Linked data server

[Maatary Okouya]

Hi, 

I'm trying to understand properly the use case scenario of stardog. I have the feeling that it pretty much oriented toward an entreprise internal usage rather than public. For instance, would it be recommended to use it as a public linked data server ?

Many thanks, 

Daniel

[Mike Grove]

On Fri, Jun 27, 2014 at 1:43 PM, Maatary Okouya <maatar...@gmail.com> wrote:

Hi, 

I'm trying to understand properly the use case scenario of stardog.

It's a semantic graph database with support for the acronym soup (OWL, SWRL, RDF, etc.) of the semantic web.  There's no single, specific use case it's geared toward.

 

I have the feeling that it pretty much oriented toward an entreprise internal usage rather than public.

I would disagree with that.  While it has many features users would expect from an enterprise class system, it's not designed solely for that usage.

 

For instance, would it be recommended to use it as a public linked data server ?

I'd recommend that everyone use it, but I'm a bit biased.  Whether or not it's the correct fit for *your* use case is not clear.  If your only requirement is that it can provide a SPARQL endpoint, then any RDF database available, Stardog included, can satisfy that.  If you have other requirements, such as OWL reasoning, RDF validation or simply scalability, Stardog is probably a good fit.

If you can give more details about your requirements, we can try to give better guidance.  I also recommend you try it out for yourself.

Cheers,

Mike

 

Many thanks, 

Daniel

--
-- --
You received this message because you are subscribed to the C&P "Stardog" group.
To post to this group, send email to sta...@clarkparsia.com
To unsubscribe from this group, send email to
stardog+u...@clarkparsia.com
For more options, visit this group at
http://groups.google.com/a/clarkparsia.com/group/stardog?hl=en

[Maatary Okouya]

Many thanks for your answer. I get it. Just to let you know briefly, i'm evaluating several option for a new project that i will be starting very soon. I'm choosing the triple store. We will have lot and lots of publication description. 

One aspect is a store that can contain that amount of data, provide simple reasoning like QL, and optionally one that can act as a linked data server. On that last part i have been more tempted in adding my own webServer and to the link data on my own, tapping on the store . This is because i need a great control over the HTML presentation that will be generated and on the information resource (publication) that will be returned. For instance in virtuso customizing the page generated is a bit annoying, i don't know about starDog. 

If you can already give me some indication here that might help. Later on i could provide you with more detail of my scenario (public endpoint, RDFA, semantic powered website etc...)

I'm trying it out the platform right now. although i can't figure out how to publish linked data for now. I have create some triples. The subject resource have the id http://localhost:5820/myDB#...

However when i click on it, i don't get an html page, or the description but  The requested URL /5080/myDB was not found on this server.

Could you guide me on how to publish something simple ?

[Mike Grove]

On Fri, Jun 27, 2014 at 3:20 PM, Maatary Okouya <maatar...@gmail.com> wrote:

Many thanks for your answer. I get it. Just to let you know briefly, i'm evaluating several option for a new project that i will be starting very soon. I'm choosing the triple store. We will have lot and lots of publication description. 

Do you have an idea of how many triples that corresponds to?

 

One aspect is a store that can contain that amount of data, provide simple reasoning like QL,

QL, as well as all the other OWL profiles, are supported by Stardog.

 

and optionally one that can act as a linked data server. On that last part i have been more tempted in adding my own webServer and to the link data on my own, tapping on the store . This is because i need a great control over the HTML presentation that will be generated and on the information resource (publication) that will be returned. For instance in virtuso customizing the page generated is a bit annoying, i don't know about starDog. 

Stardog's web console does not currently provide access to customization of the rendered pages for concepts within the database.

Also, you'll notice that, in the webconsole, a page for a resource is available as part of the fragment of the URL; [1] is an example of the URL of something from the BSBM dataset.  The route for displaying a URL is `/browse/resource/{encoded qnameOrURI}`.  Again note this goes in the *fragment* portion of the URL.

If you need fine-grained control of both the pages and the URLs, you might be better off building your own server for those purposes.  You could use stardog.js [2] to interact with Stardog to build your own site.

 

If you can already give me some indication here that might help. Later on i could provide you with more detail of my scenario (public endpoint, RDFA, semantic powered website etc...)

I'm trying it out the platform right now. although i can't figure out how to publish linked data for now. I have create some triples. The subject resource have the id http://localhost:5820/myDB#...

However when i click on it, i don't get an html page, or the description but  The requested URL /5080/myDB was not found on this server.

Could you guide me on how to publish something simple ?

I'm not entirely sure what you mean by this.  If you want to make an open public SPARQL endpoint, you can pass `--disable-security` to the `server start` command so that the endpoint does not require authentication.  There is also a lot of useful information in our documentation [3].

Cheers,

Mike 

[1] http://localhost:5820/bsbm10m#!/browse/bsbm-inst%3AProductType1%2Cbsbm-inst%3AProductType7%2Cbsbm-inst%3AProductType56%2Cbsbm-inst%3AProductType442%2Cbsbm%3AProduct/dataFromProducer1%3AProduct6

[2] http://clarkparsia.github.io/stardog.js/docs/stardog.html

[3] http://docs.stardog.com

[Maatary Okouya]

On the last point. The idea of linked data is that if you have a resource with a specific URL you should able to dereference it. However, this sounds not possible in Stardog from what you explain on the third point. The resource are not available at the address of the server as is. For instance if i have my website at http://www.myWebsite.org/net/, the resource http://www.myWebsite.org/net/SampleDatasets/John/bookdataset#LordOfTheRing should be deferencable. Basically Does not provide it as it. That is what i mean. Probably trough some URL Rewriting. But again you need something outside, like you have suggest: To have great control on the URL and the Page.

About the size of the triple store, I can't say right now. I'm going to work for a library which wants to make its data available on the Web according to linked data principle. It's the library of an organization not a hug public one. Nevertheless, they think they have million of records, what does that means in triples for now i can say. Also we might further enrich the data with more description to fit Search engine vocabulary such as schema.org.

Btw would it be easy to deploy stardog on Amazon ?

[Mike Grove]

On Fri, Jun 27, 2014 at 3:57 PM, Maatary Okouya <maatar...@gmail.com> wrote:

On the last point. The idea of linked data is that if you have a resource with a specific URL you should able to dereference it. However, this sounds not possible in Stardog from what you explain on the third point. The resource are not available at the address of the server as is. For instance if i have my website at http://www.myWebsite.org/net/, the resource http://www.myWebsite.org/net/SampleDatasets/John/bookdataset#LordOfTheRing should be deferencable. Basically Does not provide it as it. That is what i mean. Probably trough some URL Rewriting. But again you need something outside, like you have suggest: To have great control on the URL and the Page.

Correct, Stardog does not provide access to the resources in the database in that manner.  Getting the URL's for resources correct for every use case is not straightforward out of the box, and a URL rewriting approach as you note, might be a better bet.  

 

About the size of the triple store, I can't say right now. I'm going to work for a library which wants to make its data available on the Web according to linked data principle. It's the library of an organization not a hug public one. Nevertheless, they think they have million of records, what does that means in triples for now i can say. Also we might further enrich the data with more description to fit Search engine vocabulary such as schema.org.

You might find Stardog's full-text search capabilities useful.

 

Btw would it be easy to deploy stardog on Amazon ?

Yes, it requires only Java.

Cheers,

Mike

[Maatary Okouya]

Many thanks