Now mankind is often requiring the software for solving the difficult problems and the high-value software projects are rapidly developed. Therefore, software errors threaten with disasters that lead to human and environmental casualties, disasters, losses of time and finances. Most software failures were due to errors in the requirements specification [1]. So, the success of the software project's implementation (both the timely implementation of the software project within the allocated budget and the implementation of all necessary features and functions [2]) essentially depends on the early life cycle stages, therefore, the urgent task is automation of evaluating the level of early life cycle stages based on the specifications' analysis (assessing the sufficiency of specification's information as a key factor of the software projects' success [3]), and special attention should be focused on the nonfunctional characteristicscomponents of quality of software (according to ISO 25010 [4] these nonfunctional characteristics are: Reliability, Performance Efficiency, Functional Suitability, Compatibility, Maintainability, Portability, Usability, Security). The evaluation of nonfunctional characteristics-components of quality of software according to ISO 25010 standard [4] is carried out as follows: based on the attributes specified in ISO 25023 [5], subcharacteristics are evaluated, which in turn assess the nonfunctional characteristics. Then, with sufficient information for the nonfunctional characteristics in the specification, we will understand the presence in the requirements of all elements (attributes) necessary to determine the nonfunctional characteristics. The analysis of ISO 25010 [4], ISO 25023 [5] standards made it possible to conclude that there are attributes on which more than one nonfunctional characteristic depends, that is, there is a correlation of nonfunctional characteristics according to certain attributes (so according to ISO 25023 [5] subcharacteristics of nonfunctional characteristics depends on 203 attributes, but only from 138 various attributes). The relationships (correlations) between characteristics and subcharacteristics on attributes influences on the significance of such attributes [6]. If the attributes that are part of several nonfunctional characteristics are absent (there is a lack of information), the simultaneous use of these attributes will significantly affect the reliability of the estimates of nonfunctional characteristics-components of quality of the software. In such case, it is important to mitigate the influence of the mutual correlation of nonfunctional characteristics on attributes. Such mitigation is done by identifying common attributes and ensuring their availability. Thus, it was found that information about nonfunctional characteristics is conveniently presented as ontologies [7], which allow to reveal the causal relationships between concepts and to identify missing attributes, their influence on one or more nonfunctional characteristics, and also the correlation of nonfunctional characteristics for attributes. For this, we use the ontologies, which systemate the subject industry, holistic represent the subject industry, identificate gaps in knowledge, visualize missing logical connections, provide basis for intellectual agents [8][9][10].
In order to solve the problem of assessing the adequacy of the volume of information for determining the nonfunctional characteristics in the specifications of the requirements, it's necessary the basic (universal) ontology for software quality domain based on ISO 25010 [4], ISO 25023 [ 5], the components of which will be ontologies for each of the nonfunctional characteristics-components of quality of software. Basic ontologies will display the required information (attributes) that should be available in the specification of software requirements to ensure that its information about nonfunctional characteristics is sufficient. During developing the specific software, in addition to the basic ontologies, it is necessary to have the ontology of this software, which will reflect the requirements in the specifications for the specific software attributes necessary for determining the nonfunctional characteristics. Comparison of developed ontology of the specific software with basic ontologies allows to identify the deficiency (Fig. 1 [3]) or sufficiency (Fig. 2 [3]) of information about nonfunctional characteristics in the specification of requirements for the specific software. (1)
Then: if SAMI =, then requirements' information is sufficient; if SAMI ≠, then information in the SRS is insufficient, and the SRS requires the complement by attributes of nonfunctional characteristics.
In order to determine the structure and filling of the basic (universal) ontologies of software quality domain, set-theoretical and ontological models for nonfunctional characteristics are needed, which is the objective of this study.
Models of nonfunctional characteristics-components of quality of software contain information about nonfunctional characteristics, their subcharacteristics and attributes, taken from the standards [4,5]. Since nonfunctional characteristics models are used to systematize information about nonfunctional characteristics and bring it to the common (unified) view in accordance with standards, then it is expedient to develop such models in the set-theoretical form.
