Re: About a network infrastructure

Dear Luciano,

> My research field is p2p networks, DHTs and recently I'm adding  
> semantics to all these. Just a few days ago, I subscribed myself to  
> this list, when I've found the LOD project. I'm following your  
> debate about how to share the data.
>
> I think the LOD project is very interesting and goes right to the  
> point to interconect data on the web, that is why the semantic web  
> idea was created for.
>
> As I'm new in this "world", please, correct me if I'm wrong about  
> something..
>
> One thing that I'm realizing is that the manner that the RDF data is  
> found is very "centralized". Sites like "Ping the Semantic Web" and  
> "Sindice" are fed with location of RDF data by the very users of  
> these data. They're like a big index of RDF bases. Is that so?

you say "found" centralized - this is right to some extent. It is  
published decentralized and you can browse over this decentralized  
graphs, however, you also need some global index for lookups, etc. and  
that's what the projects you named are providing

> What I'm researching, and hope some feedback from you guys about, is  
> how to create a network infrastructure, using the concepts of p2p  
> (structured and unstructured), that all RDF bases can be put on top  
> of.

sounds interesting. You should definitely look at the projects at EPFL  
(Prof. Karl Aberer), e.g. ALVIS search engine [1], which is also based  
on DHTs. They did also some projects towards RDF storage (GridVine).

I think you'll have to take care not to assume people would abandon  
basic REST architecture and use P2P - that won't happen. As you said,  
building an "overlay" sounds reasonable - so, I could imagine to build  
something like Sindice & co based on DHTs.

> In a simple way, what I'm thinking of is an overlay network used to  
> spread and recover semantic data (RDF) naturally.
> To publish your data or query about some data, you just need to have  
> a node that is part of this overlay network (it could be one node  
> per community...) or to know one node that is part of it. This  
> overlay network would be composed by semantic links that  
> interconnect the nodes, having semantic values in this links.

I'm sure, at least providing LOD will rely on HTTP solely. I could  
imagine another infrastructure for corprorate environments, e.g.  
enterprise storage, separate projects, but the Linked Open Data web  
must be based on very simple already existing technologies and tools  
which are already there

Note this is my opinion only.

Best regards,
AndyL

[1] http://cosco.hiit.fi/search/alvis.html


>
> I know it is very vague idea as it was presented here, but I hope  
> you could send me some feedback.
>
> Am I getting all these correctly? Is there someone that is thinking  
> about these issues? What do you think about it? Any advices/critics?
>
> Thank you for your attention.
>
> Best regards,
>
> Luciano
>
> ---------------------------------
> Luciano Bernardes de Paula
> DCA - FEEC - Unicamp
> www.dca.fee.unicamp.br/~luciano
> ---------------------------------
>
>
>      Veja quais são os assuntos do momento no Yahoo! +Buscados
> http://br.maisbuscados.yahoo.com
>
>


http://www.langegger.at
----------------------------------------------------------------------
Dipl.-Ing.(FH) Andreas Langegger
Institute for Applied Knowledge Processing
Johannes Kepler University Linz
A-4040 Linz, Altenberger Straße 69

Received on Monday, 9 February 2009 10:08:56 UTC