Re: Given a university's name, retrieve URL for university's home page.

Yes, Freebase can be very good - the disambiguation is good.
And it is often more comprehensive than Wikipedia in terms of URIs.
This is because the Notability is different - for example, they have pages for authors simply because they are authors of books, and Wikipedia would never allow that.
In fact, recently I found a page I had created some years ago had been deleted from Wikipedia, as they trawl back through old pages, back patching the Notability rules; I asked for the text, but when they delete a page it really gets deleted (be warned, ye Wikipedia authors!).
Fortunately, however I managed to get my words back from Freebase, and in fact the page is still there, so I have a URI for the resource.

Just to make clear, on the sameAs.org link you quote:
sameAs.org keeps no strings.
The way that happens is that we query Sindice with the string, and then we use the sameAs.org data to group the URIs coming back.
You could just search on sindice.com, but actually you are right, the two seem to work very well together for this.

Cheers

On 14 May 2013, at 00:20, Daniel O'Connor <daniel.oconnor@gmail.com> wrote:

> I'd actually encourage you to use other services, rather than just querying DBPedia as a single source of truth.
> 
> Why not make use of a service like sameas.org to locate candidates:
> http://sameas.org/html?q=Cambridge+University&x=-1007&y=-244
> 
> ... and crawl those/interrogate various data sets via SPARQL to find instances of a specific University type?
> 
> 
> Alternatively, try Freebase's disambiguation control as a good starting point - I find it more accurate than dbpedia/wikipedia for searching for the right entity/kind of thing.
> 
> Demo:
> http://www.freebase.com/ => Enter "Cambridge University", get the top result of "University of Cambridge", which is a University.
> 
> Widget:
> https://developers.google.com/freebase/v1/search-widget
> 
> You can query freebase via MQL, a javascript flavoured graph query language, or RDF serialisations of the data (http://basekb.com/ ? I'm sure there are lots of others too).
> 
> Sure, it's not all pure semantic web tooling, but its pretty close - and I bet there's more interesting problems to solve after disambiguation of entities :)
> 
>

Received on Wednesday, 15 May 2013 14:40:05 UTC