1998 Fiscal Year Final Research Report Summary
Research on an integrated software design methodology
| Project/Area Number |
08680361
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
計算機科学
|
|---|
| Research Institution | Toyohashi University of Technology |
|---|
Principal Investigator |
ISODA Sadahiro Toyohashi University of Technology, Faculty of Engineering, Professor, 工学部, 教授 (50262978)
|
|---|
| Project Period (FY) |
1996 – 1998
|
| Keywords | Software analysis and design / design methodology / software architecture / object orientation |
|---|
| Research Abstract |
1. Object identification method
An experiment is conducted to prove or disprove lvar Jacobsons hypothesis that the use-case control pattern enables to construct highly robust software system. Here the use-case control pattern is defined as a kind of design pattern consisting of three kinds of object stereotypes, boundary, control and entity objects. The material used in the experiment is a series of requirements documents of a traffic information system that is under current use. The experiment goes as follows : first, the first version of the requirements is analyzed and an object model is constructed using the use-case control pattern. Then, requirements changes in the second version is applied to the object model and the amount of influence caused by each of the changes is counted. The second step is repeated as often as the number of versions available. Up to now, this experiment is repeated twice, that is the first, second and third versions have been analyzed. An interim result of the experiment is that Jacobson's hypothesis holds only when there exists some business rule that is assigned to a control object. Furthermore, the roles of control object, which is rather ambiguous in Jacobson's book, Object-oriented software engineering, includes : 1) invoking subordinate boundary and entity object according to an algorithm, 2) handling one or a few business rules concerning to more than one entity object, and 3) embodying transactions.
2. Object-oriented real-world modeling
Object-oriented real-world modeling is found to have two types of modeling techniques, direct modeling and indirect modeling. An identification method of associations is introduced for both of the modeling techniques. It is also shown that OMT, or Object Modeling Technique, is constructed without considering the fact stated above and therefore contains a few defects.
|
