JP2003029970A - Program quality management supporting device, program quality management supporting method, computer readable recording medium having program for making computer perform program quality management supporting method recorded thereon and program for making computer perform program quality management supporting method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize simple and stable program quality evaluation hardly affected by the fluctuation of a metrics value depending on the characteristics of a program source code without depending on the experience or intuition of a program developer. SOLUTION: A program specifying means 10 specifies the program (the program of the latest version, for instance) of a quality estimation object and the program (the program of a version one before the latest version, for instance) of a comparison object, and a metrics measuring means 4 measures the metrics value to a prescribed item for the program of the quality estimation object and the program of the comparison object. A quality estimating means 5 decides a difference between the metrics value of the program of the quality estimation object and the metrics value of the program of the comparison object and calculates the quality estimation value of the program of the quality estimation object on the basis of the difference and a work guide presenting means 6 presents the calculated quality estimation value to the program developer.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for supporting quality improvement work of a program in software development, and in particular, analyzing the source code of the program and measuring metrics to diagnose its quality and characteristics. Program quality control support technology that supports the proper implementation of quality control improvement work.

[0002]

2. Description of the Related Art With the rapid development of information processing technology and the expansion of application fields, the scale and function of software to be developed are generally rapidly expanding. In response to this situation, various technologies are currently being developed for improving software development efficiency and quality.

Generally, software development is carried out by procedures such as program design, production, and test (verification). However, it is necessary to prevent errors from being created in the process, and to find and correct errors as soon as possible. It is well known that it reduces backtracking work in development, which leads to improvement in development efficiency and quality. As one of the methods for managing and improving the quality of programs in software development, metrics (me
trics ... metric, sometimes referred to as metric), and the value is used to evaluate quality. Here, the metrics for a program can be calculated from the output of the program design or production work, and can be used as a scale indicating the scale and complexity of the program, the ease of understanding, the ease of maintenance, etc. Is the numerical value of.

For example, the number of steps of a program source code, which is often used as a numerical value indicating the scale of a program, is a value calculated from the program source code which is one of the outputs of the program development work, and can be calculated from the source code of the program. This is one of the metrics. Similarly, the comment rate of the source code, the number of conditional branches, etc. can also be called metrics. In this way, as a metric of a program, a value that can be calculated from a source code is generally widely used. For example, the number of pages of a program specification or the number of function points obtained from the program specification is also a metric of a program. Various metrics have been proposed as metrics and are used to evaluate the quality of programs.

However, when quality control of a program in software development is generally performed by quality evaluation using metrics, there are the following problems. That is, first, there is a problem that the program cannot judge that the quality is the same even if the values of the metrics are the same. This is a fundamental problem based on the essential characteristic of programs, that is, programs that realize the same function can be realized by various methods.
This is because the value of the metric of the developed program easily changes depending on the required performance, operating conditions, or the technical level of the developer. Therefore, as a quality evaluation method based on the value of metrics, qualitative knowledge is widely known, but there is a problem that the quantitative evaluation method is not clear. In particular, when quality evaluation is performed from a plurality of metrics, which metric is selected as a measurement item and how the measured value is evaluated tend to depend on the experience and intuition of the evaluator. For this reason, the quality evaluation work using the metrics has a problem that it is difficult to automate the operation, and it is liable to be a complicated work manually, and the basis for the quality evaluation result tends to be ambiguous. Further, when quality improvement work is performed based on the result of quality evaluation by metrics, there is a problem that it is difficult to obtain a standard of how much effort and how much work should be performed for the work.

There are already some technologies for solving these problems. As a first conventional example relating to a program quality management support technique using such program metrics, Japanese Patent Laid-Open No. 7-16
The method indicated by the 0495 "source code evaluation device" will be described with reference to FIG.

FIG. 9 is a block diagram showing a source code evaluation device of the conventional example. The source code evaluation device is a terminal 10
1 and a host 102. The host 102 includes an information collecting unit 103 and a program source code file 10.
4. Quality metric value measuring unit 105, system quality evaluation unit 106, system evaluation data file 107, dangerous program extraction unit 108, program evaluation data file 109, evaluation result file 110, information feedback unit 111, system evaluation data file Change part 1
12, program evaluation data file changing unit 113,
It is composed of a trouble information file 114.

In the conventional example, the information collecting unit 103 in the host 102 automatically collects the source code of the program forming the evaluation target system from each terminal and stores it in the program source code file 104. The source code stored in the program source code file 104 is read by the quality metric value measuring unit 105, and its quality metric value is automatically measured. The system quality evaluation unit 106 automatically evaluates the quality of the evaluation target system using the measured quality metric value of each source code based on the system evaluation data stored in the system evaluation data file 107. . It should be noted that the system quality evaluation unit 106 obtains a metric value that is significantly different between the good system model and the bad system model as the data for system evaluation, and calculates each metric value from the metric value of the source code corresponding to the evaluation target system. For each, it is determined whether it is close to a good system model or a bad system model, and the evaluation results of the system are obtained by summing up the judgment results from the viewpoint of evaluation of the metrics.

The dangerous program extraction unit 108 automatically uses the measured quality metric values of each source code based on the program evaluation data stored in the program evaluation data file 109 to automatically determine the degree of danger of the program. The dangerous program is extracted and stored in the evaluation result file 110. At this time, the dangerous program extraction unit 108 obtains, as the program evaluation data, information such as selection of metrics used for program evaluation, risk or reference value for metrics, and weighting from viewpoint of evaluation, and for each program, from the metrics value. The risk level of each program is calculated by calculating the risk level of each program and further weighting the viewpoints so that the risk levels for each metric are summarized.

The information of the quality evaluation of the evaluation target system and the degree of risk of the program stored in the evaluation result file 110 is presented to each terminal by the information feedback unit 111.

The system evaluation data file 107
The system evaluation data held by can be updated by the system evaluation data file changing unit 112. The system evaluation data file changing unit 112 uses the quality metric value measuring unit 105 from the corresponding program source code file 104 based on the good system model and the bad system model instructed by the user, for each system. An average value of the metric values is calculated, and only the metric values, which are statistically significantly different between the two, are used as the system evaluation data.
Set to 7.

The program evaluation data file 10
The program evaluation data held by 9 can be updated by the program evaluation data file changing unit 113. By using the program evaluation data file changing unit 113, the user handles the trouble information collected by the information collecting unit 103 from each terminal and stored in the trouble information file 114 with the degree of risk of each program held in the evaluation result file 110. While confirming, the current settings are examined for the program evaluation data held in the program evaluation data file 109, such as selection of metrics used for program evaluation, risk or reference value for metrics, and weighting from the viewpoint of evaluation. ,
If necessary, the setting can be changed and the contents of the program evaluation data file 109 can be updated.

As a second conventional example relating to a program quality control support technique using program metrics, the method disclosed in Japanese Patent Laid-Open No. 9-212387 "Program Development Support Device" will be described with reference to FIG.

FIG. 10 is a block diagram showing the conventional program development support apparatus. The program development support apparatus includes a program storage means 121, a quality measurement means 122, a quality storage means 123, a debug order determination means 124, and a debug means. 125, determination information storage means 128, and correction number detection means 129. Debugging means 125 includes program display means 126 and program correction means 127.

In this conventional example, the quality measuring means 122 measures metrics for each source code of the program stored in the program storing means 121, stores the value in the quality storing means 123, and stores it in the judgment information storing means 128. Output. The debug order determination means 124
Based on the judgment information obtained from the judgment information storage unit 128, the metric value for each source code held by the quality storage unit 123 is used to obtain the rank value for determining the debugging order for each source code. In the debug means 125, according to the rank value obtained by the debug order determination means 124, the program display means 126 sequentially displays the source code in ascending order of the rank value, and the user corrects the corresponding source code by the program correction means 127. By doing so, debug work (defect repair work) is performed. When the source code debugging work is completed, the modification number detection means 129 detects the modification number of the corresponding source code and outputs it to the determination information storage means 128. The judgment information storage means 128 associates the correction number of the source code passed from the correction number detection means 129 with the uncorrected metric value of the source code passed from the quality measuring means 122, and records and stores the judgment information. To do.

The program development support apparatus according to the conventional example performs the above-mentioned operation for each debugging operation of the program.
As a result, the judgment information storage unit 128 records and stores information (judgment information) on a large number of source codes as a pair of the uncorrected metric value and the correction number. Using this determination information, the debug order determination means 124 determines the rank value for determining the debug order as follows. In the debug order determination means 124, the metrics value of the target source code is obtained from the quality storage means 123, and from the determination information stored in the determination information storage means 128,
Search the judgment information that the metrics value is similar,
Obtain the corresponding correction number (estimated correction number). Then, the respective source codes are arranged in descending order of the estimated number of modifications, and the order is used as the rank value. That is, the rank value of the program source code having the largest estimated correction number is set to 1, and the rank value is increased in the order of decreasing estimated correction number.

[0017]

As described above, in the first conventional example, system evaluation and program evaluation (calculation of risk) can be automatically performed from the value of metrics. Regarding the criteria for evaluating the system, by comparing the metrics values of the good system model and the bad system model,
We have automated the selection of metrics used for system evaluation and the setting of parameters used for quantitative judgment of metrics, and by doing so, we will attempt to improve existing problems by implementing quantitative evaluation and automation of evaluation work. There is. However, the good system model and the bad system model itself for obtaining the system evaluation standard must be selected by the user, and the basis is vague, and in particular, when there is no system model similar to the evaluation target system, the correct judgment is made. There is a problem that it is difficult to select a model that can be used. Furthermore, in the evaluation of individual programs (calculation of risk levels), the selection of metrics and their criteria must be individually determined and set by the user, and the setting work is complicated, and the contents are the experience of the user. I have been unable to solve the existing problems in that I am left with the intuition.

In the second conventional example, the quality of the program is automatically based on the metric value of the program based on the past results of the correction work stored and accumulated in the determination information storage means 128 in the order of debugging. Since it operates so as to make an evaluation, the metrics values are automatically evaluated based on the results of past correction work, and the ambiguity of the basis of the judgment is eliminated. In addition, by estimating the number of modifications, it is possible to present a guideline for the amount of work for debugging the program. However, in this conventional example, since there is no clear suggestion on the method of selecting the metric to be used, the existing problem cannot be solved in that the metric selection is still left to the experience and intuition of the user. Further, in this conventional example, it is premised that a high correlation is obtained between the metric value and the number of corrections. Considering that the metric value of the developed source code easily fluctuates,
On the contrary, it can be easily inferred that the amount of defects in the program source code having similar metrics values, that is, the required number of modifications, varies considerably. Therefore, in this conventional example, the rank value based on the estimated number of corrections is not a correct guideline for quality unless the user selects and uses a “smart” metric that provides a high correlation between the metric value and the number of corrections. On the contrary, there is a problem that erroneous quality evaluation may be performed and work efficiency may decrease.

The present invention has been made in order to solve the above-mentioned problems of the prior art, is not easily affected by the fluctuation of the metric value depending on the characteristics of the program source code, and the experience of the program developer and By realizing a simple and stable program quality evaluation that does not depend on intuition and presenting appropriate suggestions for the quality improvement work performed by the program developer, the software quality evaluation can be performed easily and quickly and stably. The objective is to obtain a program quality management support device, etc. that improves the efficiency of software quality improvement measures and improves the efficiency and quality of program development.

[0020]

A program quality management support apparatus according to the present invention is a program quality management support apparatus that estimates the quality of a program in program development and supports quality management of the program, and is the first version in program development. And a program of at least one or more revisions are associated with each other to form a revision program group, and program management means for managing a plurality of versions of the program by the program revision for each revision program group and the program management means Designate any revised version program of any revised program group as a quality estimation target program, and compare programs of the same revision program group as the quality estimation target program and versions earlier than the quality estimation target program The program specified as a program With respect to the gram specifying means, the quality estimation target program and the comparison target program, a metric value measuring means for measuring a metric value for each of at least one or more kinds of metric measuring items, and a metric value measured by the metric value measuring means. Based on the measurement result, the quality estimation unit estimates the quality of the quality estimation target program, and a quality estimation unit that calculates the quality estimation value of the quality estimation target program, and the quality estimation value of the quality estimation target program calculated by the quality estimation unit. And a quality estimation value presenting means for presenting.

The metric value measuring means measures the metric value of the quality estimation target program and the metric value of the comparison target program for each metric measurement item, and the quality estimating means measures the metric value measuring means. Based on the measurement result of the metric value according to, determines the difference between the metric value of the quality estimation target program and the metric value of the comparison target program for each metric measurement item
The quality estimation value of the quality estimation target program is calculated based on the determined differences.

The program quality management support apparatus further includes a difference extracting means for extracting a difference in program content between the quality estimation target program and the comparison target program, and the metric value measuring means is provided for the difference. With respect to the difference extracted by the extraction means, a metric value is measured for each of the metric measurement items, and the quality estimation means estimates the quality of the quality estimation target program based on the measurement result by the metric value measurement means. It is characterized by calculating a value.

