KDD

KDD-2003, the Ninth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, was held in Washington, DC, U.S.A., on August 24-27, 2003. KDD is the leading international forum for the exchange of research results and practical experience in field of knowledge discovery and data mining. As the mountains of data available to organizations and individuals continue to grow without limit, and the need to extract useful knowledge from them becomes ever more intense, scientists, government workers and business people turn to the KDD community for solutions. The volume you have in your hands (or on your screen) contains a snapshot of a year of developments in this field; we hope you find it useful and rewarding.The KDD-2003 technical program featured two parallel research tracks and an industrial/government track. The latter was the result of expanding the scope of KDD's industrial track to reflect the increased importance of knowledge discovery and data mining in government and vice-versa. The program also featured three keynote speakers, nine workshops, seven tutorials and two panels. The 2003 KDD Cup competition focused on mining citation networks and data cleaning in large bibliographic repositories. Dozens of exhibits from vendors and other organizations added to the ferment, and underscored the conference's dual role as an industry and academic event.Continuing its tradition of collocation with related conferences, this year KDD was collocated with ICML-2003, the Twentieth International Conference on Machine Learning. The two conferences held a joint session on the first day of KDD (last of ICML), which featured a selection of papers from the two conferences and a joint keynote speaker. COLT-2003, the Sixteenth Annual Conference on Computational Learning Theory, was also collocated with KDD and ICML.We received a large number of submissions, and the selection process was extremely competitive. Each paper was independently reviewed by three members of the program committee for originality, significance, technical quality, and clarity of presentation. This was followed by discussion among the reviewers and final decisions. Of the 258 research track submissions received, 34 were accepted as full papers for oral presentation, and 36 were accepted for poster presentation (13% and 14% of submissions, respectively). The industrial/government track received 40 submissions, of which 12 were accepted for oral presentation and 10 were accepted for poster presentation (30% and 25%). Additionally, ne research submission was reassigned to the industrial/government track, and accepted for oral presentation there.The resulting program was notable for its diversity and vitality. Alongside traditional KDD topics like classification, clustering, frequent sets, scalability, and temporal data, we also saw papers in rapidly growing areas like graph and relational mining, data streams, semi-structured data, and privacy. Application areas included the Web, bioinformatics, health care, marketing, crime-fighting, and many others.