ACM KDDCUP 2005 Data Mining Contest: Champion


A HKUST research group has won the championship of the KDD-Cup 2005 Knowledge Discovery and Data Mining competition which is held in conjunction with the Eleventh ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. The team consists of PG students and postdoctoral fellows Dou Shen, Rong Pan, Jiantao Sun, Junfeng Pan, Kangheng Wu and Jie Yin, and their supervisor Prof. Yang Qiang, a faculty member of the Department of Computer Science at HKUST.

The ACM KDDCUP is the most rigorous annual competition in the field of machine learning and data mining. The competition is open to all academic institutes, industries as well as individuals from around the world. Since its inception in 1997, the KDDCUP competition has presented some very practical and challenging data mining problems. Many more researchers and practitioners participate in the contest over the years, and the KDDCUP datasets have become benchmarks for the data mining field. ACM KDDCUP Contest 2005 was held between May and July 2005 by the ACM Special Interest Group on Knowledge Discovery and Data Mining which was aimed at solving a difficult and practical problem in Internet search, for the classification of 800,000 realistic Internet search queries. A good solution to this problem will help improve the current internet search and information retrieval technology. Many teams around the world completed event and 37 solutions were submitted. Solutions include the design and implementation of algorithms based on artificial intelligence, statistics, database and information retrieval technologies. This year's tasks are particularly challenging due to the large size of the search problem and the lack of training data that are usually given in data mining problems. The HKUST team used an ensemble based search method that applies machine learning algorithms innovatively. They won all three prizes set up by the KDDCUP organizers, including the "Query Categorization Precision Award", "Query Categorization Performance Award", and the "Query Categorization Creativity Award". An award ceremony was held on August 21, 2005 in Chicago, USA.

See 2005 ACM KDDCUP Official Web Site for more information.