Query Previews in Networked Information Systems: the Case of EOSDIS

Catherine Plaisant, Tom Bruns, Ben Shneiderman , Khoa Doan

Human-Computer Interaction Laboratory
University of Maryland, College Park MD 20742
(301) 405-2768, plaisant@cs.umd.edu


dynamic query, query preview, network information system, visualization, direct manipulation, earth science


Dynamic queries have been shown to be an effective technique to browse information, and to find patterns and exceptions. Dynamic queries involve the interactive control by a user of visual query parameters that generate rapid (100 ms update), animated, and visual displays of database search results. The data of early implementations was stored in local memory to guarantee optimal speed. Problems arise when the data is very large and distributed over a network. To overcome the problems of slow networks and data volume we propose a two-phase approach to query formulation using query previews and query refinements [1]. Preview mechanisms have been used in the past [2] and we believe that their use will be a major component of successful networked information systems interfaces (e.g. [3]).


We use the example of NASA's Earth Observing System Data and Information System (EOSDIS) to illustrate our approach. Soon users will be able to retrieve earth science data from hundreds of thousands of datasets from centers around the country. Classic form fill-in interfaces for EOSDIS exist, but zero-hit queries are a problem and it is difficult to estimate how much data is available on a given topic.


In our prototype interface users first select rough ranges for a few attributes (time, location and parameter) in the query previewer (Figure 1). The impact of their selections is shown on the preview bars which are dynamically updated to reflect the number of datasets available: e.g. when a user selects North America, the preview bars reflect the distribution of datasets for North America. The query preview interface makes use of dataset counts maintained by providers about their holdings, and downloaded when users initiate an EOSDIS session.

When the number of dataset is small enough, the metadata (i.e standardized data about the data) corresponding to the query preview is downloaded for further exploration in the query refinement phase. A second dynamic query interface allows users to specify precise values for more attributes and further filter the result set. The timeline shows the coverage of the datasets, already zoomed on the years selected in the query preview. Large datasets appear at the top, small ones at the bottom, color coded by processing level. An active cursor highlights the corresponding attribute values: location, sensors, campaign, data center etc.

Click here for Figure 1a
Click here for Figure 1b

  • Figure 1 (a and b) The Query Previewer displays on preview bars aggregate data about all EOSDIS datasets. Users learn about the holdings of the collection and make rough selections over a few parameters (here location, parameter and time). The preview bars are updated immediately. The result bar at the bottom shows the total number of selected datasets. In b) North America and 2 parameters are selected. Next, years will be selected and the query submitted to request more details about the datasets.

  • Click here for Figure 2

  • Figure 2: In the Query Refinement users can browse all the information about individual datasets. The result set is narrowed again by making more precise selections on more attributes. Sample data can be viewed before the long ordering process.

  • The prototype shown in the video was implemented in Tcl/Tk but a partial Java implementation is also available (Figure 1 and 2) at: http://www.cs.umd.edu/projects/hcil/

    Research/1995/dq-for-eosdis.html The 2-step approach extends the use of dynamic queries to network environments. It was well received by test users from the scientific community and is being considered for prototyping in the operational EOSDIS system. We are now gathering real data for the prototype and working on data structures capable of handling 100,000 records in the query refinement [4].


    This work is supported in part by NASA (NAG 52895) and by the NSF grant NSF EEC 94-02384.


    More on EOSDIS Project - with JAVA version demos

    1. Doan, K., Plaisant, and C., Shneiderman, B. Query previews in networked information systems,Proc. of the Third Forum on Research and Technology Advances in Digital Libraries , ADL '96 (Washington, DC, May 13-15, 1996) IEEE CS Press,120-129.

    2. Heppe, Edmondson and Spence. Helping both the novice and advanced user in menu-driven information retrieval systems, Proc. of HCI85 , 92-101.

    3. North, C., Shneiderman, B., and Plaisant, C. User Controlled Overviews of an Image Library: A Case Study of the Visible Human, Proc. of the 1st ACM International Conference on Digital Libraries (Bethesda, MD, March 20-23, 1996) 74-82. Project information

    4. Tanin, E, Beigel, R., Shneiderman, B., Incremental Data Structures and Algorithms for Dunamic Query Interfaces. SIGMOD record, 25, 4, Dec. 96, pp. 21-14

    Web Accessibility