Evaluating Software Quality Metrics for Enhanced Software Management and Engineering

How can the system's quality, programs, and applications be defined? It is defined as a grade attachment to the system, this grade reflects a certain degree of the features and characteristics of that system, which requires designers CMMI v2.0 to redesign or improve the system to solve a specific issue [4]. Software feature is well-defined by the International Software Testing Qualification Board (ISTQB) as the sum of topographies and functional features of a particular creation that affect fulfilling the declared and implicit requirements of the overall product [12]. IEEE 1061:1998, Metrics are defined as a function, the output of this function expresses the degree of system quality, and the resulting value declares the system characteristics, in several stages if it is in an intermediate stage during the development of the program, or in its final stage of development [1].

The ISO/IEC 25000 standard, is distinct as a “quality archetypal”, and deals with several related characteristics, and the interrelationship between them, and provides specific mechanisms to meet quality requirements [13]. Each model consists of characteristics to determine quality, and other sub-characteristics such as reusability, reliability, etc. To contribute effectively to improving product quality, it will be through the use of standards that fit and affect the quality characteristics.

The ISO 9126 standard, which represents product quality standards for software engineering systems, has been analyzed and revised by ISO 25010: 2005 [14]. The ISO 25010:2005 standard provides a comparison of the following three newly classified quality models, namely Quality in Use (QinU), which is one of the standards for measuring the quality of web applications, represents the user’s point of view, and is defined as the degree of measuring the efficiency and performance of the system, and contains measures including system analysis and risk analysis. Software Product Quality (SPQ), is one of the widely used quality standards, such as ISO 25000 Square and ISO 9001, which provide several standards for measuring the quality of systems and the quality management system. In addition, several institutions have developed their quality management systems, such as OMC S.p.A. and EHSQ quality management system, Data Quality (DQ), is a measure that indicates the state of qualitative or quantitative information, and if the DQ score is high, it indicates that the system is “suitable for its intended uses in operations, decision-making, and planning.” DQ provides other metrics to measure data management integrity, accuracy, consistency, timeliness, and validity. It provides mechanisms to protect data from changes and exceptions through quality control to ensure the reliability and usefulness of the data [15]. The quality of any application is defined as the degree of characteristics related to the application, and each characteristic has a degree of correlation. Calculate the correlation characteristics of each model of the application with other models in ISO 25010 [15] and ISO 25012 [16]. In a previous analysis, the definition of SQM was mentioned as providing several measurement metrics, including IEEE 1061:1998, as it represents the degree of measurement of the designers’ orientation towards several characteristics related to the documents related to the project, as well as the codes on which the project is based [1]. Metrics are usually classified by designers on several levels. These levels represent the application life cycle. Requirements are the first stage of system design, as metrics designers have included these metrics at the first stage, in the early stages of the system life cycle, user requirements are divided into functional and non-functional requirements, so metrics have been created to measure functional and non-functional requirements. The second stage of the application life cycle is coding in object-oriented programming is divided into classes, subclasses, methods, and access fields, and metrics have been created to measure the coupling and cohesion between classes and subclasses. The third stage is design; several metrics have been created at the highest code level of design. The last stage is the testing level, at this level there are many proposed metrics and they are general and are used to measure the degree of generality of the system.

In SQM management, we have some paradigms specific to the code level [16]: Which can be defined as the lack of coherence in method (LCOM) as the number of classes in which a local relationship is relative to the non-intersected relationship with the global method's classes, the local instance variable is considered to be related to the degree of local methods correlation. If we want to measure the number of classes with ancestors that reflect the depth of the inheritance tree (DIT), which can be affected by this measurement. The tree consists of the nodes and the path between the nodes, where the node represents the class, and the distance of any track between any two bumps is the maximum depth, representing the DIT scale from the root of the tree to the leaves. The following two are class-level standards, LCOM and DIT. Another important measure is the detection of Defect Density (DD) in SQM, and its task is to detect the defect generated during program execution during its life cycle. When calculating the amount of flaws and dividing by the volume of the produce, Defective Density Delivered” (DDD) is extracted. These two metrics are project-level test DD and DDD metrics.

