ABSTRACT:

Traditionally, accounting information systems (AlSs) have relied on a chart of accounts (COA) for the categorization of accounting data. Contrary to this tradition, McCarthy (1979, 1982) contended that a COA is an artifact of history and is not required for the storage of disaggregated accounting data. The emergence of the object-oriented (00) paradigm in information systems once again presents a fundamental change in how AIS data can be viewed, stored, modeled and manipulated (Murthy and Wiggins 1993). While the use of this paradigm in the accounting domain has been proposed (Mui and McCarthy 1987; Kandelin and Lin 1992; Chu 1992; Adamson and Dilts 1993), the majority of this research constructs an object model directly from a COA. It is our contention that such direct mapping does not adequately capture the full potential to be realized from 00 and data disaggregation. The purpose of this article is to propose and evaluate a methodology that creates an accounting object model directly from a set of accounting transactions, ignoring the concept of debits and credits. The method is first applied manually and then, using concepts similar to Wu and Dilts (1992), a computerized natural language approach is formulated. Such a semi-automated method can significantly decrease the development time to build an object model, while si- multaneously increasing the model's consistency and auditability. The object model created by the methodology is then measured, using function point analysis, and compared to the corresponding entity-relationship (ER) model. In this example, we demonstrate that the methodology produces a measurably less complex AIS than traditional entity-relationship modeling.