Web ONtology mediateD Extraction of Relational data (WONDER) for Web-based ObdA with the Hgt-db (WOAH!)


Aim

The aim of the project is to put the OBDA system to the test by enhancing a typical genomics database—the horizontal gene transfer database (HGT-DB)—with the OBDA technologies and to equip it with a web-based end-user usable GUI to query the data through the ontology.

Approach

To achieve the aims, we are conducting the following steps.

First, the preparatory steps to make the database “OBDA-ready” were to reverse engineer a conceptual data model (in ORM) from the HGT-DB database, to transform the ORM diagram into a DL-liteA ontology and devise guidelines for reuse of this procedure, and to represent this ontology in Protégé.

The second step consists of setting up and using the basic OBDA configuration: to create the mappings from the classes and object properties in Protégé to SQL queries over the relational database through the OBDA plugin and to examine feasibility of both basic OBDA queries and those provided by the subject domain experts (posed in SPARQL in the OBDA plugin and evaluated by the QuOnto reasoner), as well as to conduct a performance analysis and optimisation.

The third phase, to make this user-usable, consists of development of a web-based GUI to graphically browse and query the ontology, which includes the development of a graphical query language. Given the progress made, we will extend the query features with the option to pose epistemic queries and add a query completion feature, and extend the ontology with new classes and object properties (which were hitherto implicitly used) so as to further simplify querying the database.

Last, but not least, the functionality of the Web-based OBDA will be evaluated against the present HGT-DB web interface.

Results


The results described here are split up into three aspects: (i) early experimentation, (ii) the graphical query interface and use case with the HGT-DB, and (iii) supplementary data for a small experiment on the feasibility with rough concepts in the HGT conceptual data model (‘application ontology’).

I. Initial results of the OBDA use case scenario



Preliminary results of the preparatory steps, the basic OBDA configuration, and a comparison with the web interface of the HGT-DB can be found in:

  • R. Alberts, D. Calvanese, G. De Giacomo, A. Gerber, M. Horridge, A. Kaplunova, C. Maria Keet, D. Lembo, M. Lenzerini, M. Milicic, R.Moeller, M. Rodriguez-Muro, R. Rosati, U. Sattler, B. Suntisrivaraporn, G. Stefanoni, A.-Y. Turhan, S. Wandelt, M. Wessel. Analysis of Test Results on Usage Scenarios. Deliverable TONES-D27 v1.0, Oct. 10 2008.

II. The WONDER system


The WONDER system was developed during Giorgio Stefanoni’s internships and BSc thesis under supervision of Werner Nutt and Marijke Keet:

Supplementary data for the working WONDER system:
  • Application ontology of the horizontal gene transfer (OWL file v1, DL axioms in pdf v1)
  • Ontology <-- Database mappings (obda file, using v1 of the application ontology and HGT-DB physical schema of August '08)
  • Performance measurements of the WONDER system (pdf).
  • Screenshots of the WONDER system:

Ontology browsing pane
ontology1

Path selection
path

Query pane
query1

Constraint adder
constraintAdder1

Constraint manager
constraintManager1

Results pane
results

III. Rough ontology experiment


Because there are multiple files used during the experimentation with rough ontologies, which cover the above-mentioned setting but also include experiments that have nothing to do with either OBDA or with the HGT-DB, the test files have been put together at this location

People (in alphabetical order)