A method for evaluating component quality of application software

Abstract

The present invention relates to a software evaluation method, and in particular to a component quality evaluation method for application software. The method organically combines the evaluation of experts and the attribute correlation obtained from defect data, comprehensively considers the weight of component attributes, and then obtains the component quality. The method comprises the following steps: (1) according to the evaluation of the importance of component attributes by experts, the subjective weight of each attribute is obtained by using a fuzzy hierarchical analysis method; (2) according to each attribute value and defect data obtained by the test, the defect data is analyzed by using a grey correlation method to construct a reference quality, and the attribute importance weight is obtained from an objective perspective based on the comparative analysis between the obtained attribute value and the reference quality; (3) using a control parameter, the ratio of subjective weight to objective weight is adjusted, and finally the comprehensive weight of each attribute is obtained. (4) according to the attribute value and comprehensive weight of each attribute, the quality of the component is calculated.

Description

Technical Field

[0001] The invention relates to a software evaluation method, in particular to a component quality evaluation method for application software. Background Art

[0002] As the scale of software grows, people have higher and higher requirements for the rapid delivery and quality of software, and component-based software development methods are increasingly widely used. Components are reusable software modules for software architecture and reusable software components. Developers can develop new application systems by assembling existing components, thereby achieving the purpose of software reuse. Generally, software components should have important properties such as availability, portability, and adaptability. Since software contains multiple components, the quality of components directly determines the quality of the software system. Failure or failure of a single component will bring great disasters to the software system. Therefore, how to evaluate the quality of components so that they can adapt to more software systems is an important issue that needs to be studied. A lot of research has been done on the quality of software components at home and abroad. Based on software quality standards, Jin Maozhong et al. established a software component quality measurement model from the perspective of external and internal quality models, using external ease of assembly as a metric. Based on the relevant systems of the Ministry of Information Industry and the requirements for software quality measurement standards, his team proposed six sub-characteristics that have a direct impact on the reusability of software components, as well as software component reusability quality measurement and component credibility quality measurement schemes. Liu Shuai and others used the fuzzy matter-element evaluation method to comprehensively evaluate component quality. He Haibo and others improved the hierarchical analysis method, and on the premise of ensuring that a single judgment matrix meets the consistency, they calculated the weight of component quality indicators, and evaluated the indicators with different weights assigned to each expert, linearly weighted the weights of quality indicators and expert weights, and when the expert weights were relatively determined, a group decision-making-based strategy was used to comprehensively evaluate component quality. However, the evaluation of experts is fuzzy and arbitrary, which will greatly and directly affect the evaluation of component quality. At the same time, the components themselves have different attributes, and there is a certain correlation between the attributes. By testing the components and obtaining defect data, and analyzing these defect data, the relationship between the defect data and these quality attributes can be established, and then the importance of the attributes can be determined from an objective perspective. Simply using expert evaluation to obtain the weight of the indicator is a relatively simple method. However, it is a technical problem to organically combine the subjective evaluation of experts with the objective information obtained from the test data, and appropriately allocate the ratio of subjective and objective weights to minimize their impact on quality evaluation. Summary of the invention

[0003] The purpose of the present invention is to provide a component quality assessment method for application software, which organically combines the evaluation of experts and the attribute correlation obtained from defect data, comprehensively considers the weight of component attributes, and then obtains the quality of the component.

[0004] To achieve the above object, the present invention provides the following technical solution: a component quality assessment method for application software, comprising the following steps:

[0005] (1) According to the user's demand for component quality, the quality attributes of the component are determined. Experts evaluate the importance of each attribute of the component, obtain the evaluation results of the software component attributes, and use the fuzzy hierarchical analysis method to obtain the importance weight of each attribute of the software component from a subjective perspective;

[0006] (2) Obtain the attribute values, defect data, and number of code lines of each attribute through testing, analyze the defect data using the grey correlation method, construct the reference quality, and obtain the attribute importance weight from an objective perspective based on the comparative analysis between the obtained attribute values ​​and the reference quality;

[0007] (3) Using control parameters, adjust the ratio of subjective weight to objective weight, and finally obtain the comprehensive weight of each attribute;

[0008] (4) The mass of the component is calculated based on the attribute values ​​and comprehensive weights of each attribute.

