I swimming the huge sea of semantic Web and Linked Open Data (LOD) to use these knowledge bases for annotating semantically biological experiment of my organisation.
My idea consists of:
- RDF-izing my local biological experiment data in my triple store
- linking them to references which point to LOD
- If some concepts are not available as LOD, completing my virtual KB locally, by creating other types and properties in my triple store linked to the LOD cloud.
- Analyzing data through federated
SPARQL
queries, to leverage my virtual KB (= local data of points 1,2 plus external LOD of point 3)
In principle, that's is exactly one of the use cases LOD were created for but, practically speaking, there are some issues to tackle.
Issue 1: are federated queries scalable for the real world problems?
Issue 2: There are many linked data vocabularies out there in the LOD Cloud. In this day and age, it is very huge and covers many fields of Life Sciences, but some of such datasets are out-of-date or do not have a SPARQL endpoint to perform federated queries. They seem not very stable.
One alternative, is using semantic stable hub like Wikidata as the principal external KB. Wikidata is a general purpose KB but it seems very active.
Could be a valid opportunity or are there other design patterns for LOD usage?