| Budget Amount *help |
¥1,800,000 (Direct Cost: ¥1,800,000)
Fiscal Year 1990: ¥900,000 (Direct Cost: ¥900,000)
Fiscal Year 1989: ¥900,000 (Direct Cost: ¥900,000)
|
|---|
| Research Abstract |
The aim of Software Engineering is "Automatic Program Generation".The wide semantic gap between requirements description of users and program description executed by a computer, however, make it so difficult. The research attacks from both directions, i. e. formalization of requirements description and high level abstraction of a program, are required to obtain "Automation" of program development.
Metaーprogramming is a kind of programming activity treating programs. Our research consists of 3 basic approaches, modelling of specification description, high level programming using software database, and customization of software parts.
As for modelling of specification description, we provide an abstract requirements model from which many related engineers can extract the object system model from their view points. Even end users can retrieve the information trough natural languageーlike descriptions. The specification system should have a multiーpoints-view since a functional model suits EDP
… More
(Electric Data Processing) while a logic model fits recursive problems, defined with constraints. For specification of the large system, we provide hierarchical specification methods based on parametrization and inheritance. More over, we mentioned reusing of existent specification description to save efforts to describe and verify. The reusing will help a lot for getting a target system since the implementation track can be followed.
The most important feature of software lies in applicability. Then, there are a lot of application fields of software systems. The wide application field makes it difficult to prepare the knowledge required to generate target systems for specification descriptions. In this research, we limited a pplication filed to EDP. The software parts to implement EDP systems are stored in a software database. The efficiency of an object program is apt to be low when we use high level and universal parts. To avoid this problem, we developed a system for parts customization. This is based on techniques similar to optimization techniques in a compiler. More global scope like among functions (or subroutines) should be checked while compilers check only one function. Less
|