Examples of SQM, which provide several attributes of quality presented in ISO 25010 and ISO 25012, can be mentioned: The first one is to measure the system failures based on the mean time between system failures (mean time between failures metric (MTBF)), is a measure established for application maintenance and represents the average execution time for a system or component to operate without failure. It is a significant degree and represents the amount of reliability of the system. The MTBF formula is designed by distributing the entire operating phase by the number of failures that occur during that period.

At the same time, MTBF be able to be classified as many quality features at the application level and its relation to SQP. The client's opinion and opinion are a prominent indication of the use of the system and knowledge of the quality metrics of the Qin [4]. Data quality often depends on quality aspects such as refactoring, and compatibility; on the other hand, integrity, recoverability, traceability, and portability are to the originality of the data in the program [3]. Another attribute is measuring portability which is defined as the transfer of source code from one software to the detached atmosphere, the ER metric is a tool to quantity system source code and create an environment target compatibility.

2.1 Related work

The literature review and the study related to SQM were discussed. Table 1 contains the reviews and literature of the studies within the scope and content of their differences. Table 1 covers the studies of the articles related to the software system quality metrics and measures. Otherwise, the literature review in this article differs from the one mentioned in the table. Table 1 offers a comparison made to include the previous works extracted based on papers titles and abstracts of research, methods used, and practical implications. In addition, our work focused mainly on measures that were applied at all levels of development that aim to increase the product quality of the program for the types of software and various applications. In the study of Fenton and Bieman [17] identified the most used metrics and measurable properties in OO software. Highlighted most used statistical techniques/models for quality evaluation. Ultimately, for systematic mapping, the aim of which is to identify the metrics of systems, applications, and statistical features, in this study 79 manuscripts out of a total of 8231 were analyzed.

The study presented in Table 1 shows that the only study similar to ours is by Santos et al. [18], which focuses on metrics related to object-oriented programs for the years 2004-2013. Through the study by Erdem et al. [19], we noticed that it is relevant to our current study. In the study, the analysis is conducted by selecting a set of system patterns that support the architectures. In the same study, research questions related to publishing and a different number of types of research were examined. Where the RQs were taken for the same work, principles related to the research, and various published articles, defined the descriptive models, taking into account the common elements in system quality models and software engineering supporting quality models. Also, there is not enough data and information available about SQM, as we referred to in this article. Then, the RQs are different within the scope and method of our current study. The years between 2009 and 2023 were chosen for detailed study and analysis because SQM contains diverse and wide-ranging content. Therefore, this study expands the scope of research questions RQs on SQM and accuracy.

In this paper by Santos et al. [18], quality measures of importance in agile software development were reviewed and discussed, as well as their applicability and challenges in assessing productivity and quality in an agile environment. Maintainability prediction by using machine learning approaches has been reviewed related to object-oriented software in the study by Erdem et al. [19], the study from 1991 to 2018 points out little indicates in the maintainability prediction area, in contrast, many other software quality metrics. In the study by Olorunshola et al. [20], machine learning is one of the useful techniques for quality metrics, used by AI designers in combination with software quality metrics, defect detection, and prediction of efficient system models. The goal of this research is to judge and analyze software quality measurement, helping developers identify weaknesses in the software. The study of Aguileta and Gómez [21] used global information and communication technology (ICT) that performs estimation in quality practice software and the output indicates enhanced capacity and product metrics quality. The empirical and mixed approaches used (repository analysis, interviews for consultations, and overview article) induced the benefits of ICT enterprise for optimizing measuring structure. In the study of Keele [22], Human-computer interaction (HCI) is adapted to correlate between the e-commerce website and the survey article and compared by using heuristic approaches to reveal the maximum serviceable website (amazon, daraz, Flipkart, Alibaba). The goal paper refers to designing and developing software quality metrics and providing good use to the application related to e-commerce with the best experience.

