I'm not sure if the sparql extension in itself would make Janus/Titan
equivalent to Rya-- Rya provides a no SQL  backend for sail graphs and a
sparql extension wouldn't necessarily ensure you can interact with Titan as
if it was a sail repository.  I imagine the sparql support in
Titan/Janus would be secondary to the native query language support for
property graphs, kind of like how Rya property graph support is secondary
to its rdf support.  So it might not be as efficient as Rya for straight
sparql queries, and you likely wouldn't have support for things like
geospatial or free text indexing.   It probably wouldn't provide any
reasoning support.  What I remember of Titan, it used external indexing
like elastic search to support certain types of queries, so cell level
security there may be another concern.

Do they have a dev list you could move this discussion to?  You might get
more meaningful responses from people who actively develop on Titan/Janus
as opposed to here where few of us really have used it extensively.

On Friday, January 13, 2017, Sam Chance <[email protected]> wrote:

> FWIW,  I say +1 for the SPARQL extension. But I am a  huge evangelist for
> semantic technologies.
>
> On Jan 13, 2017 3:29 PM, "John Smith" <[email protected] <javascript:;>>
> wrote:
>
> > Google, IBM, Hortonworks, and other companies have just decided to back
> > Janus Graph, a scalable Graph DB.  It is essentially a continuation of
> > Titan DB (if you are familiar with Titan).
> > http://janusgraph.org/
> >
> > I wonder if it would make sense to create an Accumulo backend
> > implementation for Janus, and a SPARQL extension.  Such additions would
> > largely make Janus equivalent to Rya (less cell level security).  As
> Janus
> > already supports Casandra and since Casandra and Accumulo are rather
> > similar, I imagine it wouldn't be too hard to do.  Also there already
> > exists a SPARQL extension for Titan, we would simply need to update and
> > extend it for the latest version of Janus.  What are your thoughts?
> >
>

Reply via email to