ESTIMATION OF TIME SPENT ON IDENTIFICATION OF INCORRECT SOLUTIONS DURING CLASSIFICATION OF OBJECTS USING COLORED PETRI NETS

Abstract

There is a known procedure for evaluating the effectiveness of the method of replenishing and monitoring the correctness of a hierarchical knowledge base of a decision support system. But the approaches to assessing the quality of the verification method of a software product for the classification of air objects constructed using colored Petri nets are not fully formed.

The purpose of the article is to estimate the time spent on identifying incorrect decisions when classifying objects using colored Petri nets.

The time spent on identifying one unaccounted elementary object depending on the number of features, the number of classes, and the number of generalized objects obtained is given.

It has been established that with an increase in the size of the knowledge base, the time spent on the execution of the algorithm grows polynomially and no worse than in the quadratic dependence. The adequacy of the designed test suites was evaluated using colored Petri nets. The composition of indicators that characterize the quality of test suites was determined. These indicators are indicators of test coverage and reliability of test suites. The main indicator is the completeness of the test coverage requirements, additional – the completeness of the test coverage based on the analysis of control flows. It is revealed how the method of designing test cases affects the performance.

The reliability of the designed test kits was estimated using colored Petri nets. To assess the value of the indicator, many factors, the weight of each factor and the relative errors of the initial information are taken into account. Values of the confidence index are calculated for fixed values of the relative error (= 10%, 20%, 40%, 80%). Carefully put together a variety of factors used in the design of test kits. For comparison were considered: a well-known approach to the design of test suites based on traditional decision tables; method of designing test suites based on the use of colored Petri nets.

The completeness of coverage of functional requirements with test suites at the level of system software testing was evaluated. For a colored Petri net, it makes up to 98%, and for traditional tables - up to 60%.