Re: Knowledge Graph links to Freebase

On 3 June 2012 16:05, Andreas Thalhammer <andreas.thalhammer@sti2.at> wrote:

> Dear all,
>
> I want to take the opportunity to present my findings after analysing the
> patterns of Google's Knowledge Graph.
>
> Each summary has a unique identifier. This identifier is used when linking
> to other entities, e.g.
>
> H4sIAAAAAAAAAONgVuLQz9U3MKs0Lg**IAXXSnTQwAAAA
>
> stands for the summary of Steve Jobs.
>
> The URI to get the summary is
> https://www.google.com/search?**hl=en&sa=X&q=steve+jobs&stick=**
> H4sIAAAAAAAAAONgVuLQz9U3MKs0Lg**IAXXSnTQwAAAA<https://www.google.com/search?hl=en&sa=X&q=steve+jobs&stick=H4sIAAAAAAAAAONgVuLQz9U3MKs0LgIAXXSnTQwAAAA>
>
> I found out that this key is created with 2 tools, namely base64 and gzip.
>
> We can use this key to find out what the original content was (hoping to
> find a link to Freebase).
>
> The way to go is the following:
>
> 1. Store the identifier above (H4sl..) in a file, e.g. id.b64
> 2. console:$ base64 -d id.b64 > id.gz
> 3. console:$ gunzip id.gz
> 4. console:$ cat id
>
> --> /m/06y3r
>
> Now, add the freebase namespace to that:
>
> http://rdf.freebase.com/ns/m/**06y3r <http://rdf.freebase.com/ns/m/06y3r>
>
> This redirects to:
>
> http://www.freebase.com/view/**en/steve_jobs<http://www.freebase.com/view/en/steve_jobs>
>
> Have a nice Sunday!
>

Nice find.

I wonder what the motivation for this pattern is ... I tend to use sha1 in
my hashes, but I guess that's lossy ...


>
> Andreas
>
> --
> Andreas Thalhammer
> PhD Student
> Semantic Technology Institute
> University of Innsbruck
> http://www.sti2.at/
>
> phone: +43 (0) 512507 6454
> email: andreas.thalhammer@sti2.at
>
>
>