Semantic web and ontologies at the Institute and Museum of the History of Science of Florence:
The Institute and Museum of the History of Science (IMSS) of Florence, as content provider, was part of the consortium MESMUSES in the European IST Project which ended on July 2003. Since then IMSS continued to work on ontologies in four main directions:
- Improve ontology.
- Create a consistent view of our archives to be used in ontology repository.
- Implement search engine and user interface.
- Built viewpoints and itineraries in order to use ontology as an e-learning system.
We experienced that semantic web approach to information retrieval is a way to make more effective the access to large and heterogeneous data warehouses (like IMSS archives). Furthermore, in our knowledge domain, we are currently undergoing the important effort to develop an ontology schema as basis for terms meaning and relationship.
On the short run the goal is to enable a semantic search system on IMSS databases, which will be offered to skilled users. On the long run the goal is to build-up a community for the history of science and to enable a search system based on a common ontology.
The state of the art for semantic web search engines requires three-layer software:
- Relational database engine, which gives consistent view to the archives and enables SQL access to them.
- Adaptation layer with high-level conceptual model that enables access to the databases in a consistent way (e.g. RQL for RDF query, OQL for Object orientation, STELLA for declarative logic).
- Semantic web engine, which uses the features and the models of layer (2) in order to render and navigate datasets.
Despite the coherence of this theoretical model some unsolved problems can be found in layer (2). In our experience in Mesmuses project we didnt find any product able to manage a large amount of data with affordable reliability.
We choose an alternative approach by making a direct connection between Semantic web engine and database layer using dynamically generated SQL query. This approach does not inhibit the possibility to add in layer (2) other components when eventually they will be available as RQL or OQL engine.
Our software product is made of two components: the ontology management and the semantic web portal.
Furthermore, among other activities, we are currently analysing a graphic metaphor model that will allow us to developed in the future a graphical view of semantic relationship between concepts and/or objects in our knowledge domain.
Ontology for life sciences:
A RDFS conceptual schema has been designed. It is structured by the three top-level categories of the application: official school programme for each grade, concepts of life-sciences, and selected topics of the real-world news. The resulting schema is quite complex, and its design has proven to be a challenge. It has been tested, validated and documented from end of February to end of March 2003.