[0009] Preferably, corresponding to the above quality assessment method, the specific steps are as follows:

[0010] The steps of quantitatively evaluating the importance of software component attributes according to the fuzzy hierarchical analysis method and obtaining the importance weights of software component attributes specifically include:

[0011] Step 1: Experts make subjective judgments on the relative importance of each pair of attributes, and give a judgment table of the relative importance of attributes. According to the corresponding table between importance and triangular fuzzy numbers, a fuzzy judgment matrix is ​​constructed. Assuming there are p experts and n component attributes, the fuzzy judgment matrix of the e-th (1≤e≤p) expert is: where h ij is the relative importance of indicator i and indicator j, in h ij The value of is taken using the integer scaling method of [1,9];

[0012] Step 2: Directly use the arithmetic average method to obtain the fusion matrix

[0013]

[0014] Steps: Using the Fusion Matrix Calculate the fuzzy subjective weights w1,…,w n :

[0015]

[0016] Step 4: Defuzzify the fuzzy weights to obtain the final subjective weights

[0017]

[0018] As a preferred method, relevant defect data is obtained through testing, the defect data is analyzed, and the correlation between each attribute and the defect data is obtained using the grey correlation method, and then the attribute weight is obtained from an objective perspective. The specific steps are as follows:

[0019] Step 1: Determine the reference quality. Based on the attribute values ​​of each attribute obtained from the test, collect defect data from the defect perspective. Considering the negative correlation between defect data and attributes, perform inverse processing on the defect data and calculate the reference quality.

[0020] x0=e -100num / f (4)

[0021] Among them, f is the control parameter of the component, its value is the number of code lines of the component, and num is the number of defect data;

[0022] Step 2: Calculate the correlation coefficient and the absolute difference between each attribute of the component and the reference mass:

[0023] Δ i =|x i -x0|,i=1,2,…,n (5)

[0024] On this basis, according to the formula

[0025]

[0026]

[0027] The maximum difference between the two levels Δ can be obtained max and the two-level minimum Δ min , and on this basis obtain the correlation coefficient between the attribute and the reference quality,

[0028]

[0029] ζ i is x i The correlation coefficient with x0;

[0030] Step 3: Determine the objective weight of the attribute. The higher the correlation, the higher the correlation between the trust attribute and the component defect, and the more weight allocation to the trust attribute needs to be increased. is the objective weight of the ith credible attribute, then

[0031]

[0032] As a preferred method, the fuzzy analytic hierarchy process and the grey relational method are combined to establish the comprehensive weight of component attributes; the specific steps are as follows:

[0033] Step 1: Using the expert's fuzzy judgment matrix of the component, we can obtain in represents the subjective weight of the i-th attribute;

[0034] Step 2: Based on the attribute values ​​and defect data of the software components obtained from the test data, use formulas (4)-(9) to calculate in represents the objective weight of the i-th attribute;

[0035] Step 3: In order to adjust the proportion of subjective and objective weights, it is necessary to select a suitable parameter under which the weight variability is minimized. Let u i Represents the control parameter of the i-th attribute; the final attribute weight and control parameter u are determined by the following constraints and objective function i , any attribute i, its composite weight w i for:

[0036]

[0037] satisfy:

[0038]

[0039] As a preferred method, the importance evaluation of component attributes by experts, the triangular fuzzy number comparison table, the attribute values ​​obtained by the test, the defect data and the control parameters are input into the quality assessment model, and the quality of the software component is output. The specific steps are as follows:

[0040] Step 1: Obtain the comprehensive weight w of each attribute according to formula (10): i ,i=1,2,…n;

[0041] Step 2: Let T C It represents the quality of a component C of the software system to be evaluated, and its tested attribute values ​​are recorded as y1, y2, …, y n ,but

[0042]

[0043] Compared with the prior art, the beneficial effects of the present invention are as follows: the component quality assessment method of the application software of the present invention not only combines the subjective evaluation of experts, but also combines the relationship between defect data and various attributes from an objective perspective, conducts a comprehensive analysis of the weights of various attributes of the component, and determines a reasonable subjective and objective weight distribution ratio, which not only reduces the subjectivity of experts, but also takes into account the influence of each attribute itself on the component quality, and evaluates the quality of the component more comprehensively. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0045] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] In the specific implementation, the fuzzy hierarchical analysis method is used to obtain the importance weight of each attribute of the software component from the subjective aspect;

[0047] The grey correlation method is used to analyze the contrast intensity and conflict between software component attributes and reference quality, and the attribute importance weight is obtained from an objective perspective;

[0048] The fuzzy analytic hierarchy process and grey relational method are combined to establish the comprehensive weight of component attributes.