A program quality management support apparatus according to the present invention is a program quality management support apparatus for estimating the quality of a program in program development and supporting the quality management of the program, and at least one of the first version of the program in the program development. A program management unit that manages a plurality of program versions by program revision for each revised program group by associating the revised program programs with each other, and for each revised program group managed by the program management unit. With respect to each of a plurality of versions of programs, a metric value measuring means for measuring a metric value for each of at least one or more types of metric measuring items, and a metric value measurement result by the metric value measuring means are provided in the revised program group. And a metric value measurement result storage means to be stored in and a revised version program of any revised program group managed by the program management means, which is designated as the quality estimation target program, and is the same as the quality estimation target program. From the metric value measurement result storage means, a program designating unit that belongs to a revised program group and designates a program of a version earlier than the quality estimation target program as a comparison target program,
Acquiring a measurement result of a metric value regarding the quality estimation target program and the comparison target program, estimating the quality of the quality estimation target program based on the obtained measurement result of the metric value, and estimating the quality of the quality estimation target program It is characterized by comprising a quality estimating means for calculating a value and a quality estimated value presenting means for presenting a quality estimated value of the quality estimation target program calculated by the quality estimating means.

The metric value measuring means measures the metric value for each metric measurement item for each of a plurality of editions of the revised program group, and the metric value measurement result storage means is for the revised program group. For each of the plurality of versions of the program, stores the measurement result of the metric value, the quality estimating means, from the metric value measurement result storage means, the measurement result of the metric value for the quality estimation target program and the Get the measurement result of the metric value for the compared program,
The difference between the metric value of the quality estimation target program and the metric value of the comparison target program is determined for each metric measurement item, and the quality estimated value of the quality estimation target program is calculated based on the determined difference in each metric value. It is characterized by doing.

The program quality control support device further extracts, for each program managed by the program management means, a difference in program content between each of the programs of other versions of the same revised program group. The metric value measuring means has an extracting means, and with respect to each of the metric measuring items, the difference from each of the other versions of the same revised program group extracted for each program by the difference extracting means. The metric value is measured, and the metric value measurement result storage means stores the metric value measurement result for each difference for each of a plurality of edition programs for each of the revised program groups, and the quality estimation means From the metric value measurement result storage means, Get the measurement results of the metrics values for the difference between the quality estimation target program and the comparative program, and calculates a quality estimate of the quality estimation target program on the basis of the measurement result of the obtained metrics values.

The quality estimation value presenting means presents a guide for the quality improvement work of the quality estimation target program based on the quality estimation value of the quality estimation target program.

The program designating means designates a plurality of comparison target programs, and the quality estimating means calculates a quality estimation value of the quality estimation target program corresponding to each of the plurality of comparison target programs to calculate the quality. The estimated value presenting means presents a guide for the quality improvement work of the quality estimation target program, based on each quality estimation value of the quality estimation target program.

The program designating means designates, as the quality estimation target program, a latest version of the program of any of the revised programs, and the quality estimation value calculates a quality estimation value of the latest version of the program. It is characterized by doing.

The program specifying means specifies, as the quality estimation target program, a program other than the latest version of the program that belongs to the same revised program group as the latest version of the program, and the quality estimating means ,
Calculating a quality estimation value of the other program, the quality estimation value presenting means compares the quality estimation value of the latest version of the program with the quality estimation value of the other program, and based on the result of the comparison, the latest version. It is characterized by presenting a guide to the quality improvement work of the version of the program.

The quality estimating means, when the latest version of the program is designated by the program designating means as the quality estimation target program, at least one estimated value relating to a defect occurring in the latest version of the program. It is characterized in that the above-described quality estimation value presenting means presents each of the estimation values relating to the defect calculated by the quality estimation value.

The program quality management support device further includes quality estimation value calculation rule storage means for storing at least one quality estimation value calculation rule indicating a quality estimation value calculation procedure by the quality estimation means, The quality estimation means calculates the quality estimation value of the quality estimation target program according to one of the quality estimation value calculation rules stored in the quality estimation value calculation rule storage means.

The quality estimation value calculation rule storage means stores a parameter quality estimation value calculation rule including parameters corresponding to respective metric measurement items, and the quality estimation means stores the parameter quality estimation value calculation rule in relation to the metrics value measurement means. , Instructing each of a plurality of versions of programs of a specific revision program group to measure a metric value for each metric measurement item, and the metric value measuring means, based on an instruction from the quality estimating means, The metrics value for each metric measurement item is measured for each of a plurality of versions of the revised program group, and the quality estimating unit determines each of the parameter quality estimation value calculation rules based on the measurement result by the metric value measuring unit. To the parameters of And determining the numerical values.

The quality estimating means performs a regression analysis using the measurement result by the metric value measuring means, calculates a regression coefficient for each metric measurement item, and calculates the parameter quality estimated value based on each calculated regression coefficient. It is characterized by determining a numerical value for each parameter of the rule.

When the quality estimation means calculates a quality estimation value of one of the programs at a predetermined timing and analyzes the estimation accuracy of the calculated quality estimation value, the estimation accuracy exceeds a predetermined allowable range. In addition, numerical values for respective parameters of the parameter quality estimation value calculation rule are set.

The quality estimation value calculation rule storage means stores a plurality of quality estimation value calculation rules, and the quality estimation means stores at least two or more quality estimation value calculation rules among the plurality of quality estimation value calculation rules. Is selected as a selection quality estimation value calculation rule, a quality estimation value corresponding to each of the selection quality estimation value calculation rules is calculated for one of the programs, and the estimation accuracy is analyzed and analyzed for each calculated quality estimation value. Based on the result, a specific quality estimation value calculation rule is designated from the selected quality estimation value calculation rules as a designated quality estimation value calculation rule, and the quality estimation value of the quality estimation target program is calculated according to the designated quality estimation value calculation rule. It is characterized by doing.

Each of the plurality of quality estimation value calculation rules includes a parameter corresponding to each metric measurement item, and the quality estimation means, prior to the selection of the selected quality estimation value calculation rule, the metric value measurement means. Instructing each of the multiple editions of a particular revision program group to measure the metric value for each metric measurement item,
The metric value measuring means measures a metric value for each metric measurement item for each of a plurality of versions of the program of the specific revised program group based on an instruction from the quality estimating means, and the quality estimating means, Based on the measurement result by the metrics value measuring means, determine the numerical value for each parameter, and calculate the quality estimation value corresponding to each of the selected quality estimation value calculation rule by using the determined numerical value for each parameter. Characterize.

The quality estimating means performs a plurality of types of regression analysis using the measurement results by the metric value measuring means, calculates a plurality of regression coefficients for each metric measurement item, and calculates the accuracy of each calculated regression coefficient. Perform an analysis, select a specific regression coefficient from multiple regression coefficients for each metric measurement item based on the analysis result, and use the selected regression coefficient for each metric measurement item for each parameter. It is characterized by determining a numerical value.

The program management means manages the revision cause information indicating the program revision cause of the revised version program for each revision program group, and the program designating means determines the quality estimation target program based on the revision cause information. A revised program with the same revision cause is designated as the program to be compared, and the quality estimation value calculation rule storage means classifies the quality estimation value calculation rules for each program revision cause and stores a plurality of quality estimation value calculation rules. Then, the quality estimating means calculates the quality estimation value of the quality estimation target program according to a quality estimation value calculation rule corresponding to a program revision cause of the quality estimation target program.

The program managing means manages a plurality of revised program groups, and the quality estimating means classifies programs of a plurality of versions of the plurality of revised program groups into a plurality of groups, respectively, and the quality estimation target program. The quality estimation value corresponding to each of the selection quality estimation value calculation rules is calculated for any of the programs in the same group that are classified into the same group, and the estimation accuracy is analyzed for each of the calculated quality estimation values. Based on the selection quality estimation value calculation rule, the designated quality estimation value calculation rule is designated.

The quality estimating means specifies a specific quality estimation value calculation rule for one of the programs without classifying the programs of the plurality of editions of the plurality of revised program groups into a plurality of groups. In addition to the quality estimation value calculation rule, the plurality of revision programs of the plurality of revised program groups are classified into a plurality of groups, respectively, and any one of the group programs classified into the same group as the quality estimation target program is classified. As a second quality estimation value calculation rule by designating a specific quality estimation value calculation rule, the first quality estimation value calculation rule and the second quality estimation value rule are compared, and the quality estimation value presenting means is Presenting a guide to the quality improvement work of the quality estimation target program based on the comparison result by the quality estimation means It is characterized in.

A program quality management support apparatus according to the present invention is a program quality management support apparatus which estimates the quality of a program in program development and supports the quality management of the program, and a program management means for managing a plurality of programs. , A specific program is designated as a quality estimation target program from the plurality of programs managed by the program management means, and at least two or more programs are compared from other programs other than the quality estimation target program. Program specifying means for specifying as a program, each of the quality estimation target program and the comparison target program,
A metric value measuring unit for measuring a metric value for each of at least one or more metric measuring items, and a quality evaluation coordinate composed of coordinate axes corresponding to each metric measuring item. The coordinate position in the quality evaluation coordinates is specified for each of the quality estimation target program and the comparison target program based on the measurement result, and the geometric operation is performed based on each specified coordinate position. A quality estimation unit that estimates the quality of the quality estimation target program based on the quality estimation unit and a quality estimation unit that calculates the quality estimation value of the quality estimation target program; and a quality that presents the quality estimation value of the quality estimation target program calculated by the quality estimation unit. And an estimated value presenting means.

The quality estimating means classifies each of the coordinate positions of the program to be compared into at least one coordinate position group based on the distribution of the coordinate positions of the program to be compared, and for each coordinate position group. A specific coordinate position in the coordinate position group is determined as a comparison target coordinate position, a geometric calculation is performed based on the coordinate position of the quality estimation target program and each comparison target coordinate position, and the result of the geometric calculation is determined. A quality estimation value of the quality estimation target program is calculated based on the above.

A program quality management support method according to the present invention is a program quality management support method for estimating the quality of a program in program development and supporting the quality management of the program, and at least one of the first version of the program in the program development. A program management step of managing a plurality of versions of programs by program revision for each program revision group by associating with the program of the above revised editions, and any program revision group managed by the program management step A program that specifies any of the revised programs as a quality estimation target program, and that designates a program of a version earlier than the quality estimation target program as a comparison target program that belongs to the same revised program group as the quality estimation target program. A measuring step, a metric value measuring step of measuring a metric value for each of at least one or more metric measuring items in the quality estimation target program and the comparison target program, and measuring the metric value by the metric value measuring step. Based on the result, a quality estimation step of estimating the quality of the quality estimation target program and calculating a quality estimation value of the quality estimation target program, and a quality estimation value of the quality estimation target program calculated by the quality estimation step. And a quality estimation value presenting step of presenting.

In the metric value measuring step, the metric value of the quality estimation target program and the metric value of the comparison target program are measured for each metric measurement item, and the quality estimating step is performed in the metric value measuring step. Based on the measurement result of the metric value according to, the difference between the metric value of the quality estimation target program and the metric value of the comparison target program for each metric measurement item is determined, and the quality estimation target The feature is that a quality estimation value of the program is calculated.

The program quality management support method further includes a difference extraction step of extracting a difference in program content between the quality estimation target program and the comparison target program, and the metric value measuring step comprises:
With respect to the difference extracted by the difference extracting step, a metric value is measured for each of the metric measurement items, and the quality estimating step is based on the measurement result by the metric value measuring step. The feature is that a quality estimation value is calculated.

A program quality management support method according to the present invention is a program quality management support method for estimating the quality of a program in program development and supporting the quality management of the program, and at least one of the first version of the program in the program development. A program management step of managing a plurality of program revisions for each program revision group by associating the revised program with the program revisions described above, and for each program revision group managed by the program management step For each of a plurality of versions of the program, a metric value measuring step for measuring a metric value for each of at least one or more metric measuring items, and a metric value measuring result by the metric value measuring step A metric value measurement result storing step for storing each correction program group and a revised version program of any revised program group managed by the program management step are designated as a quality estimation target program, and the quality estimation target From the program specification step of designating a program of a version earlier than the quality estimation target program belonging to the same revised program group as the program as the comparison target program, and the metric value measurement result storing step, the quality estimation target program and the comparison target Acquiring the measurement result of the metrics value related to the program, estimating the quality of the quality estimation target program based on the measurement result of the acquired metrics value, a quality estimation step of calculating a quality estimation value of the quality estimation target program, The item And having a quality estimate presenting step of presenting a quality estimate of the quality estimation target program calculated by the estimating step.

In the metric value measuring step, the metric value for each metric measurement item is measured for each of a plurality of editions of the revised program group, and the metric value measurement result storing step is performed in the revised program group. For each, storing the measurement results of the metrics value for each of the plurality of versions of the program, the quality estimation step, from the metrics value measurement result storage step, the measurement result of the metrics value for the quality estimation target program and the The measurement result of the metric value for the comparison target program is acquired, and the difference between the metric value of the quality estimation target program and the metric value of the comparison target program is determined for each metric measurement item, and the determined And calculating the quality estimate of the quality estimation target program based on Les difference metrics values.

The program quality control support method further includes a difference for extracting a difference in program content between each of the programs managed by the program management step and each of the other versions of the same revised program group. An extraction step, wherein the metric value measurement step is performed for each of the metric measurement items regarding a difference from each of the other versions of the same revised program group extracted for each program by the difference extraction step. Measuring a metric value and storing the metric value measurement result,
For each of the revised program groups, for each of a plurality of editions of the program, the measurement result of the metric value for each difference is stored, and the quality estimation step comprises:
From the metric value measurement result storing step, a metric value measurement result regarding a difference between the quality estimation target program and the comparison target program is acquired, and quality estimation of the quality estimation target program is performed based on the acquired metric value measurement result. It is characterized by calculating a value.