According to the ISO 25010: 2011 standard, such nonfunctional characteristics as Functional Suitability (Fs), Performance Efficiency (Pe), Usability (Ub), Reliability (Rb), Compatibility (Cb), Security (Scr), Maintainability (Mb), Portability (Pb) are functions of several subcharacteristics, then we represent each of the above-mentioned nonfunctional characteristics as a function of the subcharacteristic [7]:
where F_Com -Functional Completeness, F_Cor -Functional Correctness, F_Appr -Functional Appropriateness;
where T_B -Time Behaviour, R_U -Resource Utilization, Cc -Capacity;
where A_R -Appropriateness Recognisability, Lb -Learnability, Ob -Operability, U_E_P -User Error Protection, U_I_A -User Interface Aesthethics, Ab -Accessibility;
where Mat -Maturity, Avb -Availability, F_T -Fault Tolerance, Rcv -Recoverability;
where C_E -CoExistence, Ib -Interoperability;
where Conf -Confidentiality, Int -Integrity, N_R -Non Repudiation, Acb -Accountability, Auth -Authenticity;
where Mod -Modularity, Rub -Reusability, Anb -Analysability, Mdfb -Modifability, Tsb -Testability;
where At the same time, each subcharacteristic of nonfunctional characteristics is a function of certain attributes described in ISO 25023. Depending on the subcharacteristics from the attributes is represented as follows: F_Com = φ 1 (N_o_F, F_I_Cn, F_Aq, F_I_C), (10) where N_o_F -Number of Functions, F_I_Cn -Functional Implementation Completeness, F_Aq -Functional Adequacy, F_I_C -Functional Implementation Coverage; Considering the resulting sets of attributes and sub-characteristics of nonfunctional characteristics, obtained functions of dependencies of characteristics on subcharacteristics and characteristics on attributes, as well as the relationship between the concepts of ontology "depends on", we develop basic ontological models of nonfunctional software characteristics (principles of developing basic ontological models are presented in [7]).
Then, for example, base ontological model of Functional Suitability is:
O Fs = {{F_S, A, I, J, K}, "depends on", {f 1 , φ 1, φ 2, φ 3 }} = ={{fsa 1 ,…,fsa 19 }, "depends on", {f 1 , φ 1, φ 2, φ 3 }},
where fsa 1 = F_S, {fsa 2 ,…,fsa 4 } € A, {fsa 5 ,…,fsa 8 } € I, {fsa 9 ,…,fsa 13 } € J, {fsa 14 ,…,fsa 19 } € K. Ontologies (ontological knowledge base), which are developed by models (41) etc., are filled up based on the information taken from ISO 25010 [4], ISO 25023 [5] standards.
The ontological models of concrete software nonfunctional characteristics have the form, for example, ontological model of Functional Suitability is: O Fs_real = {{fsa 1 ,…,fsa m }, "depends on", {f 1 , φ 1, φ 2, φ 3 }},
where m≤19the number of attributes of subcharacteristics of functional suitability, available in the specification requirements for the specific software, as well as the number of functional suitability subcharacteristics that can be calculated based on available attributes. Ontologies (ontological knowledge bases), which are developed on the models (42) etc., are filled up based on the information taken from the specification requirements for the specific software [3,7].
Practical case of selection of accurate scope of information for evaluation of the nonfunctional characteristics is represented at [3,7].
The urgency of the task of automation of evaluating the life cycle initial phases based on analysis of specifications necessitates the development of software quality models, in particular, the models of nonfunctional characteristics-software quality components.
This article proposed set-theoretic models of nonfunctional characteristics of software used to systematize information on nonfunctional characteristics and bring it to the common (unified) form according to ISO 25010: 2011, ISO 25023: 2016 standards, as well as basic (universal) ontological models and ontological models of specific software nonfunctional characteristics, based on the requirements of ISO 25010 and ISO 25023standards. These models are basis for selecting the sufficient volume of information to evaluate the software nonfunctional characteristics.
The proposed models will become the theoretical basis for the model and method of activity of the intelligence agent on the basis of the ontological approach for assessing the specifications of the requirements for software, the operation of which will be based on the comparison of ontologies. Such the intelligence agent will compare the ontology of the specific software with basic ontology and determine how many and which attributes are not sufficient to determine the particular nonfunctional characteristic, which will make the decision on the adequacy or lack of information on nonfunctional characteristics in the specification requirements to software.