its correspondence with Description Logics (DL) [2], make it a good candidate to be the standard language for de¯ning ontologies used by Semantic Web applications. However, there are still relatively few tools that allow us to manipulate, store and query ontologies de¯ned using this language. Furthermore, Semantic Web applications, such as biological tools, use large ontologies, that is, ontologies with a large number (millions) of instances. Description logic based tools allow us to manage OWL-DL ontologies, but not very large ontologies. The development of tools for storing and querying OWL ontologies is currently under investigation. In this paper we present a plugin for the Prot¶eg¶e [3] environment for reasoning and querying OWL-DL ontologies stored in a relational database. We de¯ne a relational schema for storing OWL ontologies and a query/reasoning language, which supports reasoning on both, ontology structure and ontology instances. However, the main feature of this plugin is that developers of OWL-DL reasoners can customize it in order to use their own database and their own query language by means of several con¯gurations ¯les.
about the implementation of the language for one or more speci¯c repositories is also needed. All this information is provided by means of several con¯guration ¯les. These ¯les are XML ¯les. We have chosen XML because there exists many java libraries to manage XML. The use of RDF or RDFS could also be an option, but there are more tools for managing XML than RDF. We will consider changing to RDF in the later versions of the tool.
Finally, using all this information, the query and reasoning tool generator customizes a Protege generic graphical interface, which shows the information about the ontology using the ontology information con¯guration ¯les, and the speci¯c reasoning supported by the speci¯c query language.
Once the tool is generated, it allows users to make queries using the speci¯c operators de¯ned in the query and reasoning con¯gurations ¯les. These queries will be performed in a persistent repository. Note that the tool does not implement the query language. It is the language developer who has to give an implementation for each element of the language. 3
We have developed our plugin for querying and reasoning on OWL ontologies
stored in a relational database. Speci¯cally, it implements the ECQ (Extended
Conjunctive Queries) language [
Con¯guration Files The ¯le ontinformation.xml contains the functions for obtaining the ontology information from the relational database. For example, we de¯ne a function for obtaining all the features of a speci¯c property or a function for obtaining all the restrictions for a speci¯c class. Due to these functions are trivial, we will describe only the query and reasoning con¯gurations ¯les.
The queryLanguage element of the queryLanguages.xml ¯le contains the following information. The name of the language is RECQ, from Reasoning and Extended Conjunctive Queries. It is the language developer who chooses the name. The storage model is relational and the DBMS used for implementing it is Oracle. The ¯le RECQQueries.xml de¯nes the operators for the query mechanism. In our case, allowed operators are and, or, all, max and min. The RECQTbox.xml and RECQAbox.xml ¯les specify respectively, the Tbox and the Abox reasoning supported. We only have to specify name, description and parameters names for each inference. Finally, the RECQoracle.xml ¯le de¯nes the implementation for the queries (which will use the query operators and the Abox reasoning) and also the implementations for the Tbox inferences. Note that the ¯rst rule in the ¯le implements the connection to the Oracle DBMS.
Once the con¯guration ¯les are de¯ned, the use of the tool is really easy. First, the user must select the ontology. The content of the queryLanguages.xml ¯le is used in order to allow the user to select the query language he/she wants to use. Once the language is selected, using the con¯guration ¯les, the tool customizes the generic graphical interface for the selected language. Only reasonings supported by it are enabled, and the panel for building the queries is provided only to the supported operators. A window asking the login and password for connecting to the database is also shown. In order to see the description of a speci¯c language or reasoning, the user only has to place the mouse on the corresponding element and the tool shows its description. The user is now able to make queries over the selected repository. 4
This paper presents a Prot¶eg¶e Plugin that helps developers to build query and reasoning tools for OWL ontologies, which are persistently stored in a database. It is possible to customize the Plugin to di®erent query languages according to the needs of the users. In order to do this, some con¯guration ¯les are de¯ned. These ¯les de¯ne the query mechanism and specify the Tbox and Abox reasonings supported by the speci¯c query language. Using this Plugin it is possible to implement the same query language using di®erent storage models and di®erent DBMS easily, and to use a common graphical interface to perform queries and reasoning. We have also presented an example of how to use this tool for querying and reasoning on an OWL ontology which is stored in a relational database. 5
This work has been supported by the Spanish MEC Grant (TIN2005-09098-C0501).