Ontology Recommender Web service
To automatically recommend an ontology to use for semantic annotations
Presentation & Demonstration
As the use of ontologies for annotation of biomedical datasets rises, a common question researchers face is that of identifying which ontologies are relevant to annotate their datasets. The number and variety of biomedical ontologies is now quite large and it is cumbersome for a scientist to figure out which ontology to (re)use in their annotation tasks. NCBO develops an ontology recommender service, which informs the user of the most appropriate ontologies relevant for their given dataset. The recommender service uses a semantic annotation based approach and scores the ontologies according to those annotations. The recommender service uses the Annotator Web service . The prototype service can recommend ontologies from UMLS and the NCBO BioPortal.
Please try the NCBO Ontology Recommender service prototype: http://keg.cs.uvic.ca/ncbo/obs/OBARecommender.html.
- Community & usage: [Trish Whetzel]
- Design, utility & applications: [Nigam Shah] and [Clement Jonquet]
- Technical support: []
Documentation & References
- See also the Annotator Web service wiki page.
- Please refer to:
Versions (prototypes & releases)
- November 2009 - Second prototype (v1.1) for the recommender: http://obs.bioontology.org/recommender/Recommender1.1.html
- March 2009 - First prototype of UI for the recommender: http://keg.cs.uvic.ca/ncbo/obs/OBARecommender.html
- February 2009 - The first prototype (v0) has been released for testing and evaluation. Please check it out: http://obs.bioontology.org/recommender/Recommender.html
The original datasets used for evaluating the Recommender 1.1, as well as the recommendations generated are available here: http://obs.bioontology.org/recommender/results/
Collaboration & Acknowledgment
- The user interface for the recommender serice UI prototype is designed & developed by NCBO members from University of Victoria
- We also thank Helen Parkinson (European Biomedical Institute), Stephen Granite (Johns Hopkins University) and Wei-Nchih Lee (Stanford University) for the help in the Recommender (version 1.1) evaluation.