A program quality management support method according to the present invention is a program quality management support method for estimating the quality of a program in the program development and supporting the quality management of the program, including a program management step for managing a plurality of programs. , A specific program is designated as a quality estimation target program from the plurality of programs managed by the program management step, and at least two or more programs are compared from other programs other than the quality estimation target program. For the program specifying step to be specified as a program, and for each of the quality estimation target program and the comparison target program, a metric value measurement step for measuring a metric value for each of at least one or more types of metric measurement items. And quality evaluation coordinates composed of coordinate axes corresponding to respective metrics measurement items, and the quality evaluation coordinates for each of the quality estimation target program and the comparison target program based on the measurement result by the metric value measuring step. Identify the coordinate position in the, and perform a geometric operation based on each identified coordinate position, and estimate the quality of the quality estimation target program based on the result of the geometric operation, the quality estimation value of the quality estimation target program And a quality estimation value presentation step of presenting a quality estimation value of the quality estimation target program calculated by the quality estimation step.

A computer-readable recording medium recording a program for causing a computer to execute the program quality control support method according to the present invention estimates the quality of the program in the program development and supports the quality control of the program. A computer-readable recording medium in which a program for causing a computer to execute a management support method is recorded, and a first edition program in program development and at least one or more revised edition programs are associated with each other to form a revised program group, and revised. A program management step of managing a plurality of versions of a program by program revision for each program group, and a program of any revised version of any of the revised program groups managed by the program management step is used as a quality estimation target program. Program specification step of designating as a comparison target program, which is designated as a gram, and belongs to the same revised program group as the quality estimation target program and is a version earlier than the quality estimation target program, and the quality estimation target program and the comparison target. Regarding the program, based on a metric value measurement step of measuring a metric value for each of at least one or more metric measurement items, and a metric value measurement result by the metric value measurement step,
A quality estimation step of estimating the quality of the quality estimation target program and calculating a quality estimation value of the quality estimation target program, and a quality estimation of presenting the quality estimation value of the quality estimation target program calculated by the quality estimation step. It is a computer-readable recording medium recording a program for causing a computer to execute the program quality control support method having a value presenting step.

A program for causing a computer to execute the program quality management support method according to the present invention is for estimating the quality of the program in program development and causing the computer to execute the program quality management support method for supporting the quality management of the program. Program management step for managing a plurality of versions of the program by revising each of the revised program groups by associating the first version program in the program development with at least one or more revised version programs And a revised version of the revised program group managed by the program management step is designated as a quality estimation target program, and the quality estimation target program belongs to the same revised program group as the quality estimation target program. Program specifying step of designating a program of a version earlier than the program to be compared as a comparison target program, and measuring a metric value for each of at least one or more metric measurement items regarding the quality estimation target program and the comparison target program And a quality estimation step of estimating the quality of the quality estimation target program and calculating a quality estimation value of the quality estimation target program, based on the measurement result of the metrics value by the metrics value measurement step. And a quality estimation value presenting step of presenting a quality estimation value of the quality estimation target program calculated in the quality estimation step.

[0052]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiment 1. FIG. 1 is a program quality management support device 1 according to a first embodiment of the present invention.
It is a block diagram which shows the structure of. As shown in FIG. 1, the program quality management support apparatus 1 according to the present embodiment includes a program version management unit 2, a metrics measurement unit 4, a quality estimation unit 5, a work guide presentation unit 6, and a program designation unit 1.
It consists of zero.

Next, each component will be outlined. The program version management means 2 has a plurality of revised program groups 3 and stores and stores, for each revised program group 3, a program source code that is created and revised with the development of software. Further, when the program source code is stored and stored, the management number (version management number) of the program source code and accompanying information (version attached information) are also stored and stored. In this way, the program version management means 2 stores and saves the first version program source code and the revised program source code, and manages a plurality of versions of the program source code. The program version management means corresponds to the program management means. The program designating unit 10 designates a program source code (quality estimation target program) to be a quality estimation target and a program source code (comparison target program) to be a comparison target in quality estimation. Program designation means 10
May specify the program source code according to the input from the program developer. For example, every time the program is revised, the program specifying means 10 automatically selects the latest version of the program source code as the quality estimation target. As the program source code, the program source code of the version immediately before the latest version may be designated as the program source code to be compared. The metrics measuring unit 4 measures the metrics value for a predetermined metrics measurement item for the program source code of the quality estimation target and the program source code of the comparison target designated by the program designating unit 10. The quality estimation means 5 calculates the quality estimation value of the program source code of the quality estimation target based on the measurement result of the metrics value by the metrics measurement means 4. Here, the quality estimation value means a value indicating the possibility that the program source code has a defect, or a value indicating the possibility that the program source code has a portion to be improved or corrected. The work guide presenting means 6 presents the quality estimated value calculated by the quality estimating means 5 to the program developer, and, in a predetermined case, presents a guide to the improvement work of the program source code of the quality estimation target. The work guide presenting means 6 is
It corresponds to a quality estimation value presenting means.

The program quality control support device 1 is
Although not shown, for example, a CP such as a microprocessor
It can be realized by a computer having a recording means such as U, a semiconductor memory or the like, a magnetic disk or the like. The recording means records a program that realizes the function of each component included in the program quality control support device 1, and C
When the PU reads these programs, it is possible to control the operation of the program quality control support device 1 and realize the function of each component. It is also possible to record these programs in a computer-readable recording medium. Further, these programs can be transferred via a communication network.

Generally, in the development of a program, there is always an appropriate opportunity to revise and register the program as a turning point of the series of operations. The program version management means 2 in the present embodiment operates so as to store and accumulate the program source code created and revised with the development of software at such an appropriate timing. For example, assuming the source code of a program such as an assembler program or a C language program as the program source code, the program development work involves first describing the program as text (character string) as the source code and then using a compiler or assembler. It is converted into a machine language, then an executable file is generated, and then executed to confirm (test, verify) its operation. For this reason, the work of revising the program source code and registering it in the program version management means 2 is at least at the stage where the program source code has been created (written down) according to the implementation design of the program.
It is possible to check whether there is a grammatical error or other simple error using an assembler or a compiler, deal with the error, and convert it to a machine language (“passed” through the assembler or compiler). It will be performed at each stage, such as the stage where the new file can be generated, and the stage where the program is revised every time one or more defects found by the operation check are corrected. The program version management means 2 in the present embodiment operates so as to store the program source code saved for each revised program group at an appropriate timing in such program development work.

Next, the operation will be described. In the present embodiment, first, the program designating means 10 sets a program source code (quality estimation target program) that is a quality estimation target and a program source code (comparison target program) that is a comparison target when calculating a quality estimation value. Is specified. For example, the program source code of the latest version of any revised program group (the program source code after the revision) and the program source code of the version before the revision of the latest version are specified. Next, the metrics measuring means 4 obtains the quality estimation target program source code and the comparison target program source code designated by the program designating means from the program version management means 2, and determines each of a plurality of predetermined items (metrics). The metric value for the measurement item) is obtained and passed to the quality estimation means 5.
The quality estimating means 5 uses the metric value of the program source code of the quality estimation object and the metric value of the program source code of the comparison object passed from the metric measuring means 4 to calculate the difference (metrics value) of the metric value for each metric item. A difference value) is obtained, and a quality estimation value representing the state of the change in the quality of the program source code is obtained for the program source code of the quality estimation target based on the metric difference value, and is passed to the work guide presenting means 6. The work guide presenting means 6 presents the calculated quality estimation value to the program developer. Further, when the quality estimation values of a plurality of program source codes are calculated, the program source codes to be examined are presented to the program developer in the order of the quality estimation value. For the program source code to be examined, the program developer, based on the order presented by the work guide presenting means 6 and the respective quality estimation values, the order of performing the work on the program source code, the improvement work on each program source code, or Measures are taken after determining the necessity of implementing countermeasures such as verification work and its density (difference between intensive work and minor work).

The method of obtaining the quality estimation value from the metric difference value in the present embodiment will be described below. In the present embodiment, as the measurement items of the metrics, all the items that can be measured by the metrics measuring unit 4 or the items that are important based on general knowledge about the metrics are set. As the quality estimation value, for each measured metric item, the quality shall be increased or decreased by 1 based on the generally known qualitative guideline for that item depending on whether the metric difference value is positive or negative. The sum of the changes is used as the quality estimation value. In the case of this method, the quality estimation value indicates that the quality is improving when the value is positive, the quality is deteriorating when the value is negative, and the absolute value is the degree of change. Will be shown. The calculation formula of the quality estimation value in this method is shown below.

[0058]

[Equation 1]

As described above, in the present embodiment, since the quality of the program source code is estimated based on the known objective information using the difference of the metric values, the metric depending on the characteristics of the program source code is used. It is possible to obtain a stable software quality evaluation method that is not easily affected by fluctuations in values and does not depend on the experience and intuition of a program developer. As a result, the quality of software can be evaluated easily and quickly and stably, so that the implementation efficiency of software quality improvement measures can be increased, and higher quality software can be obtained with less effort. Become.

In this embodiment, when the program source code for quality estimation is the latest version of the program source code, the program source code for comparison uses the version corresponding to the revision immediately before the latest version. Alternatively, or alternatively, a version corresponding to a major milestone in a previous revision may be used. Further, by adopting both of these, the metric difference value is obtained for each, the quality estimated value is obtained, and the work guide presenting means 6 operates so as to present the program source code in an order in consideration of the both, so that the program developer can operate. It is possible to suggest the priority of the countermeasures against, and improve the implementation efficiency of the countermeasures.

Further, in the present embodiment, in addition to the quality estimation value for the latest version of the program source code, the quality estimation value is similarly obtained for the program source code before revision, which is one version before the latest version, and the work guide presenting means is provided. In step 6, based on the combination of these two values, the program developing worker may be operated to present a guide showing appropriate suggestions for the work. For example, if the pre-revision quality estimate indicates that the quality has deteriorated significantly, and the latest version of the quality estimate indicates that the quality has not changed much, then there is a high possibility that the defect is inherent. As a guide to encourage sufficient verification work, the quality estimate before revision indicates that the quality has not changed much, and the latest version of the quality estimate indicates that the quality has improved significantly. Suggests that there is a possibility that there is a defect in the defect detection work up to the stage before revision, and it presents a guide to revise the work content or work standard. As a result, it is possible to improve the efficiency of performing work.

In the process of selecting a guide to be presented from the above combination of quality estimation values, the combination of quality estimation values to be used and the content of the guide to be presented are held, for example, in the form of a table, and the work guide presenting means 6 is provided. May operate to obtain a work guide to be presented by searching the table. FIG. 2 shows an example of the work guide holding table 21 according to this operation. In the case of the operation based on this example,
If the quality estimation value Q> Q _th with the rule shown in the work guide holding table 21, that is, Q _th as a threshold, from the quality estimation value before revision and the quality estimation value after revision, the work guide presenting means 6 The estimated quality is "great quality improvement",
If −Q _th <Q <Q _th , the estimated quality is “small change in quality”, and if Q <−Q _th , the estimated quality is “large deterioration in quality”. Operates to find a guide for a location. As described above, when selecting the guide presented by the work guide presenting means 6, the work guide holding table 21 is searched to operate to obtain the guide, thereby flexibly determining the guide to be presented and the setting rule thereof. By properly revising the guides and rules, the work guide presenting means 6 can operate so that an effective guide can be presented to the program developer, and the work efficiency of the program developer can be further improved. You can

Further, in the above, the quality estimation values for the two versions of the program source code, that is, the latest version and the version immediately before the latest version, are used. It may operate, and the same effect can be obtained.

With respect to the method of obtaining the quality estimation value from the metric difference value in the present embodiment, the same effect can be obtained by using other methods. One example is shown below. This method weights each metric item based on the above method. The calculation formula of the quality estimation value based on this method is shown below.

[0065]

[Equation 2]

As the weighting W _i in the above equation, for example, based on general knowledge about metrics, items that are qualitatively important are set to be heavy (for example, 2) and items that are not so important are set to be light (for example, 1). There is a way. In this case, weighting can be easily determined based on publicly known qualitative information. Alternatively, using the past revision information about the program source code, W _i and q
_i (x) may be obtained. An embodiment of the present invention related to this method will be described below.

Embodiment 2. FIG. 3 is a configuration diagram showing the configuration of the program quality management support device 1 according to the second embodiment of the present invention. As shown in FIG. 3, the program quality management support device 1 according to the present embodiment has a program version management means 2, a metrics measurement means 4, a quality estimation means 5, a work guide presentation means 6, a quality estimation value calculation method holding means 7, It is composed of the program designating means 10.

In the present embodiment, the operations of the program version managing means 2, the metrics measuring means 4, the work guide presenting means 6, and the program designating means 10 are the same as those in the first embodiment, and therefore their explanations are omitted. The quality estimation value calculation method holding unit 7 corresponds to the quality estimation value calculation rule storage unit, and stores the quality estimation value calculation method (quality estimation value calculation rule) indicating the procedure for calculating the quality estimation value by the quality estimation unit 5. Hold. Here, as will be described later, the quality estimation value calculation method includes a quality estimation value calculation formula (for example, formula 3), a regression coefficient calculation formula (for example, formula 4), and a weighting coefficient. It includes a procedure for setting numerical values of W _i and a function q _i (x) indicating qualitative evaluation (for example, W _i is an absolute value of regression coefficient α _i ) and the like.

Among the operations of the quality estimating means 5, the method of obtaining the quality estimated value from the metric difference value will be described below. This method is the same method as the second quality estimation value calculation method described in the first embodiment, and weights each metric item. The calculation formula of the quality estimation value based on this method is shown below.

