SOFTWARE QUALITY FORECASTING BASED ON FUZZY LOGIC (PART 1. MEMBERSHIP FUNCTIONS OF LINGUISTIC VARIABLES)

Abstract

In today's software development environment, ensuring its high quality is an important issue. One of the effective approaches to predicting the quality of software is the use of fuzzy logic, which allows to take into account uncertainties and uncertainties and fuzziness of estimates of various influence factors. The paper implements an approach to software quality forecasting based on the use of linguistic variables to describe the factors that affect software quality. The structure of the publication includes brief theoretical descriptions of the research stages supported by practical recommendations. Due to the large volume, the research material is divided into two parts, the first of which is submitted for publication.

The author analyzes the literature sources related to the subject of the article. It is noted that publications related to software quality assessment based on fuzzy logic use an incomplete list of factors influencing the degree of software quality. This additionally confirms the relevance of our study, in which the initial set of influencing factors reproduces the characteristics of software quality given in the standard to the fullest extent possible.

The factors of influence on the quality of software, transformed by linguistic variables, whose values are described by linguistic terms that form a universal term set. For each linguistic variable, membership functions are constructed that reflect the degree of correspondence of each term to the real value of the factor. The prerequisites for the feasibility of using a fuzzy inference model to predict software quality are determined. A model of logical inference is designed, which reproduces the hierarchy of relationships between linguistic variables and degrees of software quality. The values of the membership functions of the linguistic variable «testing» are calculated, based on the analysis and processing of matrices of pairwise comparisons, the elements of which are the priority ranks of linguistic variables. The membership functions are visualized in a graphical representation, which determines the degree of involvement of a factor in a certain level of software quality depending on the defined term.