The survey was completed, and the results were obtained and presented as follows:

RQ1: What are references and research publications?

RQ1.1 What are the most cited scientific research and manuscripts between 2009 and 2023?

Research papers were taken from the Scopus database, and each paper's citations were calculated from 2009 until 2023. The citation percentage is calculated within the database, and it is the entire amount of papers indexed in the database. These documents cite another document or many documents within the repository or refer to researchers, and exclude the researchers’ self-citations.

The following law calculates the rate for normalized citation count (NCC), where 2.62 of the citations were normalized, leaving 12 papers within the average scale. For the remainder, we calculated the NCC for each archived search belonging to the country via the formulation:

for normalized citation count _{per paper} =$\frac{sum\ of\  (N C C)\   for\  all \ papers\  per\  country }{ number\  of \ papers\  for\  country }$      （1）

These papers [1, 19, 43] received the most citations, The results obtained from the previous mathematical equation, when choosing 245 citations, the NCC = 30.63, when choosing 137 citations, the results were NCC = 17.13, and finally with choosing 106, the result becomes NCC = 13.25, we notice country India contributes the most contributes of the quantitative standards, meanwhile, the United States of America appears as a prominent country contributing to product quality standards as in Figure 3. Figure 3 clearly shows the comparison between countries. As for Brazil, Germany, and China, we find that Germany and Brazil have the largest number of research that is characterized by quantitative standards, and China has the quality aspect of the standard.

RQ 1.2 Which are the highest publishing sites?

The top ten papers were identified, by counting the total number of citations for the paper, besides the entire amount of documents, as shown in Figure 4.

in this study, it was found that there are no articles offered in EUROPI2, MetriKon, and MetriSecAnd, according to the indicators of Containment and exception that were adopted in a group of our papers. The papers published through the years 2009-2023 do not contain studies published in EUROSPI2. MetriKon, MetriSec, and SAM. Keyword test "measures the quality of the program product'' within the main title and abstract parts. However, with the availability of many software standards in these conference papers, Through the analysis of the previous data, it was found that there is no close relationship between these criteria on the one hand and the quality metrics of the systems on the other hand.

RQ2: Mind mapping questions related to technical trends.

RQ2.1 What type of technique and methods/criteria are used in the articles?

In 2011, utmost of the research documents in the field of computer science were published by empirical research. Also, in 2010, analytical methods were adopted to publish research papers within the same category, as shown in Figure 5. According to the study by Riccio et al. [43]. Research papers were selected and then classified built on the results of the study process according to the revision by Yan et al. [44]. The results of this mechanism show that 67% of the total 42 research papers used experimental research to develop and adopt them in our study.

3.png

Figure 3. Countries with cited scientific articles

4.png

Figure 4. List of top 10 conferences

5.png

Figure 5. Research facets distribution

RQ2.2 What are the methods observed and used in the research papers?

In this revision, we will employ the Goal Question Metric (GQM), the most popular method [45]. Thus, the classification of 19 papers was isolated because it does not contain detailed research methods, and this was done using (evaluation papers or surveys). Parts of these papers, 6 papers, were marked because they contain a hidden Markov model as a statistical model.

19 of these papers were marked N/A as in Figure 6 and according to the version of GQM. Developmental research methods have been used, which are AHP (Analytical Hierarchical Process), and it is considered a modern approach [46]; AHP was used with another obscure [47], GQM with PSM (Practical Software Measurement) [48]. Moreover, most papers consider GQM to be an inappropriate old method but in SQM, the common and dominant method expression.

6.png

Figure 6. The research method's classification of articles

RQ2.3 What type of SQM was presented in the papers?

First, SQM data is taken and analyzed in detail, and then these data are documented for each of the 70 research papers and articles and organized in measurement details.

RQ2.3.1 What are the common metrics used and levels generally presented in the papers?

