Guest Editorial Introduction to the Special Section on the 16th International Conference on Data Engineering

HE 16th International Conference on Data Engineering (ICDE 2000) was held in San Diego, California, from 29 February through 3 March 2000. As the first major database conference of the new millenium, it marked the threshold of a new database era with a proliferation of exciting data-intensive, network-centric applications, and increasing penetration of database technology into the software infrastructure of cyberspace. The conference technical program reflected this, covering topics ranging from data mining and knowledge discovery to XML, e-commerce, and mobile computing, while also capturing advances in traditional and still critical engine-technology. The 287 submissions for the ICDE technical program required us to be very selective, but also resulted in an outstanding program. We accepted 41 research papers. In addition, 24 submissions were accepted for poster sessions. The technical program also included industrial sessions, panels, demos, and tutorials. The keynote addresses, by Jim Gray on “rules of thumb” and Dennis Tsichritzis on the “art of computer research” were thought-provoking highlights. Thanks to the active encouragement and cooperation of Farokh Bastani, editor-in-chief of the IEEE Transactions on Data and Knowledge Engineering, we were chartered to bring the best of the ICDE 2000 technical contributions to this special section of TDKE, ensuring that these fine contributions are widely available in an archival journal. This special section contains the six contributions selected as the best papers of ICDE 2000. We solicited extended versions from the authors, to provide full, thorough, and archival quality presentations. The extended papers were then reviewed a second time and we are convinced of their outstanding technical strength and presentation quality. This special section leads off with the paper selected for the ICDE 2000 Best Paper Award. This paper is “Automating Statistics Management for Query Optimizers” by Surajit Chaudhuri and Vivek Narasayya. It addresses the often neglected but pragmatically important area of automating system administration. The specific problem investigated is how to automatically select the columns of a relational database to maintain in histograms. An elegant solution is devised for choosing those columns that are essential to effective query optimization, and the solution is experimentally studied. The second paper, “A Foundation for Conventional and Temporal Query Optimization Addressing Duplicates and Ordering” by Giedrius Slivinskas, Christian Jensen, and Richard Snodgrass, develops a general algebraic framework that includes powerful aggregation and sorting constructs that integrate conventional and temporal query processing. The framework is sufficiently flexible to enable it to accommodate different architectual approaches to the support of temporal functionality. Our special section would be incomplete without papers from the increasingly important field of data mining. The first data mining paper, “DEMON: Mining and Monitoring Evolving Data” by Venkatesh Ganti, Johannes Gehrke, and Raghu Ramakrishnan, won the ICDE 2000 Best Student Paper Award. It addresses the problem of incremental mining where the underlying data gradually changes over time and it proposes “incremental” mining algorithms that exploit this gradual evolution for increased efficiency. The incremental approach is applied to a range of problems including discovery of association rules and clustering. The other mining paper, “Finding Interesting Associations without Support Pruning” by Edith Cohen, Mayur Datar, Shinji Fujiwara, Aristides Gionis, Piotr Indyk, Rajeev Motwani, Jeffrey Ullman, and Cheng Yang, reconsiders association rules, e.g., on market baskets or web traffic logs. The new insight, motivated by these discovery applications, is that requiring a minimum support is counter-productive. But, removing this requirement interferes with pruning the search space. The paper develops a family of randomized algorithms that find the most interesting rules, regardless of support, with probability converging to one. Our final two papers fall into the area of spatial and multimedia data management. The first one, “Multiple Similarity Queries: A Basic DBMS Operation for Mining in Metric Databases” by Bernhard Braunmueller, Martin Ester, HansPeter Kriegel, and Joerg Sander, aims at improving the performance of similarity queries, e.g., finding the k nearest neighbors of a given data object, on spatial and multimedia