[0070]

[Equation 3]

When the above-mentioned calculation formula is used as the quality estimation value calculation method, the quality estimation value calculation method holding means 7 holds the weighting coefficient W _i for the metrics and the function q _i (x) (i = 1 to 1) indicating the qualitative evaluation. n), and the quality estimating means 5 holds W _i and q _i (x) held by the quality estimated value calculating method holding means 7 when obtaining the quality estimated value of the program source code. ) Is used to calculate the quality estimation value. Further, the numerical values of W _i and q _i (x) held by the quality estimation value calculation method holding means 7 can be changed by the quality estimation means 5, and W _i and q _i (x).
The respective numerical values of are obtained by the quality estimating means 5 using the past revision information regarding the program source code, and are set in the quality estimated value calculating method holding means 7. These W
_i and q _i (x) correspond to the parameters included in the quality estimation value calculation rule.

Using the past revision information about the program source code, which is performed by the quality estimating means 5, W _i and q
An example of a method for obtaining each numerical value of _i (x) will be described with reference to FIG. FIG. 4 is an explanatory diagram used for explaining the present method. First, the quality estimating means 5 extracts the respective program source codes from the program version managing means 2 for the versions 1 to m of the m program source codes existing in the revised program group of the quality estimation subject to the past revision work. A metric value is obtained using the metric measuring means 4, and the metric value at this time is defined as M _ij (where i is a metric item number 1 to n, and j is a version management number 1 to m), and the difference Value ΔM _ij (= M _{ij + 1} −
M _j ) is obtained (i is the item number of the metric, 1 to
n and j are version control numbers 1 to m-1). Further, based on version attached information such as a comment at the time of revision held by the program version management unit 2 or based on failure information for each version that is separately managed, a defect detected in the revision from version j to version j + 1 Is calculated, and this value is set as F _j (j is a version management number 1 to m−1). Then, these values are applied to the following linear model of the objective variable F, the explanatory variable ΔM _i , and the error ε.

[0073]

[Equation 4]

Next, the quality estimating means 5 performs m-2 sample [F] by multiple regression analysis which is one of the statistical methods.
_{j + 1} , ΔM _1j , ΔM _2j , ..., ΔM _nj ] (j = 1 to m−
From 2), the regression coefficient α ₀ ... α _n in the same linear model is obtained. Since the above linear model is a model for estimating the number of defects F _{j + 1} to be treated in the version j + 1 to the version j + 2 from the metric value changed in the revision from the version j to the version j + 1, the obtained α _i is the measurement of the metrics. The difference value of the item i is a coefficient that contributes to the detected defect. Since the defect detected here can be considered to have a strong correlation with the defect that occurs, the obtained α _i can be estimated as a coefficient that strongly correlates with the defect that the difference value of the measurement item i occurs, so that the value of α _i is positive. When it If metric value increases in the quality decreases, also the absolute value of the value sets the value of q _i (x) as the quality increases the metric value increases if it is negative, also alpha _i, or absolute The value of W _i is set according to the order of the magnitude of the value.

When the absolute value of the regression coefficient α _i corresponding to the metric measurement item i obtained above is relatively small with respect to other regression coefficients, the measurement item i affects the calculation of the quality estimation value. However, it may be excluded from the measurement items when the quality estimation is performed, and the measurement work can be simplified.

In the above, the multiple regression analysis that handles a plurality of explanatory variables at the same time is used as the analysis method, but the regression coefficient may be obtained by using the single regression analysis for each metric measurement item. Further, when only one metric measurement item is used, n = 1, and thus the regression coefficient is inevitably obtained by simple regression analysis.

Further, in the above, the sample is obtained from the past revision information for the program source code, but the number of samples used at this time has a regression coefficient at a sufficient significance level (for example, 95%) based on the knowledge of statistics. It is only necessary to calculate using the required sample size, and it is not necessary to calculate using all revision information.

Similarly, the respective numerical values of W _i and q _i (x) in the quality estimation value calculation formula do not have to be recalculated each time when the quality estimation value is calculated, but may be calculated from time to time at appropriate timing. Is also good. In this case, for example, when several new versions of the program source code in which the number of detected defects have been clarified become available in line with the progress of software development, The quality estimation value is calculated, the obtained quality estimation value is compared with the actual quality, the estimation accuracy is evaluated, and when the estimation accuracy is out of the allowable range, W _i and q _i in the quality estimation value calculation formula are calculated. It suffices to recalculate each numerical value of (x). In order to evaluate the estimation accuracy, for example, the order in which the quality of the program source code, which is determined by the quality estimation value, is considered to be good, and it is compared with the order in which the actual quality is good (the order with the fewest defects) The degree of coincidence may be digitized as a correlation coefficient between the two using knowledge of statistics, or another method may be used.

Further, it may be possible to obtain it by using the past revision history of another revised program group which is not the revised program group which is the subject of the quality estimation, or the program source code in the other program development work performed in the past. Alternatively, the revision information may be used. In this case, for example, the determination of the numerical values of W _i and q _i (x) is
When the program development work is started, the work load for determining the numerical values of W _i and q _i (x) is sufficiently small by obtaining the value based on the result of the previous program development work. can do.

Alternatively, instead of the detected number of defects, the analysis may be performed by using the score of the defect weighted according to the content of the defect. For example, the detected defects are classified into three levels from 1 to 3 (the most serious is 3 and the least is 1) according to the seriousness of the content, and the value is detected as the score of each defect. It is a method of performing analysis by using the sum of the scores of each defect instead of the number of detected defects. In this case, the metric difference value is correlated with the defect in consideration of the influence on the quality, and the influence on the quality is more emphasized and W _i and q _i (x)
The numerical value of can be set, and the effect that a more appropriate quality estimation value can be expected can be obtained.

The value of α ₀ when the multiple regression analysis is performed in the above linear model (Equation 4) represents the number of defects detected by the revision work regardless of the change in the metrics value. If the value becomes significantly larger,
It can be inferred that a defect is detected for each revision regardless of the change in the metrics value, that is, the quality verification accompanying the revision is insufficient. Therefore, when the value of α ₀ becomes a significantly large value when performing the multiple regression analysis in the linear model (Equation 4), the work guide presenting means 6 is used to notify the program developer of the value of α ₀ . Since it is large, it may operate so as to present a guide to the effect that it is better to review the contents of the quality verification work accompanying the revision. In this case, the content of suggestions to the program developer can be enhanced, and the implementation efficiency of the measures can be improved. Note that this operation is _Wi
And it not only performed in association with the time of determining the value of q _i (x), to make a more aggressive testing, appropriate alone multiple regression analysis is performed only alpha ₀ of the value of the linear model (Equation 4) It may operate to perform an inspection.

The result of the multiple regression analysis in the linear model (Equation 4) can be similarly applied to the parameter setting in other quality estimation value calculation methods. Examples of other quality estimation value calculation methods are shown below. In the calculation formula of the quality estimation value in this method, W _i = −α _i may be set based on the regression coefficient α _i obtained by the multiple regression analysis, and q _i (x)
The calculation is simplified because is not used.

[0083]

[Equation 5]

Further, the parameters in the quality estimation value calculation formula can be set by using other regression analysis methods instead of the above-mentioned regression analysis method. For example, in the multiple regression analysis using the above linear model (Equation 4), the sample to be used is m
-Two samples [(F _j + F _{j +1} ), ΔM _1j , ΔM _2j ,
, ΔM _nj ] (j = 1 to m−2), the multiple regression analysis may be performed. In this case, the linear model (Equation 4) is
From version j to version j + 1 is a model for estimating the number of defects F _j + F _{j + 1} detected from version j to version j + 2 from the metric value changed. The α _{i obtained by} this model is a coefficient that contributes to the problem that the difference value of the measurement item i of the metric is detected as in the above model. This is different from the above-mentioned model in that it is also associated with the generated defect F _j, and thus the accuracy of the correlation of the obtained coefficients may be enhanced, and a more appropriate quality estimation value can be expected. Further, in general, when the metric value greatly changes, it is considered that the quality has a great influence. Therefore, the explanatory variable is not a first-order ΔM _i but a second-order (ΔM _i ) ² or a third-order (ΔM _i ) ³ analysis. You can go and get the same effect.

Furthermore, a plurality of regression analysis methods and a plurality of quality estimation value calculation formulas may be used in combination. In this case, by performing the following work in advance, an appropriate method can be selected as a quality estimation value calculation method used for quality estimation. This process will be described with reference to FIG. FIG. 5 is a flowchart showing the operation of the step S10 of selecting the quality estimation value calculation method.

In the step S10 of selecting a quality estimation value calculation method, the quality estimation means 5 first selects a candidate quality estimation value calculation method (selected quality estimation value calculation rule) and becomes the selected candidate. In order to set the parameter in the quality estimation value calculation method, an appropriate regression coefficient for the metric value is obtained by regression analysis on the program source code (S11), and then the parameters of each candidate quality estimation value calculation method are calculated from the regression coefficient. Set (S1
2) After that, the quality estimation value calculation method used for actual quality estimation is selected by evaluating each quality estimation value calculation method (S13).

Details of each process will be described below. In the process S11 of obtaining an appropriate regression coefficient for the metric value by the regression analysis on the program source code, the quality estimating unit 5 firstly, for each of the plurality of regression analysis methods,
Analysis work is performed using the program source codes stored in the program version management means 2 to obtain the regression coefficient and its accuracy (S111, S112, S11).
3). Specifically, as described above, the metrics measuring means 4 obtains the metrics value of each metrics measurement item with respect to a specific plurality of program source codes, performs regression analysis based on the measurement result of the metrics value, and with respect to each regression coefficient. To seek accuracy. Next, based on the most accurate analysis result in each regression analysis method, the combination of the metric measurement item and the regression coefficient used thereafter is determined (S114). Next, in the process S12 of setting the parameters of each quality estimation value calculation method, the process S12 is performed for each of the plurality of candidate quality estimation value calculation methods.
From the combination of the metric measurement item and the regression coefficient obtained in 11, the numerical values of the parameters used in each calculation method such as the weighting W _i and q _i (x) in the above calculation method (Equation 3) are set (S121, S12
2, S123). Then, in the selection process S13 of the quality estimation value calculation method used for the actual quality evaluation, first, the quality can be compared from the program source code stored in the program version management means 2 based on, for example, the number of detected defects. A plurality of past program source codes are selected (S131), and a quality estimation value is calculated for each program source code for each of a plurality of candidate quality estimation value calculation methods, and the result and each program source are calculated. Each calculation method is evaluated by comparing it with the actual quality of the code (S132, S133, S134, S
135), based on the evaluation result, one or several kinds of quality estimation value calculation methods used for actual quality estimation are designated (S).
136). The quality estimation value calculation method designated by the above processing is the designated quality estimation value calculation rule. The process S1
As an evaluation of the quality estimation value calculation method performed in step 34, for example, the order of the program source code that is determined to be good in quality determined from the quality estimation value is obtained, and is determined as the order in which the actual quality is good (for example, the order in which there are few defects). The comparison may be performed and the degree of coincidence between the two may be digitized as a correlation coefficient between the two using statistical knowledge, or another method may be used.

Although the selection process S10 may be a relatively complicated process, it can basically be automatically executed, and when the program development work is started, the result of the previous program development work is Since it is necessary to recalculate the quality estimation value when the accuracy of the quality estimation value has deteriorated, it is possible to reduce the frequency of execution sufficiently, and It can be operated so that it does not become a large workload. As a result, it is possible to avoid a situation in which an incorrect quality estimation value is used by continuing to use an inaccurate quality estimation value calculation method, which makes it difficult to improve work efficiency. Note that the verification of the accuracy of the quality estimation value is performed by the same processing as that performed in the evaluation of the quality estimation value calculation method described above, that is, by using a plurality of past program source codes capable of comparing the quality by the number of detected defects, for example. It suffices to select it, obtain a quality estimation value for each source code, and compare both the quality estimation value and the actual quality.

When the quality estimation value calculation method is set by performing the above multiple regression analysis, the model used in the process of the analysis is a calculation formula capable of estimating the number of defects F detected from the metric difference value ΔM _i. Therefore, when the quality estimating means 5 uses the work guide presenting means 6 to present the program source code to be examined for quality estimation to the program developer in the order of the quality estimation value, the defect is also estimated. The number F may be presented. In this case, the program developer considers the number of defects F estimated in addition to the order presented by the work guide presenting means 6 and the quality estimation value, and considers the order of implementation of the work for the program source code and the content of measures. Therefore, the implementation efficiency of the measures can be improved. Further, since the program developer can use the estimated number of defects F as a guide for detecting defects and ending the treatment work, unnecessary work can be reduced and work efficiency can be improved.

In the above, the number of defects detected as the objective variable F was used in the multiple regression analysis (Equation 4) for setting the quality estimation value calculation method, but this is another numerical value related to quality. Possible parameters, such as the number of tests or man-hours required for defect detection, or the man-hour required for treatment of detected defects, may be used, and similar effects can be obtained.

Alternatively, the regression coefficient α _i , that is, the coefficient indicating the correlation between the difference value of the measurement item i and the quality is derived from various criteria by performing multiple regression analysis on each of these plural different parameters as the objective variables. You may In this case, the weighting given to the quality by the measurement item i can be comprehensively evaluated and set by using the regression coefficient for the parameter for which high accuracy is obtained or by using a plurality of regression coefficients. The accuracy of can be improved.

Alternatively, a plurality of quality estimation value calculation methods may be set based on the results of multiple regression analysis using these plurality of different parameters as objective variables, and the quality estimation means 5 may operate to use them in combination. . In this case, the content of suggestions to the program developer can be enhanced, and the implementation efficiency of the measures can be improved. Further, as in the case of using the number of defects F described above, the values of the respective parameters subjected to the multiple regression analysis can be estimated from the metric difference values, so that the respective estimated values for the respective parameters are provided as work guide presenting means. You may make it display by adding 6 together. Since these estimated values can be used as a guide for finishing the treatment work to be performed, like the above-described estimated value of the defect number F, unnecessary work can be reduced and work efficiency can be improved.