The following are the most important fixed criteria for quality metrics:

• Two quality-related applied in method-level measures are McCabe and Halstead, and two quality-related applied in class-level measures are Chidamber and Kemerer (CK).
• QMOOD and MOOD are two of the most popular and widely used measures in papers. Another metric related to the security level of Microsoft's software development life cycle is Michael Howard.
• A significant measure related to the size of the program to clarify errors.
• In languages that use object-oriented programming such as Java, metrics have been proposed at the class level and subclasses, and the relationships and dependencies between these classes are shown by Kitchenham et al. [49].
• A very common metric is coupling and coherence between classes to measure understanding, quality of reusability, and maintainability, and other metrics related to an inheritance that the parent class inherits to descendants, which are prominent features of comprehension, reusability, and flexibility.
• The studies [43, 49] contain metrics for object-oriented software that define quality attributes.

$ Extendibility =0.5(abstraction - coupling + inheritance + polymorphism) $     （2）

$\begin{gathered}Effectiveness =0.2 \times(abstraction + encapsulation + composition + inheritance +  polymorphism )\end{gathered}$      （3）

$Flexibility=0.25 (encapsulation - coupling + composition + polymorphism) $     （4）

$\begin{gathered} Understandability =0.33(-  abstraction +  encapsulation -  coupling +  cohesion -  polymorphism- complexity -  design )\end{gathered}$      （5）

$\begin{aligned}  Functionality =(0.12 \times cohesion )+(polymorphism + messaging + design size +  hierarchies )\end{aligned}$      （6）

One of the simplest measures is Easy Education rate period for education the custom of the component, Interface Density is a design level metric, Index for Interface Packet Usage Transformation (IIPUT), Error Message Clarity, Interface Packet Extension Index (IPEI), and another metric for Compatibility Count Interface benchmarks, Packet Change Effect Index (PCEI), Design Patterns for Four Aggregates, Interface Packet Usage Index (IPUI).

• In most cases, some metrics are used for project management to know the variance of effort, including productivity, COCOMO, and other metrics are schedule variance, schedule level metric.
• Software requirements for development and production use a measure, mostly, some program variable requirements (requirements fluctuation), and the analysis between labor market requirements and production requirements are tracking measures.
• The software measures at the systems level are considered the most common, and it is necessary to measure the software product, namely: defects resulting from the use that the user discovers, the level of user awareness, and the nature of dealing with the system, MTTF (Mean Time to Failure), MTTR (Mean Time to Repair).

RQ2.3.2 In the field of systems application, what are the required metric levels that can be applied?

Seventy research papers were selected, and it was found that 10 of them are subject to the quality of an application using the web, and four types of research related to safety were extracted, which are among the necessary applications, and from the mobile applications are three applications, the embedded applications were extracted two papers, one paper is related with modernity and technology, where artistic tools are used. Among the metrics captured in our paper pool are the number of connections and comments in different domains of web applications within our database, Weighted Methods per Category (WMC), and LCOM [50].

In the applications scope of directory levels and classes, a number of the following measures have been applied:

Physical LOC measures directory-level systems, counts statements in the source code, and spaces in the code, which are blank lines, and counts comments in programs.

LCOM, RFC, and Coupling Between Objects (CBO) Metrics for class-level source code. Message Complexity (MC), DIT, Coupling Weight for a Class (CWC), Class Complexity (CLC), and Average Method Complexity for each Class (AMCC).

• McCabe and Halstead's measures are defined at the method level, CK at the class level. Other commonly used metrics for source code: LOC, Number of Methods (NOM), Cyclomatic Complexity (CC), Number of Fields (NOF), LCOM, number of comment lines, active association, and Association Between Class (ABC) [51].
• Among the metrics of interest are the class-level metrics in the study by Wohlin [52].
• Some object-oriented programs have two metrics in the study by Moser et al. [53] which are DIT and LCOM, and they are usually unsuitable in design and complexity. In the use of CBO [54], it was found that the lower values indicate that the better classes are a few numbers from other classes. Also DIT and NOC measures, their lower values indicate that the use of inheritance for classes and their reuse is not completely reliable in libraries.
• Maintenance measures are among the important measures, as shown in Table 6, and they can be considered a common quality with the measurement characteristic of all kinds of systems. Excluding artificial intelligence applications. Safety measures are also one of the common quality measures for all systems and applications, except embedded applications.

