A METHOD OF TESTING CODE USING NEURAL NETWORK MODELS

Abstract

One of the main aspects affecting the competitiveness of software is its quality, which is a multifaceted and broad concept. Testing as a means of achieving software quality is an urgent problem that contributes to the endless search for solutions and implementation of these solutions, and is fundamental research in the field of information technology, in particular with the use of artificial intelligence.

To ensure and maintain the quality of program code, which is the main factor of software quality, testing is used, which in turn is divided into manual and automated.

Testing as the main way to control the quality of software solutions can be defined as the process of executing a program to check the correspondence between its real and expected behavior. Testing is performed on a finite set of test cases selected in a certain way. In the most general case, the testing process can be represented by a set of four main activities: work planning, test development, test execution, and analysis of software testing results.

To solve these tasks when solving automated testing problems, a method based on the use of artificial neural networks can be proposed. In practice, the use of neural networks shows more than decent results in a wide variety of applications. Since software can be viewed as a function in a multidimensional space, neural network methods can be applied to such a model.

Improving techniques and methods of testing complex software products deserves great attention from software development companies due to the undeniable importance of providing the proper level of quality of the released software.

This article presents an advanced method of maintaining the quality of program code using artificial intelligence, namely neural network models as an error analyzer.