Apache moves forward with machine-learning project

By Alex Handy

November 25, 2009 — (Page 1 of 2)
When Bradford Cross and the FlightCaster team launched their startup earlier this year, they had already spent months building systems that could learn from the FAA and airline information. Elsewhere, NetFlix paid an independent developer a million dollars to build a new recommendation engine for their movie rental system. Both of these systems were built on machine learning, also known as artificial intelligence, but both of these learning machines were built from scratch.

AI has been around since the early days of software development, but it is only recently that machine learning has begun to become a viable business tool. As such, a group of Apache Lucene developers decided earlier this year to build out the tools needed to create competent machine-learning systems on top of the big data systems being created at Apache.

In mid-November, the Apache Mahout project reached version 0.2, and with it comes some of the first building blocks for creating systems like the NetFlix recommendation engine. While this is only the second release for the project, and even its creators say it is still in its infancy, Mahout is functional and can be put to use now.

Mahout is an attempt to build libraries and algorithms useful to developers attempting to create machine-learning applications. It is typically used on top of Apache Hadoop, where large data sets can be stored and analyzed, and where the actual trained machine-learning routines can be executed across that data.

Grant Ingersoll is one of the cofounders of the Mahout project. He's also cofounder of Lucid Imagination, a company founded at the beginning of this year and dedicated to providing commercial support and service to the Apache Lucene project.

He said that Mahout “grew out of some frustrations all of us had had with various other machine-learning packages. There are others, and some of them are quite good, but we didn't feel they all address the issues important to us."

Those important factors were the existence of a vibrant community, the scalability of the solution, and the licensing of the tools under a business-friendly license.

Related Search Term(s): Apache, Hadoop, Mahout

Pages 1 2

Share this link: http://sdt.bz/33941

Technorati

Digg

Slashdot

Facebook

Friendfeed

Twitter

del.icio.us

The Apache Software Foundation Announces Apache ACE as a Top-Level Project Open Source OSGi software distribution framework especially suited for the Cloud and embedded computing markets.

Apache graduates Hadoop incubator projects Foundations announces project promotions, and pushes Hadoop, Lucene sub-projects into the incubator

Five Apache projects you may not know about From content management to XMPP to extracting metadata, these projects may be worth a second (or first) look

The Apache Software Foundation Announces Apache ACE as a Top-Level Project
Open Source OSGi software distribution framework especially suited for the Cloud and embedded computing markets Read More...