Database Querying Through Direct Manipulation

NOTE: Start time for Defense is 9:15 a.m.:

Database systems are tremendously powerful and useful, as evidenced by their popularity in modern business. Unfortunately, for non-expert users, to use a database is still a daunting task due to its poor usability.

This PhD dissertation examines stages in the information seeking process and proposes techniques to help users interact with the database through direct manipulation, which has been proven a natural interaction paradigm. For the first stage of information seeking, query formulation, we proposed a spreadsheet algebra upon which a direct manipulation interface for database querying can be built. We developed a spreadsheet algebra that is powerful (capable of expressing at least all single-block SQL queries) and can be intuitively implemented in a spreadsheet. In addition, we proposed assisted querying by browsing, where we help users query the database through browsing. For the second stage, result review, instead of asking users to review possibly many results in a flat table, we proposed a hierarchical navigation scheme that allows users to browse the results through representatives with easy drill-down and filtering capabilities. We proposed an efficient tree-based method for generating the representatives. For the query refinement stage, we proposed and implemented a provenance-based automatic refinement framework. Users label a set of output tuples and our framework produces a ranked list of changes that best improve the query. This dissertation significantly lowers the barrier for non-expert users and reduces the effort for expert users to use a database.