Does modeling real-world objects in object-oriented systems result in well-structured systems?

Proponents of object orientation claim that its ability to model real world objects minimizes system complexity and therefore promotes the development of information systems that can be more easily maintained to meet changing business requirements. In this vein, object oriented design methodologies prescribe using common business domain concepts, as identified by domain experts, as the basis for a system's design. Though the work of cognitive psychologists like Rosch on category theory and basic level categories provides some theoretic rationale for believing that such prescriptions can reduce system complexity, to date no research has actually examined if this is actually so. My research examines whether modeling commonly-used business domain concepts in object oriented systems results in well-structured systems. Specifically my research determines if systems that are consonant, that is systems that reflect a high proportion of commonly used business domain concepts, are more well-structured than those that do not. To this end, I perform a field study on 10 object oriented Smalltalk systems in which I (a) identify consonant and non-consonant portions of a system based on system-designer and end-user input, (b) measure goodness of system structure for these same systems using Chidamber and Kemerer's metrics suite for object oriented design (Chidamber and Kemerer, 1994) and (c) determine if there is a relationship between the two. Research findings indicate that systems that are more consonant tend to have better structure. These findings, however, are not significant due to other factors like the rigorous application of object methodologies during development, and the use of patterns and refactoring that have a greater impact on structure than consonance. At a class level, findings are significant and indicate that classes that are in themselves consonant have poorer structure than those that are not. Both findings are not inconsistent with category theory which, in general, states that the world as a whole is less complex because we use categories to understand it, but states nothing about the relative complexity of commonly-used categories versus not so commonly-used categories.Ph.D.Applied SciencesBusiness educationCommunication and the ArtsComputer scienceEducationInformation scienceUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/131016/2/9825271.pd