RQ2.4 Which statistical approach are adopted to validate or build new metrics?

A genetic algorithm was used [12, 22] as shown in Figure 7. The swarm algorithm was used as a type of optimization [50]. One research paper relied on historical data and used statistical methods to predict future events [47], one of which is statistical extrapolation. One of the most important methods used in the field of statistics and the most common is descriptive analysis. When comparing statistical methods, it was found that machine learning methods give good predictions and better results when compared with other statistical methods.

Some authors suggested studying a new set of larger data than the previous one but using less exploratory learning methods, and that was in the years 2014 and 2015, where the genetic algorithm and another algorithm, the optimization colonies, were used to obtain more appropriate results [51]. When searching using keywords, it was found that no answer was found. This indicates that neither the genetic algorithm nor the swarm algorithm has been adopted for optimization and has not been mentioned in the past five years; however, to obtain papers that use statistical algorithms and the SQM type [52], by directly writing keywords that contain the name of the approach and specifying the direction used by the algorithm. It was necessary to find a statistical method or algorithm that is suitable for work and able to reach predictions of errors in the early stages of the project or system, as mentioned by Moser et al. [53], which matches with quality attributes. 

7.png

Figure 7. Statistical methods and the number of papers affiliated with it

Table 6. Different domains of research references

8.png

Figure 8. Standards and distribution of relevant papers

9.png

Figure 9. Distribution of quality features for the SQM standard

10.png

Figure 10. Qualitative characteristics and their distribution within the QinU sample

11.png

Figure 11. Quality attributes and their distribution in DQ sample

RQ2.5 What are the patterns and metrics approved in articles that are used to measure the quality of systems?

It was found that a large ratio of the papers does not refer to metrics, which is 74% of the papers, as well as models that do not contain measures of product quality. When looking at the papers, on the other hand, it was found that a more common model was used, which is CMMI. ITIL (Information Technique Infrastructure Library) is referenced in 4 papers as Figure 8, ISO/IEC 1458. By analyzing the feature requirements and integrating them with other traditional quality models, the study by Wahyuningrum and Mustofa [54] was established, which represents the Component Quality features (CQF) model. The resulting model contains six characteristics related to quality features and 23 sub-features related to quality as well, including reliability, maintainability, portability, functionality, and efficiency.

RQ2.6 Determine the software quality measure based on the quality metrics used in the papers.

ISO 9126 [48] is the most popular and used model, and in 2011 the ISO 25010 metrics were used, but the ISO 9126 model is the most present, the most quality, and the most widely used. Where two of the papers noted the use of the Mc Call and FURPS model. As for the others, the Design Quality Model (DQM) was used, and object-oriented programs also used the Quality Design Model (QMOOD).

The previously asked questions (RQ2.7-RQ2.9) aim to determine the quality attributes and their contribution to the quality models (SPQ, and QinU Quality Model).

RQ2.7 Can SPQ's quality metrics and attributes be used in ISO 25010?

The SQM methodology has the characteristics of quality of maintenance and is associated with reliability, as in Figure 9. We also noticed when reviewing the features in the papers published in 2015 that the same features are used mostly, and this was shown when we compared the analysis of our time with the results of the article [17]. Search and inference can be made by searching for keywords and through the Internet repository according to each quality attribute [22].

RQ2.8 What are the QinU quality standards in the ISO 25010 standard quality models that are used in the majority of papers?

From Figure 10, it can be seen that the maximum number of research and metrics in QinU are risk safety and user satisfaction.

RQ2.9 How can the use of DQ quality attributes in ISO 25012 be consistent with the quality model applicable to most papers? The information mentioned in Figure 11 shows that it does not contain papers associated with the use of quality standards such as reliability, effectiveness, validity, accessibility, compliance, recovery, and portability. The most widely used metrics are understandability and accuracy, used to assess data quality. Several sub-features will be addressed and are considered among the promised quality metrics:

(1) Coherence and coupling are prominent measures of maintainability quality attributes, and mostly comprehensibility and reusability quality measures.