[0049] Input the experts' evaluation of the importance of component attributes, the triangular fuzzy number comparison table, the attribute values ​​obtained from the test, the defect data and the control parameters into the quality assessment model, and output the quality of the software component.

[0050] (1) The steps of quantitatively evaluating the importance of software component attributes according to the fuzzy hierarchical analysis method and obtaining the importance weight of the software component attributes include:

[0051] Step 1: According to the judgment table of relative importance of attributes given by experts, as shown in Table 1, construct a fuzzy judgment matrix.

[0052] Table 1. Importance ratio of attribute i and attribute j

[0053]

[0054] For the evaluation table given by the experts, a fuzzy judgment matrix with triangular fuzzy numbers as elements is established using a comparison table similar to that in Table 2.

[0055] Table 2. Meaning of judgment matrix elements and their corresponding triangular fuzzy numbers

[0056]

[0057]

[0058] Assuming there are p experts and n evaluation indicators, the fuzzy judgment matrix of the e-th (1≤e≤n) expert is: where h ij is the relative importance of indicator i and indicator j, in h ij The value of is determined using the integer scaling method proposed in [1,9].

[0059] Step 2: Directly use the arithmetic average method to obtain the fusion matrix

[0060]

[0061] Step 3: Using the Fusion Matrix Calculate the fuzzy subjective weights w1,…,w n :

[0062]

[0063] Step 4: Defuzzify the fuzzy weights to obtain the final subjective weights

[0064]

[0065] (2) Obtain relevant defect data through testing, analyze the defect data, and use the grey correlation method to obtain the correlation between each attribute and the defect data, and then obtain the attribute weight from an objective perspective. The specific steps are as follows:

[0066] Step 1: Determine the reference quality. After testing the build, the property values ​​obtained are shown in Table 3.

[0067] Table 3. Component property values

[0068]

[0069] Defect data is collected from the defect perspective. Assume that the collected defect data is num, and the number of code lines of the component itself is f. Considering the negative correlation between defect data and attributes, the defect data is inverted and the reference quality x0 is calculated.

[0070] x0=e -100num / f (4)

[0071] Step 2: Calculate the correlation coefficient. Assuming that the component has n attributes, calculate the absolute difference between each attribute of the component and the reference mass:

[0072] Δ i =|x i -x0|,i=1,2,…,n (5)

[0073] On this basis, according to the formula

[0074]

[0075]

[0076] The maximum difference between the two levels Δ can be obtained max and the two-level minimum Δ min , and on this basis obtain the correlation coefficient between the attribute and the reference quality,

[0077]

[0078] ζ i is x i The correlation coefficient with x0.

[0079] Step 3: Determine the objective weight of the attribute. The higher the correlation, the higher the correlation between the trust attribute and the component defect, and the weight allocation of the trust attribute needs to be increased. is the objective weight of the ith credible attribute, then

[0080]

[0081] (3) Combine the fuzzy analytic hierarchy process and the grey correlation method to establish the comprehensive weight of component attributes. The specific steps are as follows:

[0082] Step 1: Using the expert's fuzzy judgment matrix of the component, we can obtain in represents the subjective weight of the i-th attribute.

[0083] Step 2: Based on the attribute values ​​of the software components obtained from the test, use formulas (4)-(9) to calculate in represents the objective weight of the i-th attribute.

[0084] Step 3: In order to adjust the proportion of subjective and objective weights, it is necessary to select a suitable parameter under which the weight variability is minimized. Let u i To represent the control parameter of the i-th attribute. The final attribute weight and control parameter u are determined by the following constraints and objective function. i, any attribute i, its composite weight w i for:

[0085]

[0086] satisfy:

[0087]

[0088] (4) Experts evaluate the importance of component attributes, the triangular fuzzy number comparison table, the attribute values ​​obtained from the test, the defect data, and the control parameters to the quality assessment model, and output the quality of the software component. The specific steps are as follows:

[0089] Step 1: Obtain the comprehensive weight w of each attribute according to formula (10): i ,i=1,2,…n;

[0090] Step 2: Let T C represents the quality of a component C of the software system to be evaluated, and the attribute values ​​tested are recorded as y1, y2, …, y n ,but

[0091]