When a plurality of quality estimation values are used together, the work guide presenting means 6 can select and present an appropriate work guide according to the type of quality estimation value estimated to be of poor quality. For example, if it is estimated that the quality is poor in terms of the number of detected defects, it is suggested to focus on the detection work. It may be operated such as suggesting that the source code review should perform the trouble handling by a plurality of people. As a result, the contents of suggestions to the program developer can be enhanced, and the implementation efficiency of the measures can be improved.

Also, in the above, when obtaining the quality estimation value calculation method for the program source code, each version was handled without considering the contents of the program development work that caused the revision of the version. The quality estimation value calculation method may be determined and used according to the cause. This is, for example, based on version attached information such as a comment at the time of revision held by the program version management means 2, revision of the version by function realization (function extension), revision of the version by program review, and revision of the version by fault correction. The quality estimation value calculation method is determined by using the version of the program source code revised according to the same work content for each of them, and the quality estimation value calculation method holding unit 7
Holds a plurality of quality estimation value calculation methods by analyzing the quality estimation value calculation method for each revision cause. When calculating the quality estimation value of the program source code, the quality estimation means 5
Operates to use a quality estimation value calculation method corresponding to the work content that has caused the revision of the target version. By operating in this way, the distribution of differences (variations) in the metric values for each program source code is handled as a whole without considering the work content that caused the revision of the version. It can be expected to be smaller than that in the case, and improvement in the accuracy of quality estimation can be expected.

In the above, the program source code stored in the program version management means 2 is handled as a program source code, but the quality estimating means 5 operates as a program source code of each revised program group. If the code is divided into several groups with different tendencies of the metric values, and if you want to evaluate a certain program source code or set the quality estimation value calculation method, the program source code in the group to which the program source code belongs You may operate so that it may be closed and the above-mentioned process may be performed. For example, S11 to S11 shown in FIG. 5 for the program source code belonging to the same group as the program source code to be quality estimated.
The process of S12 can be performed to set the quality estimation value calculation method for each group. Further, the quality estimation means 5 can calculate the quality estimation value of the program source code of the quality estimation target by using the quality estimation value calculation method set in this way. In this case, the degree of dispersion of the metric values for each program source code in the group is smaller than that in the case where the entire program source code is handled collectively, so that the accuracy of quality estimation is improved.

As a method of dividing the program source code groups, for example, there is a method of focusing on the difference in the functions (graphic user interface, database access, state transition processing, numerical operation, etc.) realized by the program source code. Alternatively, the creators of the program source code may be divided into groups. Further, cluster analysis may be performed using statistical knowledge based on the metric values of the program source code, and as a result, groups that can be significantly separated may be used. In the method using cluster analysis, the degree of similarity of the metric values can be evaluated statistically exactly, and therefore the effect of accuracy improvement can be expected to be further enhanced. Note that the version of the program source code when measuring the metrics value for classification is, for example, the stage where the grammatical error detected by the assembler or compiler has been corrected, or the defect detected during the first operation check has been completed. It is better to unify the versions that correspond to the same trigger in program development work, such as stages
The possibility of more effective classification can be increased.

The quality estimation value calculation method obtained by collectively handling the program source code held in the program version management means 2 and the quality estimation value calculation method obtained by dividing each group into groups are as follows. Can be regarded as the information indicating the correlation between the metric measurement item and the quality estimation value, and by comparing the two, it is possible to obtain the difference in the influence of the metric measurement item on the quality estimation value. The characteristics of can be extracted. Especially when grouping by creator, characteristics that depend on the creator can be extracted. For example, a work guide is presented in which a specific creator has a higher correlation between a specific metric measurement item and a quality estimate than the overall tendency. The above information may be useful. Therefore, the quality estimation means 5 obtains the first quality estimation value calculation method by collectively handling the program source code held in the program version management means 2, and the program corresponding to the individual creator. A second quality estimation value calculation method was obtained using only the source code, and by comparing the two, the effect of the metrics measurement items on the quality estimation value depending on the creator was obtained, and the same information was presented as a work guide. Even if the work guide presenting means 6 operates to select and present an appropriate guide in consideration of the difference in characteristics of the creator based on the same information passed from the quality estimating means 5. well,
In this case, the effect of the presented work guide can be enhanced, and the work efficiency can be further improved.

When the program source code is divided into groups and handled, instead of the quality evaluation method described above,
It is also possible to use the quality evaluation method shown below. This quality evaluation method will be described with reference to FIG. FIG. 6 is an explanatory diagram for explaining the concept of this quality evaluation method.

In this quality evaluation method, first, as a preparation, the program source codes are grouped and representative points are calculated for each group. In grouping, the quality of the program source code stored in the program version management means 2 is first determined based on version-related information such as a comment at the time of revision or separately managed failure information for each version. The program designating means 10 compares the plurality of program source code versions that can be judged to be good (for example, the version of the program source code developed in the past program development work after all the troubles have been dealt with) by the program source code to be compared. , And the quality estimating means 5 measures the metric value using the metric measuring means 4 for each. Then, the quality estimating means 5 groups the program source codes by using cluster analysis based on the measured metric values, selects significant metric measurement items ₁ to n, and weights W ₁ to
Set W _n . For weighting, W _{1 to} W _n use the same method as described above, that is, the following linear model of the objective variable F, the explanatory variable ΔM _i , and the error ε.

[0100]

[Equation 6]

That is, in the above linear model, the objective variable is the number of defects, and the explanatory variable ΔM _i is the metric measurement item i.
Multiple regression analysis is performed as a metric difference value of (i = 1 to n), and W _i = α _i is determined based on the obtained regression coefficient α ₀ ... α _n . At this time, if the absolute value of the regression coefficient α _i is sufficiently smaller than the others, the corresponding metric measurement item i may be omitted from the metric measurement target as described above.

By defining n metric measurement items as a scale, the program source code to be quality-estimated and the program source code to be compared are:
It is associated with one point on the n-dimensional quality measurement space (quality evaluation coordinates) with each metric measurement item as a coordinate axis, which is determined by the metric value of each metric measurement item.
At this time, in order to correct the distance along each axis in space,
The measured value M _i of metrics measurement item i with weights W _i, M _'i = W _i · M _i and with variable transformation, the M' quality estimation unit 5 using the _i is in the space of each point Determine the coordinate position. Then, for example, a representative point (comparison target coordinate position) of each group is set as the center point of a sphere containing 95% of the program source code belonging to each group. The quality estimating space is held by the quality estimating means 5.

FIG. 6 is a diagram showing an example of the quality measurement space and the result of the above processing when two measurement items 1 and 2 are used as metrics measurement items. In the figure, the point indicating the program source code to be examined, that is, the program source code to be quality estimated is shown in the center. Further, a large number of white dots are points corresponding to respective program source codes to be compared, and in this example, they are classified into two groups A and B as a result of cluster analysis. Black dots indicate representative points of each of the groups A and B. The dotted circle is a circle (sphere in the two-dimensional space) that contains 95% of the program source code (program source code used for the cluster analysis) belonging to the group centering on the representative point of each group A and B. ) Is shown.

Next, a method of calculating a quality estimation value for a program source code based on this evaluation method will be described.
In the following, an example in which two measurement items 1 and 2 are used as the metric measurement items will be shown in association with FIG.

The program source code to be examined is located at the coordinates (M ′ ₁ , M ′ ₂ ) in the space = (W ₁ · M ₁ , W ₂ · M ₂ ) according to its metric values M ₁ and M _2. Points (large white points in the figure), the distance between this point and the representative point of each group A and B in the space is first obtained, and the closest representative point is selected (group A in this figure). Select the representative point of). Then, the distance between the point corresponding to the program source code to be examined and the representative point of the closest group A is set as the quality estimated value of this program source code. In this method, since the representative point of the group is the representative of the program source code of good quality, the quality is worse as the distance is longer, that is, the quality estimation value is larger as the quality estimation value is larger. Evaluate the quality of. While the metric difference value or the metric value for the difference in the program source code represents the estimation for the change in quality, the distance between the point corresponding to the difference in the program source code and the representative point of the group is the estimation for the quality itself. As a result, it is possible to obtain more direct guidance in quality evaluation. That is, it is possible to improve the implementation efficiency of the measures by giving the program development worker a guideline more appropriate as the priority of the measures. In addition, performing grouping based on the distance in the quality measurement space makes it easy for the program developer to intuitively understand the positions of representative points and quality estimation values to be compared, and the quality estimation values and It can be expected that the psychological resistance and repulsion to the suggested guide will be reduced.

Alternatively, as the quality estimation value, a metric difference value (difference between the metric value before revision and the metric value after revision) is used instead of the distance between the point of the program source code to be considered and the representative point of the group. The quality estimation value may be obtained based on the vector Δ on the space shown and the vector D obtained from the difference between the metric value before revision and the metric value of the representative point of the group. In this case, the quality estimation value is the length L _Δ of the vector Δ of the metric difference value, and the vector D obtained from the difference between the metric value before revision of the metric difference value vector Δ and the metric value of the representative point of the group. It can be obtained as a set of component lengths L _{D · Δ} with respect to the direction.

[0107]

[Equation 7]

However, L _{Δ of the} quality estimation value is a numerical value corresponding to the amount of work involved in revision, and L _{D · Δ} can be regarded as a numerical value corresponding to the effect of improving the quality in the work ( The quality estimating means 5 evaluates the quality of the program source code on the basis of the values of both (if the value is positive, the quality is improved, and if the value is negative, the quality is decreased). Further, the work guide presenting means 6 is appropriate based on the values of both, for example, in the case where L _{D · Δ} is small and L _Δ is large, it is assumed that a lot of work that cannot be measured as a change in quality has been carried out and the verification work is reinforced. It is possible to operate to present various guides. With this method, it is possible to infer that work that cannot be measured as a change in quality has been performed, so we suggest a guide that strengthens the verification work as it is possible that inappropriate work is being performed, Alternatively, it is possible to present a guide that prompts the review of the work content itself that has been carried out, whereby the content of suggestions to the program development worker can be enhanced, and the implementation efficiency of the measures can be improved. In addition, performing grouping based on the distance in the quality measurement space makes it easy for the program developer to intuitively understand the positions of representative points and quality estimation values to be compared, and the quality estimation values and It can be expected that the psychological resistance and repulsion to the suggested guide will be reduced.

In the above, some of the metric measurement items to be used have a strong correlation with each other, and if there is a metric measurement item that does not provide mutual independence suitable for use as a coordinate axis of a space, if there is a metric measurement item, the main item is It may be used as a representative, or by using principal component analysis, which is one of the statistical methods, the metrics measurement items are summarized into synthetic variables that are uncorrelated with each other, and the synthetic variables are used as coordinate axes of the space. May be. As a result, the deviation of the distribution of the measurement points in the space is reduced, and the effect of improving the accuracy of quality evaluation can be obtained.

Further, in the above, as the method of selecting the representative point of the group to be compared, the representative point having the closest distance to the point corresponding to the program source code of the inspection object is selected. , For example, the points (M _1j , ... M _nj ) (j is the version control number) corresponding to each metric value are plotted on the quality evaluation space for the revision of the program source code that has already been performed, and the change From the state (rough direction of each difference vector), the direction in which the point corresponding to the future version is likely to exist is inferred, and the representative point of the immediate group existing in that direction is set as the representative point of the comparison target group. You may use the method of selecting. When this method is adopted, for example, in the example of FIG. 6, the representative point closest to the point representing the inspection target is the representative point of group A,
From the state of change (vector of metric difference value), it can be inferred that the process proceeds in the direction of group B, so the representative point of group B is selected as the representative point for comparison. Further, in the above description, one representative point of the group is selected and the processing is performed, but this may be performed for all the representative points. In addition, in the grouping performed as a preliminary preparation, if a group that can be clearly separated is not obtained by the cluster analysis, the whole may be treated as one group and processed. Further, the above-described quality estimation value calculation method may be used in combination with another quality estimation value calculation method. With this method, it is possible to select an appropriate representative point based on the state of revision from the past, rather than just a representative point at a certain moment, so the accuracy of the obtained quality estimation value improves,
Therefore, it becomes possible to present more appropriate guidelines to the program development worker, and it is possible to improve the efficiency of implementing the measures. In addition, performing grouping based on the distance in the quality measurement space makes it easy for the program developer to intuitively understand the positions of representative points and quality estimation values to be compared, and the quality estimation values and It can be expected that the psychological resistance and repulsion to the suggested guide will be reduced.

Third Embodiment FIG. 7 is a configuration diagram showing the configuration of the program quality management support device 1 according to the third embodiment of the present invention. As shown in FIG. 7, the program quality management support apparatus 1 according to the present embodiment includes a program version management unit 2, a metrics measurement unit 4, a quality estimation unit 5, a work guide presentation unit 6, a metrics measurement information storage unit 8, and a program designation. It comprises means 10.

In the present embodiment, the operation of the program version management means 2 is the same as that of the first embodiment, so its explanation is omitted.

As in the first embodiment, the metrics measuring means 4 uses the program version managing means 2 to update the latest program source code of a series of program source codes subject to quality examination. Obtain the (revised program source code) and the previous version (pre-revised program source code) of that program source code, and obtain the metric values for each of a plurality of predetermined items. The program source code is registered and stored in the metrics measurement information storage means 8 (metrics value measurement result storage means) in association with each identification information (name), version management number, and measured metrics value. In addition,
When there are a plurality of program source codes before revision, a metric value is obtained for each of the plurality of program source codes, and each metric value is registered and accumulated.

