giraph-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dionysis Logothetis <>
Subject Library of ML and Graph Mining algorithms
Date Fri, 28 Feb 2014 15:36:20 GMT
Hi all,

We were going to send an e-mail to the giraph mailing lists soon, but
it's good to see you already discovered the site. is a
general project on tools for ML and graph mining that we started at
Telefonica Research and for the most part revolves around Giraph. In
this context, we've been developing a library of algorithms for Giraph
that we call Okapi.

We started developing and using Okapi internally for research
purposes and we thought it'd be good to open source it. We've been
working on this for some time, and after some final testing, we're
ready to release a first version of it (a couple of the algorithm
implementations are still under testing).

So far we've developed some state-of-the-art machine
learning/recommendation algorithms and graph mining algorithms. In
general, we'd like to build more applications in these areas and
potentially new categories of algorithms that fit well on top of
Giraph. If you have any suggestions, let us know!

As you'll see we're releasing Okapi under the Apache license. We hope
to get contributions from the Giraph community and make this a rich
and useful library for everybody. This could be in the form of new
algorithms, fixes etc. In the long term we'd like Okapi to become an
Apache project if this makes sense, something similar to Mahout. If
you have any feedback about this or would like to get involved more
please do get in touch with us.


On Fri, Feb 28, 2014 at 9:26 AM, Pavan Kumar A <> wrote:
> +1 :)
>> Date: Fri, 28 Feb 2014 00:00:55 -0800
>> From:
>> To:
>> Subject: Re: Fyi: Graphos
>> Thanks for the link, this looks pretty neat.
>> On 2/27/14, 11:29 PM, Sebastian Schelter wrote:
>> > Hi,
>> >
>> > It seems a team from Telefonica built a machine learning library on
>> > top of Giraph:
>> >
>> >
>> >
>> > Looks pretty interesting to me :)
>> >
>> > Best,
>> > Sebastian

View raw message