(2) The measure of inheritance is often used in terms of complexity and is adopted as a measure of the attributes of flexibility, reusability, understanding, and error tolerance.

(3) An innovative metric for predicting change during software implementation to reduce testing cost, Harrison et al. [37] offers WMC, LOC, and RFC metric combinations.

(4) A data-intensive software maintainability check metric provided by [50] shows a strong relationship between NODBC and CCR metrics.

(5) Some quality metrics are not suitable for measuring quality because the coupling factor is not suitable, and MOOD confirms this [55].

(6) In the study by Otero et al. [56], the measure of maintainability of programs and systems was used by focusing on varying weights for each attribute of quality in terms of the source code's ability to analyze, change, testability, and stability.

(7) In Figure 7, four new measurements were presented within the study for this paper.

RQ2.10 At what phase of the software life cycle (SDLC) is the majority of papers adopted?

In Figure 12, most of the models of SQM are shown. These models are built on several stages, including the code writing stage, the system design stage, and then the model building stage, the ability to predict errors and process these errors. Processing is done proactively, not retroactively.

In Chart 12, more of the SQM models concentrate on the code and design phase of the program life cycle that adopts errors and tries to overcome them proactively.

RQ2.11 Has SQM been subject to any type of program, and what are the most important types of programming adopted by the articles?  

One of the stages of SDLC is the code, and 43 of the 70 research papers analyzed were identified at the code level. In object-oriented programming, 35 papers were revealed out of 43 research papers that were used to increase the class of object-oriented programming (OOP) metrics, as shown in Figure 13. Another type of object-oriented programming (OOP) is aspect-oriented programming (AOO). Three articles related to aspect-oriented programming metrics were discovered, as well as three related research papers on basic programming and one article on code-phase scripting. And the rest of the research papers, five of the articles were distinguished as related to other programs that depend in their work on Compound-Based Programming or independent programs.

RQ2.12 Can SQMs be used with SDLC models?

Only two articles mentioned the correlation between metrics and the Agile sample [8, 55]. In the SDLC stage, the measurement was not used as a basic parameter and within steps in the majority of papers. Other articles used advanced measures instead of the previous ones, and they were re-analyzed and evaluated relative to the activities of the agile methods. Moreover, in agile methods, some studies consider dealing with quality. When we use the development measures in [8, 43], they must be evaluated for validation and re-estimated using the 3C model (continued integration, continued measurement, and continued improvement) of the actions of the agile prototypical.

RQ 2.13 Is it possible to specify a metric threshold limit in the paper warehouse or not?

As it is known, the deviation of the values from the purpose of product quality is of great importance in adapting the threshold in the metrics data. Therefore, the minimum amount of metric data must be specified for consideration. If the system has exceeded the metrics data, the application will respond by sending a caution to execute an emergency plan. In this study, the focus was on researching data related to the threshold value of the data.

The study confirms that 22 articles from 70 research papers evaluated the threshold within their studies. Threshold values are mostly suggested values based on experience and other criteria [57]. For the threshold values to be chosen, they should be more down or more heightened values relative to the design of the source code. To define general thresholds that can be used for various types of measurements, the RTT tool (Relative Threshold Tool) [55] was executed. The Qualities Corpus data set was used to validate the tool, which is an open-source systems tool, and for practical studies [58]. The RT tool [8] calculation method is used to define the relative threshold weight for the programs used according to the code structure. When the penalty rate exceeds the given rate, the system proposes a starting point for a new threshold value that contributes to improving the design.

12.png

Figure 12. Distribution of SDLC stages in articles

13.png

Figure 13. Number of articles for programming kind

14.png

Figure 14. Plans offered by the papers

RQ2.14 What information is available about product quality measures, what are the measurement management tools, and are there any details regarding product quality measures for programs?