Next, the program designating means 10 designates a program source code to be a quality estimation target and a program source code to be a comparison target. The quality estimating means 5 determines the metrics value corresponding to the program source code to be quality estimated and the metrics value corresponding to the program source code to be compared based on the identification information (name) and the version control number of the program source code. Then, the metrics measurement information storage means 8 is searched to obtain the respective values. At this time, if the metric value for the corresponding version management number has not been accumulated yet, the metric value relating to the program source code of the relevant version is measured by the metric measuring means 4 via the metric measurement information accumulating means 8. The metric measuring means 4 obtains the program source code of the corresponding version using the program version managing means 2, measures the metric value thereof, and registers it in the metric measurement information accumulating means 8, whereby the quality estimating means 5 is obtained. Gets the required metric values.

After that, the quality estimating means 5 uses the metrics value of the program source code of the quality estimation object and the metrics value of the program source code of the comparison object obtained from the metrics measurement information accumulating means 8 in the first embodiment. Similarly, the difference between metrics values (metrics difference value) is calculated for each metric item, and the quality estimation value representing the state of the change in the quality of the program source code is calculated based on the metric difference value. 6
Hand over to. Note that the method of obtaining the quality estimation value from the metric difference value in the present embodiment is the same as in the first and second embodiments, and therefore its explanation is omitted. Further, the operation of the work guide presenting means 6 is also the same as in the first and second embodiments, and therefore its explanation is omitted.

As described above, in the present embodiment, the quality of the program source code is estimated based on the known objective information using the difference between the metrics values, as in the first embodiment. , Is not easily affected by the fluctuation of the metric value depending on the characteristics of the program source code, and does not depend on the experience and intuition of the program developer.
It is possible to obtain a stable software quality evaluation method. As a result, the quality of software can be evaluated easily and quickly and stably, so that the implementation efficiency of software quality improvement measures can be increased, and higher quality software can be obtained with less effort. Become.

Further, since the metric value corresponding to each version of the program source code operates so as to be accumulated and stored in the metric measurement information accumulating means 8, the metric value measured by the metric measuring means 4 for each version of the program source code. Since the value measurement work only needs to be done at most once for each version of each program source code, the amount of processing can be greatly reduced, which can shorten the processing time required for software quality evaluation. It is possible to obtain the effect that the work efficiency is improved.

Fourth Embodiment FIG. 8 is a configuration diagram showing the configuration of the program quality management support apparatus 1 according to the fourth embodiment of the present invention. As shown in FIG. 8, the program quality management support apparatus 1 according to the present embodiment has a program version management unit 2, a metrics measurement unit 4, a quality estimation unit 5, a work guide presentation unit 6, a program revision location extraction unit 9, and a program designation. It comprises means 10.

In this embodiment, the operation of the program version management means 2 is the same as that of the first embodiment, and therefore its explanation is omitted.

First, the program designating means 10 designates a program source code to be a quality estimation target and a program source code to be a comparison target. Next, the program revision point extraction means 9 extracts the difference in the program content between the designated program source code for quality estimation and the program source code for comparison, and the difference information indicating the extracted difference. To the metrics measuring means 4. The program revision location extraction means 9
It corresponds to the difference extracting means. The metrics measuring means 4 is
Metric values are obtained for a plurality of predetermined items for the difference information and passed to the quality estimating means 5.

The quality estimation means 5 obtains a quality estimation value representing the state of the quality change of the program source code based on the metrics value passed from the metrics measurement means 4,
Hand it over to the work guide presenting means 6. The work guide presenting means 6 presents the calculated quality estimation value to the program developer. Further, when the quality estimation values of a plurality of program source codes are calculated, the program source codes to be examined are presented to the program developer in the order of the quality estimation value. For the program source code to be examined, the program developer, based on the order presented by the work guide presenting means 6 and the respective quality estimation values, the order of performing the work on the program source code, the improvement work on each program source code, or Judging the necessity of implementation of countermeasures such as verification work and its density (difference between focused work and minor work),
Take measures.

A method of obtaining a quality estimation value from a metric value in this embodiment will be described below. In this embodiment,
As the metric measurement items, all items that can be measured by the metric measurement unit 4 or items that are important based on general knowledge about the metric are set. In addition, in the measurement of metrics, the metrics value is measured for each of the contents deleted by the revision and the contents added by the revision, which are included in the difference information between the revised program source code and the unrevised program source code. To do. For the parts that have been changed due to the revision, the contents before the revision will be included in the contents deleted due to the revision, and the contents after the revision will be included in the contents added due to the revision.

The calculation formula of the quality estimation value in this method is shown below.

[0124]

[Equation 8]

Further, the coefficient W _i in this calculation formula is obtained by using the following linear model of the objective variable F, the explanatory variable ΔM _i , and the error ε, as in the quality estimation value calculation method in the second embodiment.

[0126]

[Equation 9]

That is, in the above linear model, the objective variable is the number of defects, and the explanatory variable ΔM _i is the metric value ΔMd of the metric measurement item i (i = 1 to n) with respect to the difference information.
The difference between _i and ΔMa _i is the sample [F _s , ΔM _1s , ΔM
_2s , ..., ΔM _ns ] (s is a sample number, ΔM _is = ΔMd
_is −ΔMa _is ), multiple regression analysis is performed, and W _i = −α _i is determined based on the obtained regression coefficients α ₀ ... α _n . At this time, if the absolute value of the regression coefficient α _i is sufficiently smaller than the others, the corresponding metric measurement item i may be omitted from the metric measurement target as in the first embodiment, and the measurement work is simplified. be able to.

As described above, the present embodiment is based on the calculation method determined by using the difference information of the program source code, that is, the metric value based on the revision information, and the objective information of the number of detected defects. Since it operates so as to estimate the quality of the program source code, it is unlikely to be affected by the fluctuation of the metric value depending on the characteristics of the program source code, as in the first embodiment, and the experience of the program developer or It is possible to obtain a stable software quality evaluation method that does not depend on intuition. As a result, the quality of software can be evaluated easily and quickly and stably, so that the implementation efficiency of software quality improvement measures can be increased, and higher quality software can be obtained with less effort. Become.

Further, since the metrics value is measured and used for each of the content deleted by the revision and the content added by the revision included in the difference information of the program source code, different from the first embodiment, It can operate so as to obtain the metric value even for the information that disappears from the program source code due to the revision, more reliably grasp the change in quality due to the contents of revision, and improve the accuracy of software quality evaluation. The effect of being enhanced is obtained.

Regarding the method of obtaining the quality estimation value from the metric value in the present embodiment, another quality estimation value calculation method may be used as in the second embodiment. In any case, the above calculation is performed. The same effect as the method can be obtained. An example of the calculation formula of the quality estimation value by another method in the present embodiment is shown below.

[0131]

[Equation 10]

The coefficient W _i in this calculation formula is obtained using the following linear model of the objective variable F, the explanatory variable ΔM _i , and the error ε.

[0133]

[Equation 11]

That is, in the above linear model, the objective variable is the number of defects, and the explanatory variable ΔM _i is the metric value ΔMd of the metric measurement item i (i = 1 to n) with respect to the difference information.
_As the sum of _i and ΔMa _i , the sample [F _s , ΔM _1s , ΔM
_2s , ..., ΔM _ns ] (s is a sample number, ΔM _is = ΔMd
_is + ΔMa _is ), multiple regression analysis is performed, and W _i = −α _i is determined based on the obtained regression coefficients α ₀ ... α _n .

Further, as in the second embodiment, a plurality of quality estimation value calculation methods are used together, or an appropriate one or a plurality of methods are selected from the plurality of quality estimation value calculation methods based on the accuracy of the quality estimation. May be used as a second
The same effect as the operation in the above embodiment can be obtained. Alternatively, the quality estimation value obtained in this embodiment, that is, the quality estimation value based on the difference information associated with the revision of the program source code, and the quality estimation value obtained in the first or second embodiment, that is, the program source code Each of the quality estimation values based on the metric difference value before and after the revision may be used together, and the quality of the program source code can be grasped more multilaterally, and the accuracy of the quality evaluation can be further improved. can get.

Further, in the present embodiment, as in the third embodiment with respect to the first embodiment, the metrics measurement information storage means 8 may be added. In this case, the program revision point extracting means 9 extracts the difference between the program source code of each version and the program source code of another version, and the metrics measuring means 4 extracts the metrics corresponding to the difference information due to the revision of each version. The value is measured, and the metric value based on the measured difference information of each version operates so as to be accumulated and stored in the metric measurement information accumulating unit 8. By doing so, similarly to the third embodiment, the processing amount of the measurement of the metric value performed by the metric measuring unit 4 can be significantly reduced, and thereby the processing time required for the software quality evaluation can be shortened. You can
The effect that work efficiency improves can be obtained.

In the above first to fourth embodiments, the program quality management support apparatus according to the present invention has been described, but the program quality management support method according to the present invention can also be realized by an equivalent processing procedure.

Here, the features of the program quality management support apparatus and the program quality management support method described in the first to fourth embodiments are summarized as follows.

The present program quality control support device is composed of a metric measuring means, a quality estimating means, and a work guide presenting means. The metric measuring means includes a program source code before revision and a program source code after revision. It operates so as to measure each metric value, and the quality estimation means obtains the difference (change amount) between the metric value of the program source code before revision and the metric value of the program source code after revision, and based on that change amount. The operation guide presenting means operates so as to estimate the quality of the program source code, and the work guide presenting means operates so as to present the quality of the program source code estimated by the quality estimating means.

Also, the program quality control support device
The program revision portion extracting means, the metrics measuring means, the quality estimating means, and the work guide presenting means are provided, and the program revision portion extracting means obtains the difference information between the program source code before revision and the program source code after revision. The metric measuring unit operates to measure a metric value for the difference information, and the quality estimating unit operates to estimate the quality of the program source code from the metric value for the difference information. The guide presenting means operates so as to present the quality of the program source code estimated by the quality estimating means.

It is characterized in that the work guide presenting means operates so as to select and present an appropriate suggestion for the work based on the quality information for the program source code passed from the quality estimating means.

The quality estimating means operates so as to obtain respective quality estimation values corresponding to a plurality of revisions of the program source code, and the work guide presenting means determines the combination of the quality estimation values of the respective editions. On the basis of this, it operates so as to select and present an appropriate suggestion for the work.

The quality estimating means uses the program version managing means for holding the revision history of the program source code,
Obtain the program source code corresponding to each revision edition,
A metric value corresponding to each is obtained using the metric measuring means, and a quality estimated value calculation method for estimating the quality of the program source code from the metric value based on these values is obtained, and the quality estimated value calculation method is obtained. Is used to estimate the quality of the program source code based on the metric value measured by the metric measuring means.

When obtaining the quality estimation value calculating method for estimating the quality of the program source code from the metrics value, the quality estimating means performs regression to obtain correlation information between the metrics value and the quality of the program source code. It is characterized in that it operates by using an analytical method.

The quality estimation means uses the quality estimation value calculation method obtained from the metric value measured based on the program source code corresponding to each revision edition, to the program source code for which information on quality is clear. The quality is estimated appropriately, and the accuracy of the quality estimation value calculation method is verified by comparing the result with the information of the actual quality of the program source code, and the accuracy of the quality estimation value calculation method is out of the allowable range. In this case, the operation is performed so that the quality estimation value calculation method is calculated again.

The quality estimating means operates so as to use a plurality of quality estimated value calculating methods together as the quality estimated value calculating method for estimating the quality of the program source code. It presents a plurality of quality information for the program source code passed from the quality estimating means, and operates to select and present an appropriate suggestion for the work based on the value of each quality estimation value or the combination of the values. It is characterized by

The quality estimation means is characterized in that each of the program source code or a plurality of revisions of the program source code is divided into some groups and handled.

When obtaining the quality estimation value calculating method, the quality estimating means obtains the quality estimation value calculating method for each group to which the program source code or a plurality of revisions of the program source code belongs. When estimating the quality of the program source code, the quality is estimated by using the quality estimation value calculation method corresponding to the program source code or the group to which the revised edition of the program source code belongs. It operates.

When the plurality of revisions of the program source code are divided into groups and handled, the quality estimating means operates so as to divide the revisions into groups according to the difference in the contents of the program development work that causes the revisions. Is characterized by.

The quality estimating means obtains the first correlation information regarding the metrics value and the quality by collectively handling the program source code, and handles the program source code by grouping for each creator to handle the metrics value and the quality. Regarding the creator by obtaining the second correlation information regarding the creator and comparing the first correlation information with the second correlation information, and the characteristic information regarding the creator is passed to the work guide presenting means. The means is operable to select and present an appropriate suggestion for the work by using the characteristic information about the worker.

The quality estimating means treats the program source code as a point in the quality evaluation space based on the measured metric value, and when estimating the quality of the program source code, the quality based on the geometric operation in the space is estimated. It is characterized in that it operates so as to estimate its quality using an estimated value calculation method.

The metrics measuring means obtains the program source code before the revision and the program source code after the revision from the program version management means, and the measured metrics value corresponds to the version control number of the program source code. Thus, the quality estimation means obtains the metric value stored in the metric measurement information storage means based on the version control number of the program source code, and the metric value is stored in the metric measurement information storage means. And operating to estimate the quality of the program source code.

