Visually querying object-oriented databases

Abstract

As database requirements increase, the ability to construct database queries efficiently becomes more important. The traditional means of querying a database is to write a textual query, such as writing in SQL to query a relational database. Visual query languages are an alternative means of querying a database; a visual query language can embody powerful query abstraction and user feedback techniques, thereby making them potentially easier to use. In this thesis, we develop a visual query system for ODMG-compliant object-oriented databases, called QUIVER. QUIVER has a comprehensive expressive power; apart from supporting data types such as sets, bags, arrays, lists, tuples, objects and relationships, it supports aggregate functions, methods and sub-queries. The language is also consistent, as constructs with similar functionality have similar visual representations. QUIVER uses the DOT layout engine to automatically layout a query; QUIVER queries are easily constructed, as the system does not constrain the spatial arrangement of query items. QUIVER also supports a query library, allowing queries to be saved, retrieved and shared among users. A substantial part of the design has been implemented using the ODMG-compliant database system O₂, and the usability of the interface as well as the query language itself is presented. Visual queries are translated to OQL, the standard query language proposed by the ODMG, and query answers are presented using O₂ Look. During the course of our investigation, we conducted a user evaluation to compare QUIVER and OQL. The results were extremely encouraging in favour of QUIVER.