W3C home > Mailing lists > Public > public-lod@w3.org > August 2008

Re: Linked Movie DataBase

From: Yves Raimond <yves.raimond@gmail.com>
Date: Fri, 1 Aug 2008 09:42:34 +0100
Message-ID: <82593ac00808010142o4c27c632o5051edaa274f0558@mail.gmail.com>
To: "Juan Sequeda" <juanfederico@gmail.com>
Cc: "Oktie Hassanzadeh" <oktie@cs.toronto.edu>, public-lod@w3.org, "Mariano Consens" <consens@cs.toronto.edu>


>> We are pleased to announce the release of the preview version of the
>> "Linked Movie DataBase" (LinkedMDB): http://www.linkedmdb.org
>> LinkedMDB aims at publishing the first open linked data dedicated to
>> movies. It currently contains over three million RDF triples with hundreds
>> of thousands of RDF links to other LOD project data sources and
>> movie-related websites. Please check the LinkedMDB website for a description
>> of the data source, a brief overview of the interlinking methodology used,
>> and detailed statistics.

That's really really exciting! Is the underlying data coming from
Freebase, btw? Could your approach be generalised to other data in
Freebase? This would also be really neat :-)

I'd be also glad to try out ODDLinker! I really wonder how well
approximate joins techniques perform. My personal experiments with
really ambiguous resource (Jamendo/Musicbrainz) and string-based
distances on the different literals attached to the resource were
giving quite bad results (33% of false-positives on a 60 resources
sample in the best case), so I tried to come up with a different
approach exploring linked data as long as we can't take an
interlinking decision [1]. It is much slower though, but has the
advantage of giving almost no false positives (it dropped down to 0 on
a 60 resources sample).


[1] http://events.linkeddata.org/ldow2008/papers/18-raimond-sutton-automatic-interlinking.pdf

>> We welcome any kind of feedback and comment from the LOD community, either
>> off-list or on the list (if of interest to everyone).
>> And, if you liked this project, please don't forget to follow this link
>> and vote for us!:
>> http://triplify.org/Challenge/Nominations
>> Cheers,
>> Oktie Hassanzadeh and Mariano Consens
>> admin@linkedmdb.org
>> PS. To know more about us and our research, please check our homepages:
>> http://www.cs.toronto.edu/~oktie
>> http://www.cs.toronto.edu/~consens
Received on Friday, 1 August 2008 08:43:09 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:20:41 UTC