The Effects on End-User Query Performance of Incorporating Object-Oriented Abstractions in Database Accounting Systems

Abstract

Object-oriented (OO) advocates assert that concepts such as generalization-specialization hierarchies (GSHs) and abstract data types (ADTs) make information systems more usable by increasing the level of abstraction of the data structure. This study analyzes the effects of GSHs and ADTs on the performance of end-users of accounting information systems. Two groups of experimental participants interactively developed Structured Query Language (SQL) queries to answer ten business questions. The control group (n = 28) used data stored in a traditional relational schema. The experimental group (n = 31) used the same data stored in an OO schema that included GSHs and ADTs. Both schemas implemented the same database accounting model of the sales cycle of a hypothetical company. Participants using the higher abstraction (OO) schema with GSHs and ADTs made fewer semantic errors than did participants using the traditional relational schema. The OO participants also required less time to formulate their queries. These results have several important implications. First, relational database vendors should continue, if not accelerate, their efforts to incorporate OO features such as GSHs and ADTs into their database systems. Second, users of accounting information systems need to improve their understanding of the implications of various data structures on their interactive queries. Third, research should investigate the effects of other abstraction mechanisms, including classification/instantiation and aggregation/decomposition, on query performance.