Semi-Automatic Generation of Software ...

Research>Semiautomatic ...

Home

Curriculum

Research

Semi- Automatic Generation Of Software

Functional Architectures

General Information on Research

Publications

General Information on Research:
This research aims to eliminate a time consuming step in the software life cycle: the building of the software representation, by translating the requirements specifications into a functional architecture model.
To achieve this objective, the requirements specification is first translated into a simplified grammar which contains two types of elements: tokens and keywords. Tokens are the objects upon which the software operates, the functions in the software and the interfaces. Tokens, which are unique names provided by the user, provide an excellent means for drawing relational inferences. Keywords, on the other hand, are the operators. The next step is the chaining of the requirement statements to develop a functional architecture. This functional architecture captures both functional and non-functional requirements and can be used to evaluate the failure mode probabilities of the functions in the architecture using a Bayesian quantification framework.
The above approach is applied manually to two requirements specifications. The algorithm for the semi-automate translation of the simplified grammar specification into the architecture is also developed. The results from this work can be used to develop prototype software to automate the process. An additional benefit to this work is that the architecture can be used as an input to tools for model-based automated black box testing.

Publications :
1. A. Anant, An Approach To The Semi-Automatic Generation Of Software Functional Architectures, Master’s Thesis, University of Maryland, 2000
2. A.Gupta, Student Poster accepted at ISSRE, 2000.
3. A.Anant, C. Smidts, An Approach to the Automatic Generation of TestMaster’s Finite State Machine Graph, Proceedings of Teradyne Users Group, San Diego, April, 2000.