HCLS/Semantic Web Step By Step
Step by step: Creating interoperable Semantic Web resources
for life science and health care (outline)
- Matthias Samwald (MatthiasSamwald)
- Kei Cheung
(please add your name if you contribute text)
This guide was inspired by experiences with the conversion of the SenseLab databases and the creation of other Semantic Web resources. It does not represent any consensus among the community (at least not in its current version).
- A web server with root access
If you are new to Semantic Web technologies and ontologies, the following documents might be helpful for you:
- Ontologies for Biomedicine – How to make and use them -- describes the principles of ontology design in biomedicine.
- RDF primer -- describes RDF, a fundamental standard of the Semantic Web.
- OWL Web Ontology Language Guide -- describes OWL, another standard that extends RDF.
- Introducing SPARQL: Querying the Semantic Web -- describes SPARQL, a forthcoming standard for a RDF query language.
How to set up a development / deployment environment
- Download Protege 4
- Set up a public Subversion repository (SVN)
Searching and choosing existing foundational ontologies and metadata standards
- Look through existing biomedical ontologies in the OBO Foundry
- Choose a foundational ontology: RO, BFO, DOLCE...
- Look through existing standard Semantic Web ontologies: Dublin Core, FOAF, SIOC
- Use a Semantic Web Search (http://sindice.com or http://swoogle.umbc.edu/) engine to find other existing domain ontologies.
- Choose a PURL-based URI scheme
- Set up a purl.org account and create redirects to the SVN
Developing the ontology / ontologies
- Check consistency. If ontology is inconsistent (e.g. classes are not satisfiable), change ontology and check consistency again.
Automatic conversion of data into RDF/OWL ontologies
- Select programming platform that has a useful API for your source data (e.g., a relational database or XML) and RDF/OWL. Here is a list of some recommended RDF/OWL APIs for different scripting languages:
Setting up a SPARQL endpoint
- Download and install Virtuoso (open source edition) on your server
- Upload the recent versions of your ontologies into labeled graphs. The label of each graph should be identical with the PURL of each ontology.