Despite the availability of abundant measurement tools, there is a lack of metrics management tools, as well as a loss of integrated quality metrics. Most articles gather and analyze data to estimate different quality features, and the analysis is carried out on several available scales of various kinds, as listed [58, 59].

RQ3: What are the future opinions and current trends of the current articles?

Our current study is employed for the next step and provides a proposal for improving the technique (39%), and we seek to increase the detailed scientific research (26%) by increasing the study of new existing articles in Scheme 14 as shown in Figure 14. In this regard, about 39% of the articles and research can improve standards and techniques through the application of systems, and Prefigure No. 1 lists the search engines that include the databases, and thus the levels, trends, and constraints are classified into three groups.

4.1 Measurement/Models

Most of the research papers studied refer to low values as well as models with low levels, such as ISO 2010 and ISO 916. To raise the quality measures and enhance the usability of each product quality measures through the preparation of additional research articles to address the low values of the measures in the different fields of research. From a practical point of view, the new researchers should work on implementing the quality features by integrating several aspects of the quality models [60].

It can also be noted that there is an essential contrast between metric and value prototype criteria. This opposes the use of familiar criteria in research papers and articles. We have seen the variance in prototypes as one of the significant obstacles to the continued advancement of the standards. Despite this, the abundance of papers and articles presented in conferences and workshops for discussion may be good for noting the contradiction between standards on the one hand and academic articles on the other hand.

Wahyuningrum and Mustofa [54] proposes new quality measures with marketing characteristics as well as sub-characteristics, which are related to cost, time to market, development time, target users, and target affordability. In the article [61], the authors proposed new criteria for quality attributes with a developmental feature depending on the parameters of consumer culture, user perspective, and behavior. After all that has been found, the new quality attributes and characteristics can be approved and included in ISO 25010.

Machine learning has a history in this field, and artificial intelligence can be used through designing programs and subsequent development processes and by using quality models for programmers and developers. After including the modern features resulting from the use of the smart model and related to quality, it will be formed in the scheme automatically. Researchers and developers can be guided by the new model and help them with the activities, and by using the resulting model, we have reduced the error rate for the testers [59].

4.2 Metrics/Techniques

There is a pure application for each field of research that has a specific application context, and therefore there is a need to choose the best set of criteria and standards. Accordingly, different techniques combine instrument-based metrics and predictive functions [62].

By looking at COSMIC (Common Software Measurement International Consortium), more research has been highlighted to identify the drawbacks of size metrics used within different software application domains and by applying other, more quality metrics that belong to more evolutionary applications.

Where more diverse and complex measures have been adopted and depend on solid information such as dependence on program cost, time consumption, effort, and data diversity.

With the existence of new technologies and applications of new technology trends such as artificial intelligence, therefore, we need new product quality measures that as the Internet of Things (IoT), machine learning, artificial intelligence (AI), and Robert [58].

Security measure has much literature, and we can say that the application or system must be fully secured by analyzing the criteria and standards for any type of security attack for one or more. Accordingly, it is necessary to build more quality and reliable security measures to enable us to identify several aspects of different security and to enable us to monitor unintended threats [63]. There is a technology that contains security weaknesses, so it is necessary to provide a product quality assurance measure that increases safety as much as it relates [10].

In the papers, CMM v3.0 [4] and PMBK [64] emphasize the use of agile metrics and processes. These two papers only track metrics connected to Agile. Our study confirms the adoption of developmental processes with the contribution of DevOps with measures that contribute to increasing the speed of development and launching the product to customers. It is necessary to increase the quality of measures for agile projects. Here is what is related to metric thresholds, and because of a lack of relative knowledge about them, the program becomes negatively affected by the performance of measurements [60]. It was necessary to produce new metric threshold values, and unfortunately, our paper repository has little about metric thresholds, new metric thresholds have to fit modern application domains with different programming languages and varying project sizes and SDLC methods.

The paper [27] used standards and measures that help organizations calculate performance and presented EFQM standards, which often systematically compare curricula with other relevant organizations. In the same context, operating software companies should encourage their customers to use standards that agree with similar companies, thus enabling us to improve their operations.

