An Automated Technique to Support Generation and Validation of Software Specifications

Abstract

This study presents a new technique to support software specification and validation. Such validation by means validate requirement specification itself prior to proceeding the next phase of software engineering lifecycle. The study based on the data types by means of behavioral formal specifications that describe the functional attributes of the data type, but give no insight into how the data type may be implemented; specification is represented by axioms and rules, where axioms reflect the behavioral of the data type for simple data values and rules define their behavioral inductively for more complex data values. These axiomatic specifications are validated for completeness against independently generated data. Our challenge is to ensure that the specifications of abstract data types are valid and reflect all the relevant requirements and constraints for the completeness, consistency, and minimality because any faults that arise in this phase it might cause a negative impact on all subsequent phases and increase the cost of a software product. The methodology that we are follows is defined and specified abstract data types by using axiomatic representation, specified and checked them by using proposed language, and validated generated specification by using proposed validation tool. In order to satisfy the important criteria of a software specification such as simplicity, formality, and abstraction we have proposed a behavioral model that is used to specify abstract data types. This model is as follows: 1) an input space which defines a set of input space signals that abstract data type can accept; which his has two types: a) v operations that are expected to return or produce a value; b) o operations which are do not return or produce a value, but can affect future behavior of abstract data type. The relation between an input space and an output space is defines the name and the type of the abstract data type. Alneelain Specification Language has been proposed to implement the proposed model. Two components of its compiler have been developed; lexical analyser and syntax checker. Six samples of abstract data types have been selected as examples specified by Alneelain Specification Language, and checked its compiler. Alneelain Validation Tool which relied on the output of Alneelain Specification Language has been developed to use to validate completely independent validation data. The results showed the possibility of using Alneelain specification language to specify abstract data types and the possibility of using Alneelain Validation Tool to validate the specification of those abstract data types.