CSE PG Candidates Nathan LIU, Bin CAO, Luheng HE, and Prof. Qiang YANG Won 3rd Place in 2011 ACM KDDCUP Challenge

A joint team of HKUST and Shanghai Jiaotong University (SHJTU) researchers and students has won the 3rd place in KDD-Cup 2011 Knowledge Discovery and Data Mining Competition which was held in conjunction with the 17th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. The HKUST team consisted of PG candidates Nathan Liu, Bin Cao, Luheng He, and their supervisor Prof. Qiang Yang.

The competition was open to all academic institutes, industries as well as individuals around the world. Since its inception in 1997, the KDDCUP competition has presented practical and challenging data mining problems. The KDDCUP datasets have become benchmarks for the data mining field. The 2011 ACM KDDCUP was organized by Yahoo! Labs.

The theme this year is prediction models for personalized recommendation for musical items based on billions of user ratings amassed at Yahoo! music over the years. When properly analyzed, the raw ratings encode information on how songs are grouped. The hidden patterns link various albums, which artists complement each other, and above all, which users would like to listen to. Good solution to the problem can form the core technology to support personalized music listening as well as other types of recommendation systems.

More than 1,000 teams worldwide participated in the competition. It marks the KDDCUP this year one of the most competitive ones in its history. The HKUST team and their SHJTU peers proposed an ensemble of many large scale machine learning and data mining techniques to tackle the challenge. Their algorithm was also rated the best among all the submissions for single-model category.

Jiang Xiao from SHJTU, Nathan Liu, and Prof Yang at the KDD-Cup 2011 Award Presentation Ceremony