Interactive query formulation techniques for databases
Guogen Zhang · 1998
Building a query tool for end-users to query a database has been an important aspect of the development of database technologies. To advance the state of the art of this technology, a set of techniques for interactive database query formulation is developed in this dissertation and a prototype system is implemented. The techniques include the following: (1) High-level query formulation for simple queries; (2) Incremental query answering for complex queries; (3) Associative query answering for relevant information; and (4) Semantic query guidance for incremental query sessions. This set of techniques is aimed to make formulation of ad hoc queries, including both navigational and aggregational queries, easier for end-users. Our approach to query formulation is based on a semantic graph model, which is a semantic representation of the data in the database augmented with a probabilistic information measure. This graph model can be semi-automatically generated from the database schema. The high-level query formulator allows users to formulate their queries by specifying simple requests and constraints, without having to use the database query language. The formulator completes queries based on user input, ranks candidate queries according to probability, and uses English-like query descriptions for resolving ambiguity during the formulation process. Incremental query answering allows users to obtain the necessary information by constructing a series of simple queries rather than being burdened with query syntax and structures for complex queries. Furthermore, associative query answering provides additional relevant information for queries; query guidance provides suggestions for subsequent queries at each step in an incremental query session. All the techniques provide users with useful assistance in finding their desired information. Case-based and probabilistic reasoning techniques are used for associative query answering and query guidance. The query formulation techniques can be incorporated with a GUI or with a natural language interface for voice input.