Development of an intelligent agent for analysis of nonfunctional characteristics in specifications of software requirements

Abstract

One of the urgent present-day tasks consists in ensuring an automated analysis of specifications of software requirements for sufficiency of information on nonfunctional component characteristics of software quality. Analysis of known intelligent agents based on the ontological approach has shown that these agents do not solve the problem of quantifying sufficiency of information in the specification of software requirements for determining nonfunctional characteristics of software. The objective of this study was to implement the intelligent agent based on the ontological approach for analyzing information on nonfunctional characteristics in specifications of software requirements. A model of intelligent agent activity has been developed based on the ontological approach for evaluating specifications of software requirements. It reflects features of assessing information sufficiency for determining nonfunctional component characteristics of software quality. The developed model is a theoretical basis for implementing the intelligent agent based on the ontological approach for evaluating specifications of software requirements. The intelligent agent based on the ontological approach has been implemented for evaluating information on nonfunctional characteristics in the specifications of software requirements. The implemented agent forms conclusions on sufficiency or insufficiency of information about nonfunctional component characteristics of software quality in the specification of requirements to actual software. In addition, it quantifies the level of information sufficiency in the specification of requirements to actual software for determining each of nonfunctional characteristics of software and determining all nonfunctional component characteristics of software quality in aggregate. The agent provides a list of attributes that should supplement the specification of requirements for increasing the level of sufficiency of its information as well as visualization of gaps in knowledge of all nonfunctional component characteristics of software quality. The results of functioning of the implemented agent have shown an increase in the level of information sufficiency in the specification of software requirements. The developed intelligent agent makes it possible to partially eliminate human participation in information processing, avoid loss of essential information and minimize occurrence of errors at the early stages of the software life cycle.