Objectives. The objectives of this workshop were to provide an opportunity to (1) assess the IDM program and highlight its many successes, (2) identify areas for improvement, (3) permit researchers to be informed of targeted government strategic directions, (4) advise funders on community-generated research directions, (5) encourage interdisciplinary interaction between the structured database, unstructured database, and information retrieval communities and (6) cooperatively set out a visionary yet realistic research agenda for the future.

Workshop Structure. Plenary presentations and demos provided wide sharing of information and achievements, while workgroup discussions provided ample opportunities for exchanging views and more in-depth debates of selected issues. To fulfill the objectives of transferring information from government and industry to the academic research community, briefings and panels by invitees were presented. The meeting was opened with presentations from the NSF including: Juris Hartmanis, Assistant Director for the Computer and Information Sciences and Engineering Directorate; Michael Lesk, Director of the Information and Intelligent Systems Division; Maria Zemankova, manager of the Information and Data Management Program, and Les Gasser, Director of the NSF Computation and Social Systems Program, CISE Coordinator for Knowledge and Distributed Intelligence. These presentations covered aspects of NSF's policy and programmatic frameworks, areas of IDM research and possible research directions, stressing the need for evaluation and measurement of success for research and technology. Thomas Kalil, from the White House National Economic Council, provided a motivational talk on high-level U.S. Information Technology policy, which stressed the long-term need for research to leverage government funding in making new headway. From industry, the talks by Avi Silberschatz, Lucent/Bell-Laboratories, and David Lewis, AT&T Research, together with the industrial/ academic panel on "Achievements and Vision'' provided perspectives on potential collaborations between academic researchers and computer industry and between information management and data management research communities.

1. Research topics fundamental in making progress towards omni-present, omniscient data can be grouped into the following three broad areas:

  (i) Next Generation Information Access which includes intuitive and efficient access-by-content to multi-dimensional and multimedia/multi-modal data. The focus will move beyond text and business data to multilingual text, the incorporation of advanced linguistically oriented techniques, multidimensional numeric data, spoken documents, music, images, and video management and retrieval. This will imply user-defined Quality-of-Service requirements with respect to data consistency, completeness, relevance, timeliness and cost, and new query types on data, resources and processes, including data mining for implied rules and relationships hidden within data, using both statistical and rule-based techniques.

  (ii) Information Presentation and Visualization which includes collaborative evaluation of modes of interaction, development of evaluation models and strengthening the designer-user and the user-user interface, support for rapid construction of visualizations by non-experts, spatial and temporal integration of data of various scales and resolutions, exploitation of spatial metaphors for interaction with non-spatial information and summarization of results using both text and visualization. This will involve leveraging investments in studies of user needs to ensure that information presentation technologies match user-driven specifications.

  (iii) Large-scale Management and Integration of Distributed and Migrating Data which include application-aware information management and integration of data from autonomous and heterogeneous sources. New languages and middleware are needed to deal with data from different media and of structures such as text, images, signals, presentations, videos, audio clips, dynamic data, scientific data, and software. Further, new techniques are needed for large-scale and efficient storage of multimedia data and metadata, scalable and flexible management of migrating workflows, transactions across autonomous systems and within different computing and communication environments. These include the Internet/Web, high-speed networking environments and wireless and mobile environments.

By articulating a common challenge, the workshop successfully initiated and stimulated future collaborations among the researchers from the different research communities involved in IDM. We anticipate that this will result in new research methods and innovative technologies. The discussions during the workshop have made clear the need for close collaboration with researchers from other disciplines including other sciences, engineering, medicine and the humanities who are involved in building new data intensive applications. NSF currently actively encourages such collaborations with several initiatives. What also became clear was the need for closer collaborations within computer science, for example with the human-computer interaction, artificial intelligence, natural language processing, image processing, process and workflow management, agent and distributed systems, networks and languages research groups. Although implicitly encouraged, such collaborations should be more actively initiated by NSF through both large-scale multi-disciplinary initiatives and inter-division programs for supporting integrative basic research and experimentation.