This program quality control support method is composed of a metric measuring step, a quality estimating step, and a work guide presenting step. The metric measuring step includes a program source code before revision and a program source code after revision. It operates to measure each metric value, and the quality estimation step is the difference (change amount) between the metric value of the program source code before revision and the metric value of the program source code after revision.
Is calculated and the quality of the program source code is estimated based on the amount of change, and the work guide presenting step operates to present the quality of the program source code estimated by the quality estimating step. .

The program quality control support method is as follows:
It consists of a program revision point extraction step, metrics measurement step, quality estimation step, and work guide presentation step. The program revision point extraction step obtains difference information between the program source code before revision and the program source code after revision. The metric measuring step operates to measure a metric value for the difference information, the quality estimating step operates to estimate the quality of the program source code from the metric value for the difference information, The work guide presenting step is characterized in that it operates so as to present the quality of the program source code estimated by the quality estimating step.

The work guide presenting step is characterized in that the work guide presenting step operates so as to select and present an appropriate suggestion for the work based on the quality information for the program source code passed from the quality estimating step.

The quality estimating step operates so as to obtain respective quality estimation values corresponding to a plurality of revisions of the program source code, and the work guide presenting step calculates a combination of the quality estimation values of the respective editions. Act to select and present appropriate suggestions for work, based on
It is characterized by

In the quality estimation step, the program source code corresponding to each revision version is obtained from the program version management step that holds the revision history of the program source code, and the quality measurement step is performed using the metrics measurement step. Obtaining a metric value, generating a quality estimated value calculating step for estimating the quality of the program source code from the metric value based on these values, and using the quality estimated value calculating step, the metric measured by the metric measuring step. It operates to estimate the quality of the program source code based on the value.

In the quality estimation step, when obtaining the quality estimation value calculation step for estimating the quality of the program source code from the metrics value, regression is performed to obtain correlation information between the metrics value and the quality of the program source code. It is characterized in that it operates by using an analytical method.

The quality estimation step uses the quality estimation value calculation step obtained from the metric value measured based on the program source code corresponding to each revision version, to the program source code whose quality information is clear. The quality is appropriately estimated, and the accuracy of the quality estimation value calculation step is verified by comparing the result with information on the actual quality of the program source code, and the accuracy of the quality estimation value calculation step is outside the allowable range. In this case, the operation is performed so that the quality estimated value calculating step is calculated again.

The quality estimation step operates so that a plurality of quality estimation value calculation steps are used in combination as the quality estimation value calculation step for estimating the quality of the program source code. It operates to present a plurality of pieces of quality information for the program source code passed from the quality estimation step, and to select and present an appropriate suggestion for the work based on the value of each quality estimation value or the combination of the values. It is characterized by

The quality estimation step is characterized in that each of the program source code or a plurality of revisions of the program source code is divided into some groups and handled.

In the quality estimating step, when obtaining the quality estimated value calculating step, the quality estimated value calculating step is obtained for each group to which the program source code or a plurality of revisions of the program source code belongs. When estimating the quality of the program source code, the quality estimation value calculating step corresponding to the program source code or the group to which the revised edition of the program source code belongs is used to estimate the quality. It operates.

In the quality estimating step, when each of a plurality of revisions of the program source code is divided into groups and handled, it is divided into groups according to the content of the program development work that causes the revisions. Is characterized by.

In the quality estimating step, the program source codes are collectively processed to obtain the first correlation information regarding the metrics value and the quality, and the program source codes are grouped for each creator to be handled and the metrics values and the quality are evaluated. Regarding the creator by obtaining the second correlation information about the creator, and comparing the first correlation information and the second correlation information to obtain characteristic information about the creator, and pass it to the work guide presenting step. The step is characterized in that the characteristic information regarding the worker is used to operate to select and present an appropriate suggestion for the work.

In the quality estimation step, the program source code is treated as a point in the quality evaluation space based on the measured metric value, and when estimating the quality of the program source code, the quality based on the geometric operation in the space is used. It is characterized in that it operates to estimate its quality using an estimated value calculation step.

In the metrics measuring step, the program source code before revision and the program source code after revision are obtained from the program version management step, and the measured metrics value corresponds to the version management number of the program source code. Then, it operates so as to be stored by the metric measurement information storage step, and the quality estimation step obtains the metric value stored by the metric measurement information storage step based on the version control number of the program source code, and calculates the metric value. And operating to estimate the quality of the program source code.

[0167]

According to the present invention, since the quality of the program source code is estimated based on the known objective information by using the difference of the metric values, the metric value dependent on the characteristics of the program source code is used. It is possible to obtain stable software quality evaluation that is not easily affected by fluctuations and does not depend on the experience and intuition of the program developer. As a result, software quality can be evaluated easily, quickly and stably.
The efficiency of implementing the software quality improvement measures can be improved, and higher quality software can be obtained with less labor.

Further, according to the present invention, the metric value corresponding to each version of the program source code is measured in advance, and the measured metric value is accumulated and stored. At most once for each version of the code. Therefore, the processing amount can be significantly reduced, which can reduce the processing time required for software quality evaluation and improve work efficiency.

Further, according to the present invention, since the difference between the program source codes is extracted and the metric value with respect to the difference is measured, the contents deleted by the revision of the program source code and the contents added by the revision are respectively Metric values can be measured against. Therefore, it is possible to operate so as to obtain the metric value even for the information that disappears from the program source code due to the revision, and to more reliably capture the change in quality due to the content of revision, and to evaluate the quality of software. The accuracy can be improved.

Further, according to the present invention, since the guide for the improvement work of the program source code to be examined is presented to the program developer, the program developer can study the contents of the countermeasure based on the guide, and The implementation efficiency can be improved.

Further, according to the present invention, since a guide for improvement work is presented to the program developer based on a plurality of quality estimation values, a more accurate guide can be presented.

Further, according to the present invention, when presenting a guide for improvement work, an estimated value relating to a defect that occurs in the program source code is also presented, so that the program developer considers the estimated value relating to the defect and details of the countermeasure. Can be considered, and the efficiency of implementing the measures can be improved. Furthermore, the program developer can use the estimated value of the defect as a guide for detecting the defect and ending the remedial work, thereby reducing unnecessary work.
Work efficiency can be improved.

According to the present invention, the most appropriate quality estimation value calculation rule can be set among the plurality of quality estimation value calculation rules, and the quality estimation value can be calculated using the set quality estimation value calculation rule. You can Therefore, it is possible to avoid a situation in which an incorrect quality estimation value is used by continuing to use an inaccurate quality estimation value calculation method, which makes it difficult to improve work efficiency.

Further, according to the present invention, when the numerical values for the parameters included in the quality estimation value calculation rule are determined,
Since the regression coefficient by the regression analysis is used, the weighting showing the influence of the metric value on the quality can be comprehensively evaluated and set, and the accuracy of the quality estimation value can be improved. Furthermore, in this case, by using a regression coefficient with high accuracy, or by using a plurality of regression coefficients, it is possible to more accurately evaluate and set the weighting indicating the influence of the metric value on the quality. It is possible to further improve the accuracy of the estimated value.

Further, according to the present invention, the quality estimation value can be calculated by paying attention to the cause of revision of the program source code, and the dispersion of the difference (fluctuation) in the metric value for each program source code is revised. It can be expected to be smaller than the case where the entire version of the program source code is handled collectively without considering the work content that caused the, and the accuracy of quality estimation can be improved.

Further, according to the present invention, the program source code can be handled by being divided into a plurality of groups,
Since the distribution of metrics values for each program source code in the same group is smaller than when handling the entire program source code at once,
The effect of improving the accuracy of quality estimation is obtained.

Also, according to the present invention, the program source code can be handled by being divided into a plurality of groups,
Moreover, since it is possible to select and present an appropriate guide in consideration of the difference in characteristics of each group, it is possible to enhance the effect of the presented guide and further improve the work efficiency.

Further, according to the present invention, the point corresponding to the program source code and the representative point of the group are obtained in the quality measurement space, and the distance in the quality measurement space is used as the quality estimation value, so that the quality evaluation is performed. You can get more direct guidance. As a result, it is possible to improve the implementation efficiency of the measures by providing the program development worker with more appropriate guidelines as the priority of the measures.

Further, according to the present invention, by using the vector Δ on the quality measurement space indicated by the metric difference value, it is possible to estimate that the work that cannot be measured as the quality variation is performed, and thus it is inappropriate. It is possible to present a guide to reinforce the verification work because there is a possibility that various work is being carried out, or a guide to revise the contents of the work itself, which is suggested to the program developer. The contents of can be enhanced and the implementation efficiency of measures can be improved.

Further, according to the present invention, when a representative point is obtained in the quality measurement space, it is possible to select an appropriate representative point based on the state of revision from the past, rather than just the representative point close at a certain moment. As a result, the accuracy of the obtained quality estimation value is improved, so that more appropriate guidelines can be presented to the program developer, and the efficiency of implementing the measures can be improved.

Further, if the grouping is performed based on the distance in the quality measurement space, the position of the representative point or the quality estimation value to be compared can be intuitively understood by the program developer, and the obtained quality can be obtained. It can be expected that the psychological resistance and repulsion to the estimated value and the guide suggested by it will be reduced.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration example of a program quality management support device according to a first embodiment.

FIG. 2 is a diagram showing an example of a work guide holding table.

FIG. 3 is a diagram showing a configuration example of a program quality management support device according to a second embodiment.

FIG. 4 is a diagram illustrating an example of a process of deriving a regression coefficient.

FIG. 5 is a flowchart showing a procedure of setting processing of a quality estimation value calculation method.

FIG. 6 is a diagram showing an example of a quality measurement space.

FIG. 7 is a diagram showing a configuration example of a program quality management support device according to a third embodiment.

FIG. 8 is a diagram showing a configuration example of a program quality management support device according to a fourth embodiment.

FIG. 9 is a diagram illustrating a first conventional example.

FIG. 10 is a diagram illustrating a second conventional example.

[Explanation of symbols]

1 program quality control support device, 2 program version control means, 3 revised program groups, 4 metrics measurement means, 5 quality estimation means, 6 work guide presenting means, 7
Quality estimation value calculation method holding means, 8 metrics measurement information storage means, 9 program revision location extraction means, 10
Program designation means.

   ─────────────────────────────────────────────────── ─── Continued front page    (54) [Title of Invention] A program quality control support device, a program quality control support method, and a program quality control support method                     A computer-readable recording medium that stores a program to be executed by a computer.                     Program for causing a computer to execute a program for supporting quality control of programs

Claims (32)