[0092] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A component quality assessment method for application software, characterized in that: The steps include: (1) According to the user's demand for component quality, the quality attributes of the component are determined. Experts evaluate the importance of each attribute of the component, obtain the evaluation results of the software component attributes, and use the fuzzy hierarchical analysis method to obtain the importance weight of each attribute of the software component from a subjective perspective; (2) The attribute values, defect data, and number of lines of code of each attribute are obtained through testing. The defect data are analyzed using the grey correlation method to construct the reference quality. Based on the comparative analysis between the obtained attribute values ​​and the reference quality, the attribute importance weight is obtained from an objective perspective. (3) Using control parameters, adjust the ratio of subjective weight to objective weight, and finally obtain the comprehensive weight of each attribute; (4) The mass of the component is calculated based on the attribute values ​​and comprehensive weights of each attribute.

2. The component quality assessment method of application software according to claim 1, characterized in that: Corresponding to the above quality assessment method, the specific steps are as follows: The steps of quantitatively evaluating the importance of software component attributes according to the fuzzy hierarchical analysis method and obtaining the importance weights of software component attributes specifically include: Step 1: Experts make subjective judgments on the relative importance of each pair of attributes, and give a judgment table of the relative importance of attributes. According to the corresponding table between importance and triangular fuzzy numbers, a fuzzy judgment matrix is ​​constructed. Assuming there are p experts and n component attributes, the fuzzy judgment matrix of the e-th (1≤e≤p) expert is: where h ij is the relative importance of indicator i and indicator j, in h ij The value of is taken using the integer scaling method of [1,9]; Step 2: Directly use the arithmetic average method to obtain the fusion matrix Step 3: Using the Fusion Matrix Calculate the fuzzy subjective weights w1,…,w n : Step 4: Defuzzify the fuzzy weight to obtain the final subjective weight w1 s , w2 s ,…,w n s 3. The component quality assessment method of application software according to claim 1, characterized in that: Obtain relevant defect data through testing, analyze the defect data, use the grey correlation method to obtain the correlation between each attribute and defect data, and then obtain the attribute weight from an objective perspective. The specific steps are as follows: Step 1: Determine the reference quality. Based on the attribute values ​​of each attribute obtained from the test, collect defect data from the defect perspective. Considering the negative correlation between defect data and attributes, perform inverse processing on the defect data and calculate the reference quality. x0=e -100num / f (4) Among them, f is the control parameter of the component, its value is the number of code lines of the component, and num is the number of defect data; Step 2: Calculate the correlation coefficient and the absolute difference between each attribute of the component and the reference mass: Δ i =|x i -x0|,i=1,2,…,n (5) On this basis, according to the formula The maximum difference between the two levels Δ can be obtained max and the two-level minimum Δ min , and on this basis, the correlation coefficient between the attribute and the reference quality is obtained. ζ i is x i The correlation coefficient with x0; Step 3: Determine the objective weight of the attribute. The higher the correlation, the higher the correlation between the trust attribute and the component defect, and the more weight allocation to the trust attribute needs to be increased. is the objective weight of the ith credible attribute, then 。 4. The component quality assessment method of application software according to claim 1, characterized in that: The fuzzy analytic hierarchy process and the grey relational method are combined to establish the comprehensive weight of component attributes; the specific steps are as follows: Step 1: Using the expert's fuzzy judgment matrix of the component, we can obtain in represents the subjective weight of the i-th attribute; Step 2: Based on the attribute values ​​and defect data of the software components obtained from the test data, use formulas (4)-(9) to calculate in represents the objective weight of the i-th attribute; Step 3: In order to adjust the proportion of subjective and objective weights, it is necessary to select a suitable parameter under which the weight variability is minimized. Let u i Represents the control parameter of the i-th attribute; the final attribute weight and control parameter u are determined by the following constraints and objective function i , any attribute i, its composite weight w i for: satisfy: 。 5. The component quality assessment method of application software according to claim 1, characterized in that: Input the importance evaluation of component attributes by experts, the triangular fuzzy number comparison table, the attribute values ​​obtained by the test, the defect data and the control parameters into the quality assessment model, and output the quality of the software component. The specific steps are as follows: Step 1: Obtain the comprehensive weight w of each attribute according to formula (10): i ,i=1,2,…n; Step 2: Let T C represents the quality of a component C of the software system to be evaluated, and its tested attribute values ​​are recorded as y1, y2, …, y n ,but 。

Images