4.3 Tools

One of the very difficult things is calculating measures that have been collected to calculate another measure, and at the same time, it takes a long time. There are several types of measuring metrics to evaluate the metrics for measuring product quality. The question here is whether it is possible to use one general tool to evaluate metrics and be multi-use for various applications instead of using many and multiple tools, and that would add credibility and reliability to the metric data and increase the possibility of its analysis and completeness. The answer is that there is a possibility of integrated general measures that calculate automatically and at different levels and within relative metric thresholds that support machine and statistical learning [64]. In this context, the metric thresholds need time to calculate, as the time can be reduced more effectively by analyzing and identifying the root cause and giving appropriate solutions to values that exceed the metric threshold values. Researchers and laboratory workers can access the scales to calibrate the systems at any time [65].

4.4 Discussion

After studying the papers and articles related to SQM, the results are presented for the period from 2009 to 2023, and the results are represented in the form of detailed maps. A large picture was formed, which allows us to quickly understand the process of diagnosing gaps. Using this technique, greatly helped us in the analysis of the current study and allowed us to improve the issues associated with SQM. The graphical structure is drawn in Figure 15, and looking at this figure will give researchers an idea of the information they can benefit from about TQM trends in the literature. The standards and trends for each section are represented by a percentage. The percentage output was computed with the subsequent equation:

$\begin{aligned}

&\text { Dispersal of courses per RQs (x\%) = }\left(\frac{ sum\  of\  category\ (x)}{sum\  of\  category\ (x, y, z)}\right) \times 100

\end{aligned}$      （7）

Taking maintainability characteristics as an example.

RQ 2.7= [Quantity of maintainability metrics associated to manuscript (46)]/[Calculation of other groups (1+4+26+5+12+9+46+11+)] $\times$ 100,(46 / 114) $\times $100= 41.45%.

Questions, considerations, and issues related to SQM trends over the past ten years are examined and shown in Figure 15.

When we match the outcomes of our study with the results presented by Wahyuningrum and Mustofa [54], we notice that the results match the quality models. The ISO 9126 standard is still general compared to the ISO 25010 standard, among 2006 and 2019. From this, it is clear that published papers and articles did not shed light on the use of the new standard during this period, as shown in Figure 15. It is also obvious, that the ISO 25010 value module issued in 2011 is static in use, while ISO 9126 is still more common. For the novel description of the feature module to be used increasingly, the guidelines can be adjusted to be very clear and easy to diagnose some problems.

It has been shown that the most commonly used quality standards are reliability, safety, and reusability of the system. The most common criteria for measuring QinU are effective system performance, customer satisfaction, and freedom from risk. The last standard is DQ. The most common quality standards are accuracy, ease of understanding, and continuous integration. It was also noted that compatibility, reliability, and robustness were the most frequently used in studies after examining articles in 2007 and 2014. The reliability standard is still a very important factor for measuring product quality to meet customer satisfaction. There remain criteria that are necessary to measure the data quality of systems, which are: accessibility, efficiency, portability, availability, and confidentiality of compliance. Through our study, we noticed that there are no attributes within the standards related to quality models to measure the culture, attributes, attitudes, and points of view of developers, managers, and process models.

In our study, while analyzing quality standards, it was found that the maximum popular ordinary is IEEE 1061:1998, and the module adopted is CMMI. A large number of countries have received certificates from CMMI, and our study has made a significant contribution to SQM. The effective contribution to the use of these metrics, standards, and modules will encourage researchers to pay to the literature related to comprehensive quality product management. The textbook proposed a set of metrics: “Software Engineering, Modern Curriculum” [3] and CMMI [4], and these metrics: Mass metrics to measure the mass of the software code, effort, production cost, variation between software producers, the number of defects according to their harshness, extracting defects and calculating The percentage of defects per hour during program execution, the reuse rate of the application, customer and developer satisfaction requirements, average failure time, maintainability, and testing rate.

15.png

Figure 15. Metrics chart of software quality