A competition to create an algorithm to predict user ratings of songs was won by a team from National Taiwan University.
The KDD Cup is an educational data mining competition sponsored by Yahoo! Labs which took place for the first time in 2010.
This year’s competition was to create an algorithm to search through over 300 million Yahoo! Music ratings then predict the user ratings of songs. The competition was divided into two tracks.
Track One required the algorithm to learn how to predict how users would rate songs, while Track Two’s task was to separate out songs the users would like based on what they already like.
More than 2,100 entrants took part, and the winning solution on both tracks came from the National Taiwan University who submitted literally at the eleventh hour (23:43:48) on the final day of the competition.
The second and third place winners were:
- Commendo, a team of information scientists from Australia
- The Art of Lemon from China Academy of Science, Hulu Labs, Beijing
- InnerPeace from Shanghai Jiao Tong University, Hong Kong University of Science and Technology
Having won, the teams then had to submit papers describing the algorithm and methods used and present them at dedicated workshops and/or poster sessions during the Association for Computing Machinery’s (ACM) Special Interest Group on Knowledge Discovery and Data Mining (SIGKDD) conference held in August. Several other teams also attended to give talks or present poster papers.
"The KDD Cup offers the data scientist community an unprecedented opportunity to boost scientific efforts in a certain area of research and vertical application,”
said Yehuda Koren, Senior Research Scientist at Yahoo! and KDD Cup 2011 Co-Chair.
In 2010 the National Taiwan University also won the KDD cup and you can read the paper describing the winning algorithm here and this year's submissions can be read following the links from the Workshop Poster Presentations.
The KDD cup seems to be an excellent stimulant for getting people interested in advanced algorithms.
If you would like to be informed about new articles on I Programmer you can either follow us on Twitter or Facebook or you can subscribe to our weekly newsletter.