What advantages do RDF storages have against other NoSQL storages?
The one advantage I can see is inference engines integration. But I'm not sure, may be there are other NoSQL storages types with this "feature" enabled.
If you are dealing with data in RDF format, it's better to have an RDF triple/quad store. You will not need to worry about dealing with parsing different serializations, IRIs, encoding IRIs, building your own indexes, implementing a SPARQL query engine, etc.
NoSQL storage systems tend to be key-value stores and have very simple (but hugely scalable) ways to retrieve your data (i.e. direct access or parallel scan (see: HBase)) with limited support for indexes.
Personally, I don't see RDF triple/quad store and NoSQL as two opposed communities/technologies/trends.
Indeed, they might work well together, for example (as inspiration):
- A Project Voldemort storage layer for TDB. -- http://github.com/afs/TDB-V (kudos to Andy Seaborne)
I would like to see more RDF stored in HBase :-)
- http://www.semanticoverflow.com/questions/716/storing-rdf-data-into-hbase
- http://www.semanticoverflow.com/questions/715/translating-sparql-algebra-into-piglatin-script
See also: