University of Maryland, Computer Science Dept., Dissertation
HCIL-2001-22, CS-TR-4292, UMIACS-TR-2001-70, ISR-TR-2005-18
Companies, government agencies, and other organizations are making their data available to the world over the Internet. These organizations store their data in large tables. These tables are usually kept in relational databases. Online access to such databases is common. Users query these databases with different front-ends. These front-ends use command languages, menus, or form fillin interfaces. Many of these interfaces rarely give users information about the contents and distribution of the data. This leads users to waste time and network resources posing queries that have zero-hit or mega-hit results.
Generalized query previews forms a user interface architecture for efficient browsing of large online data. Generalized query previews supplies distribution information to the users. This provides an overview of the data. Generalized query previews gives continuous feedback about the size of the results as the query is being formed. This provides a preview of the results.
Generalized query previews allows users to visually browse all of the attributes of the data. Users can select from these attributes to form a view. Views are used to display the distribution information. Queries are incrementally and visually formed by selecting items from numerous charts attached to these views. Users continuously get feedback on the distribution information while they make their selections. Later, users fetch the desired portions of the data by sending their queries over the network. As they make informed queries, they can avoid submitting queries that will generate zero-hit or mega-hit results.
Generalized query previews works on distributions. Distribution information tends to be smaller than raw data. This aspect of generalized query previews also contributes to better network performance.
This dissertation presents the development of generalized query previews, field studies on various platforms, and experimental results. It also presents an architecture of the algorithms and data structures for the generalized query previews.
There are three contributions of this dissertation. First, this work offers a general user interface architecture for browsing large online data. Second, it presents field studies and experimental work that define the application domain for generalized query previews. Third, it contributes to the field of algorithms and data structures.