Re: sameas.org

Hi all,
Thanks for the responses.
Nice to know it is being used at all, and some want it, although that means I need to do some work to keep it going :-)

As far as making the data available is concerned, some thought is required.
I need to decide on a format - at the moment it is in a mySQL DB with a schema and access designed for speed of query.
It's 15 years since I looked at it :-)

So neither activity will happen soon, but I won't just let it die, given the responses.

If anyone wants to run a sameas server of their own, we wrote an open source version and put it at
https://github.com/seme4/sameAs-Lite
In fact, it would be very exciting if someone wanted to take on some ownership from this old man :-)
It is over 8 years ago now, which was already 6 years into my retirement.
There has been some bitrot since then, so it would need some maintenance, but its core is good, and is designed for performance.


Yousouf, thank you for the links.
I have a comment or two.
You say one of your hypotheses is to "(2) Test if the sameAs.org service can handle and allow a large number of API calls."
I can't seem to find the answer - can you perhaps tell me briefly what it was please?
And do you have any comparison with doing the same thing on other sites, such as Wikidata or elsewhere?
(sameas.org runs on an unmaintained old shared VM, so it would be nice to know if it was satisfactory for you.)

IN your last sentence you talk about "hops".
You do realise that the sameas.org server is forming the closure/equivalence class of all the pairs it receives, effectively making all possible hops into 1?
That is part of the reason for its existence.

You say "13 The sameAs.org service sends back the equivalence links, if any, plus the requested URI."
This isn't quite true.
You do some work around rewriting the main URIs.
This sort of thing is exactly what the server is designed to do - recommend a particular URI as a "canon".
So if you look at the N3 that sameas.org returns, you will find that neither
http://sameas.org/n3?uri=http%3A%2F%2Fdbpedia.org%2Fresource%2FEidyn
nor
http://sameas.org/n3?uri=http%3A%2F%2Fsouthampton.rkbexplorer.com%2Fid%2Fperson-07113
return the query URI as the subject of the sameAs ref.
You may well have thought that they do, because the server is set up to recommend the Wikidata URI (if it has one), as the canon, and you were using Wikidata URIs as the query!
The recommended canon is server configurable.
You might find life a bit easier if you could get the server to give you the RDF you want, possibly by running a sameAs Lite server (above).

sameAs servers also have a way of deprecating URIs (so they become lookup keys, but are never returned in the results sets).
I wonder if there is an interesting interaction between the canon selecting and deprecating that the server can do, and the CN decisions that you make, especially in line 9 of Algorithm 2.
(I can't recall if the sameAs Lite server has that facility - our Seme4 production version certainly has, and more things.)

That is enough, I think.

Best
Hugh


> On 18 Oct 2023, at 13:20, Yousouf TAGHZOUTI <yousouf.taghzouti@emse.fr> wrote:
> 
> Dear all,
> 
> I also think it is nice to have this service as well as the dataset (a recent use could be found here [1,2]).
> 
> [1]: Content Negotiation in a Decentralised Semantic Context Utilising Equivalence Links: https://dmkg-workshop.github.io/papers/paper9797.pdf [2]: https://github.com/YoucTagh/decentralised-cn
> 
> Best regards,
> Yousouf
> 

-- 
Hugh
+44 7595 334155

Received on Friday, 20 October 2023 15:18:25 UTC