RDF graphs usually are not only self-contained data structures. Visualising just the information explicitly encoded in a given RDF graph yields unsatisfactory results. There are formalized semantics (RDF, RDFS, OWL) that help dealing with the data.
Ontologies define specific terms using these semantics. For instance, the FOAF Vocabulary defines terms relating to social relations between people. By taking into consideration the FOAF ontology, the triple <Alice> foaf:knows <Bob> . implies, that <Alice> and <Bob> are both persons (i.e. of type foaf:Person).
This kind of background knowledge should not be included directly into graph (i.e. as a triple <Alice> rdf:type foaf:Person), but could still be considered for styling (i.e. Alice could be rendered as an user icon. By using a reasoner to apply styling rules, Odruba can leverage this background knowledge.
Typically, Odruba aims to visualize small to medium sized RDF graphs. One use case is to select a small sub-graph of a larger knowledge base, i.e. all movies of a certain director out movie database.
Visualisation libraries on modern computers can display force-directed graphs with some 1000 nodes (though limitation apply). However, Odruba does not primarily focus on visualising graphs of that size.
Not exacly. Since Odruba uses a reasoner and rules, an 'extended' use case is to define certain domain specific rules that might result in a specific color to be applied to a node. The rule-based approach is flexible enough to indicate missing or critical information visually.
Yes, and no. Odruba employs a client-server architecture to connect the Java backend with the Javascript front-end. It can be useful to fetch certain information via an HTTP request. However, the Odruba-API will change somewhat to support multitenancy. Expect a better documented API description in the near future.
No, Odruba is in early alpha stage. Not too much work has been put into usability, deployment or interaction with an existing triplestore.
Right now, Odruba is in an early alpha-stage (see previous question). However, the Jena libraries provides components that enable SPARQL queries quite easily so you can defininitely except this to be included.