[Claims] 1. A program quality management support device for estimating the quality of a program during program development and supporting the quality management of the program, wherein a first version program and at least one or more revised version programs in the program development are associated with each other. As a revised program group, and a program management means for managing a plurality of versions of the program for each revised program group, and a revised version program of any of the revised program groups managed by the program management means. Program designation means for designating as a quality estimation target program, and designating a program of a version earlier than the quality estimation target program and belonging to the same revised program group as the quality estimation target program as a comparison target program, the quality estimation target program, and The ratio Regarding the comparison target program, a metric value measuring means for measuring a metric value for each of at least one or more types of metric measurement items, and the quality of the quality estimation target program based on the measurement result of the metric value by the metric value measuring means. And a quality estimation value presenting means for presenting the quality estimation value of the quality estimation target program calculated by the quality estimation means. A program quality control support device characterized by: 2. The metric value measuring means measures, for each metric measurement item, a metric value of the quality estimation target program and a metric value of the comparison target program, and the quality estimating means determines the metric value. Based on the measurement result of the metric value by the measuring means, the difference between the metric value of the quality estimation target program and the metric value of the comparison target program is determined for each metric measurement item, and the quality is determined based on each determined difference. The program quality management support apparatus according to claim 1, wherein a quality estimation value of the estimation target program is calculated. 3. The program quality management support device further comprises a difference extracting means for extracting a difference in program content between the quality estimation target program and the comparison target program, and the metric value measuring means, For the difference extracted by the difference extracting means, measure a metric value for each of the metric measurement items, the quality estimating means, based on the measurement result by the metric value measuring means, of the quality estimation target program. The program quality management support device according to claim 1, wherein a quality estimation value is calculated. 4. A program quality management support device for estimating the quality of a program during program development and supporting quality management of the program, wherein a first version program and at least one or more revised version programs in the program development are associated with each other. And a program management means for managing a plurality of versions of programs by program revision for each revised program group, and each of the plurality of versions of programs for each revised program group managed by the program management means, Metric value measuring means for measuring a metric value for each of at least one or more kinds of metric measuring items, and metric value measurement result storage for storing a metric value measurement result by the metric value measuring means for each revision program group. And a revised version program of any revised program group managed by the program management means as a quality estimation target program, and belongs to the same revised program group as the quality estimation target program, and the quality estimation target Program specification means for specifying a program of a version earlier than the program as a comparison target program, and the metrics value measurement result storage means, and obtains the measurement result of the metrics value for the quality estimation target program and the comparison target program. A quality estimation unit that estimates the quality of the quality estimation target program based on the measurement result of the acquired metrics value and calculates a quality estimation value of the quality estimation target program; and the quality estimation target calculated by the quality estimation unit. Present a quality estimate for the program Program quality management support apparatus characterized by having a quality estimate presenting means. 5. The metric value measurement means measures metric values for each metric measurement item for each of a plurality of versions of programs for each of the revised program groups, and the metric value measurement result storage means stores the revised value. For each program group, stores the measurement results of the metrics value for each of a plurality of versions of the program, the quality estimation means, from the metrics value measurement result storage means, the measurement result of the metrics value for the quality estimation target program And the measurement result of the metric value for the comparison target program is obtained, and for each metric measurement item, the difference between the metric value of the quality estimation target program and the metric value of the comparison target program is determined, and each determined Metric Program quality control assistance device according to claim 4, characterized in that for calculating the quality estimate of the quality estimation target program based on a difference of the scan values. 6. The program quality control support device further extracts a difference in program content between each program managed by the program management means and each of the other versions of the same revised program group. The metric value measuring means, with respect to the difference from each of the other versions of programs of the same revised program group extracted for each program by the difference extracting means, to each of the metric measurement items. The metric value measurement result storage means stores the metric value measurement result for each difference for each of a plurality of edition programs for each of the revised program groups, and the quality estimation Means, from the metrics value measurement result storage means, It is characterized in that a measurement result of a metric value for a difference between the quality estimation target program and the comparison target program is acquired, and a quality estimation value of the quality estimation target program is calculated based on the acquired measurement result of the metric value. The program quality control support device according to claim 4. 7. The quality estimation value presenting means presents a guide for a quality improvement work of the quality estimation target program based on the quality estimation value of the quality estimation target program. Program quality control support device. 8. The program designating unit designates a plurality of comparison target programs, and the quality estimating unit calculates a quality estimation value of the quality estimation target program corresponding to each of the plurality of comparison target programs, 8. The program quality management according to claim 7, wherein the quality estimation value presenting means presents a guide for a quality improvement work of the quality estimation target program based on each quality estimation value of the quality estimation target program. Support device. 9. The program designating unit designates, as the quality estimation target program, a latest version of a program of any revised program group, and the quality estimation value is a quality estimation value of the latest version of the program. The program quality control support device according to claim 1 or 4, wherein 10. The program specifying means specifies, as the quality estimation target program, a program other than the latest version of the program that belongs to the same revised program group as the latest version of the program, and the quality estimation The means calculates a quality estimation value of the other program, the quality estimation value presenting means compares the quality estimation value of the latest version of the program with the quality estimation value of the other program, 10. The program quality control support device according to claim 9, wherein a guide for quality improvement work of the latest version of the program is presented based on the program. 11. The quality estimating means at least an estimated value relating to a defect that occurs in the latest version of the program when the latest version of the program is designated as the quality estimation target program by the program designating means. 10. The program quality management support device according to claim 9, wherein the quality estimation value presenting means calculates one or more and presents each of the estimation values regarding the failure calculated by the quality estimation value. 12. The program quality control support device,
Furthermore, the quality estimation means includes a quality estimation value calculation rule storage means for storing at least one quality estimation value calculation rule indicating a procedure for calculating the quality estimation value by the quality estimation means, and the quality estimation means is configured to calculate the quality estimation value. 5. The program quality management support apparatus according to claim 1, wherein the quality estimation value of the quality estimation target program is calculated according to one of the quality estimation value calculation rules stored in the rule storage means. 13. The quality estimation value calculation rule storage means stores a parameter quality estimation value calculation rule including a parameter corresponding to each metric measurement item, and the quality estimation means stores the parameter quality estimation value calculation rule in the metrics value measurement means. On the other hand, instructing to measure the metric value for each metric measurement item for each of the plurality of versions of the program of the specific revision program group, the metric value measuring means, based on the instruction from the quality estimating means, A metric value for each metric measurement item is measured for each of a plurality of versions of programs of a specific revised program group, and the quality estimating means calculates the parameter quality estimated value calculation rule based on the measurement result by the metric value measuring means. For each parameter of 13. The program quality control support device according to claim 12, wherein a numerical value for the program is determined. 14. The quality estimating means performs regression analysis using the measurement result by the metric value measuring means,
14. The program quality control support according to claim 13, wherein a regression coefficient is calculated for each metric measurement item, and a numerical value for each parameter of the parameter quality estimation value calculation rule is determined based on each calculated regression coefficient. apparatus. 15. The quality estimating means calculates a quality estimation value of one of the programs at a predetermined timing, analyzes the estimation accuracy of the calculated quality estimation value, and the estimation accuracy is within a predetermined allowable range. 14. The program quality management support device according to claim 13, wherein when the number of parameters exceeds the numerical value, a numerical value for each parameter of the parameter quality estimation value calculation rule is set. 16. The quality estimation value calculation rule storage means stores a plurality of quality estimation value calculation rules, and the quality estimation means stores at least two or more quality estimation values among the plurality of quality estimation value calculation rules. The calculation rule is selected as the selection quality estimation value calculation rule, the quality estimation value corresponding to each of the selection quality estimation value calculation rules is calculated for one of the programs, and the estimation accuracy is analyzed for each calculated quality estimation value. , Specifying a specific quality estimation value calculation rule from the selected quality estimation value calculation rules as a designated quality estimation value calculation rule, and according to the designated quality estimation value calculation rule, the quality estimation value of the quality estimation target program 13. The program quality management support device according to claim 12, wherein 17. Each of the plurality of quality estimation value calculation rules includes a parameter corresponding to each metric measurement item, and the quality estimation means prior to selecting the selected quality estimation value calculation rule. For measuring means,
Instructing each of a plurality of versions of programs of a specific revision program group to measure a metric value for each metric measurement item, the metric value measuring means, based on the instruction from the quality estimating means, The metrics value for each metric measurement item is measured for each of the programs of the plurality of versions of the program group, and the quality estimating unit determines and determines the numerical value for each parameter based on the measurement result by the metric value measuring unit. 17. The program quality management support device according to claim 16, wherein the quality estimation value corresponding to each of the selected quality estimation value calculation rules is calculated using the numerical values for the respective parameters. 18. The quality estimating means performs a plurality of types of regression analysis using the measurement results by the metric value measuring means, calculates a plurality of regression coefficients for each metric measurement item, and calculates each regression coefficient. Perform an accuracy analysis, select a specific regression coefficient from multiple regression coefficients for each metric measurement item based on the analysis result, and use the selected regression coefficient for each metric measurement item. 18. The program quality control support device according to claim 17, wherein a numerical value for the parameter is determined. 19. The program management means manages revision cause information indicating a program revision cause of a revised version program for each revision program group, and the program designating means, based on the revision cause information, the quality estimation target. A revised version of the program having the same revision cause as the program is designated as the comparison target program, and the quality estimation value calculation rule storage unit classifies the quality estimation value calculation rules for each program revision cause to classify a plurality of quality estimation value calculation rules. 13. The quality estimation unit calculates a quality estimation value of the quality estimation target program according to a quality estimation value calculation rule corresponding to a program revision cause of the quality estimation target program. Program quality control support device. 20. The program management means manages a plurality of revised program groups, and the quality estimation means classifies programs of a plurality of versions of the plurality of revised program groups into a plurality of groups, respectively, and estimates the quality. The quality estimation value corresponding to each of the selection quality estimation value calculation rules is calculated for any of the programs of the same group that are classified into the same group as the target program, and the estimation accuracy is analyzed for each of the calculated quality estimation values and analyzed. The program quality management support apparatus according to claim 16, wherein the designated quality estimation value calculation rule is designated from the selection quality estimation value calculation rules based on the result. 21. The quality estimating means specifies a specific quality estimation value calculation rule for any of the programs without classifying a plurality of versions of programs of the plurality of revised program groups into a plurality of groups, respectively. With the first quality estimation value calculation rule, the programs of the plurality of editions of the plurality of revised program groups are each classified into a plurality of groups, and any of the group programs classified into the same group as the quality estimation target program. Whether a specific quality estimation value calculation rule is designated as a second quality estimation value calculation rule, the first quality estimation value calculation rule and the second quality estimation rule are compared, and the quality estimation value is presented. The means provides a guide for the quality improvement work of the quality estimation target program based on the comparison result by the quality estimation means. Program quality control assistance device according to claim 20, characterized in that the. 22. A program quality management support device for estimating the quality of a program in program development and supporting quality management of the program, the program management means managing a plurality of programs, and the program management means managing the program quality management means. A program designating unit that designates a specific program from among the plurality of programs as a quality estimation target program, and designates at least two or more programs from among programs other than the quality estimation target program as comparison target programs, For each of the quality estimation target program and the comparison target program, metric value measuring means for measuring a metric value for each of at least one or more metric measurement items, and coordinates corresponding to each metric measurement item. Having quality evaluation coordinates composed of, the coordinate position in the quality evaluation coordinates is specified for each of the quality estimation target program and the comparison target program based on the measurement result by the metrics value measuring means, and each specified A quality estimating unit that performs a geometric operation based on a coordinate position, estimates the quality of the quality estimation target program based on the result of the geometric operation, and calculates a quality estimation value of the quality estimation target program; And a quality estimated value presenting means for presenting the quality estimated value of the quality estimation target program calculated by the above. 23. The quality estimating means classifies each of the coordinate positions of the comparison target program into at least one coordinate position group based on the distribution status of the coordinate positions of the comparison target program, and each coordinate position group. About a specific coordinate position in each coordinate position group is determined as a comparison target coordinate position, geometric calculation is performed based on the coordinate position of the quality estimation target program and each comparison target coordinate position, and the geometric calculation 23. The program quality management support device according to claim 22, wherein a quality estimation value of the quality estimation target program is calculated based on the result. 24. A program quality control support method for estimating program quality in program development and supporting program quality control, wherein a first version program and at least one or more revised version programs in program development are associated with each other. As a revised program group, and a program management step for managing a plurality of versions of the program for each revised program group, and a revised version program of any of the revised program groups managed by the program management step. A program specifying step of specifying as a quality estimation target program, a program belonging to the same revised program group as the quality estimation target program and having a version earlier than the quality estimation target program as a comparison target program; and the quality estimation target program Ram and the program to be compared, the quality estimation target based on a metric value measurement step of measuring a metric value for each of at least one or more metric measurement items, and a metric value measurement result by the metric value measurement step. A quality estimation step of estimating the quality of the program and calculating a quality estimation value of the quality estimation target program; and a quality estimation value presentation step of presenting the quality estimation value of the quality estimation target program calculated by the quality estimation step. A method for supporting program quality control, comprising: 25. The metric value measuring step comprises:
For each metric measurement item, measure the metric value of the quality estimation target program and the metric value of the comparison target program, the quality estimation step, based on the measurement result of the metric value by the metric value measurement step, respectively, For the metric measurement item, the difference between the metric value of the quality estimation target program and the metric value of the comparison target program is determined, and the quality estimation value of the quality estimation target program is calculated based on each determined difference. 25. The program quality control support method according to claim 24. 26. The program quality control support method,
Further, there is a difference extracting step for extracting a difference in program content between the quality estimation target program and the comparison target program, the metrics value measuring step, for the difference extracted by the difference extracting step, A metric value is measured for each metric measurement item, and the quality estimation step calculates a quality estimation value of the quality estimation target program based on a measurement result by the metric value measurement step. Two
4. The program quality control support method described in 4. 27. A program quality control support method for estimating program quality in program development and supporting program quality control, wherein a first version program and at least one or more revised version programs in program development are associated with each other. As a revised program group, a program management step of managing a plurality of versions of the program by the program revision for each revised program group, and each of the plurality of versions of the program for each revised program group managed by the program management step, A metric value measuring step of measuring a metric value for each of at least one or more types of metric measuring items, and a metric for storing a metric value measurement result by the metric value measuring step for each revised program group. The measurement value measurement result storing step, and a revised version program of any revised program group managed by the program management step is designated as a quality estimation target program, and is set to the same revised program group as the quality estimation target program. A program specifying step for specifying a program of a version earlier than the quality estimation target program as a comparison target program, and a metric value measurement for the quality estimation target program and the comparison target program from the metric value measurement result storage step. And a quality estimation step of estimating the quality of the quality estimation target program based on the measurement result of the obtained metrics value and calculating the quality estimation value of the quality estimation target program, and the quality estimation step. Item Program quality management support method characterized by having a quality estimate presenting step of presenting a quality estimate of the estimated target program. 28. The metric value measuring step comprises:
For each of a plurality of versions of the program for each revised program group, measure a metric value for each metric measurement item, the metric value measurement result storing step, for each revised program group, each of the plurality of versions of the program. The measurement result of the metric value for the quality estimation step, the quality estimation step, from the metric value measurement result storage step, the measurement result of the metric value for the quality estimation target program and the measurement result of the metric value for the comparison target program. Is obtained, the difference between the metric value of the quality estimation target program and the metric value of the comparison target program is determined for each metric measurement item, and the quality estimation is performed based on the determined difference in each metric value. Program quality control support method according to claim 27, wherein calculating the quality estimate of the target program. 29. The program quality control support method,
Further, the method includes a difference extracting step for extracting a difference in program content between each program managed by the program managing step and each program of another version of the same revised program group, and the metric value measuring step. Measures a metric value for each of the metric measurement items for a difference from each of the other versions of the same revised program group extracted for each program by the difference extraction step, and obtains the metric value measurement result. The storing step stores the measurement result of the metric value for each difference for each of the plurality of versions of the program for each of the revised program groups, and the quality estimating step includes the metric value measurement result storing step, Quality estimation target 28. The quality estimation value of the quality estimation target program is calculated based on the measurement result of the metric value of the difference between the gram and the comparison target program, and the quality estimation value of the quality estimation target program is calculated based on the acquired measurement result of the metric value. Described program quality control support method. 30. A program quality management support method for estimating program quality in program development and supporting program quality management, comprising: a program management step of managing a plurality of programs; and a program management step managed by the program management step. A program designating step of designating a specific program from among the plurality of programs as a quality estimation target program, and designating at least two or more programs as comparison target programs from among programs other than the quality estimation target program, For each of the quality estimation target program and the comparison target program, a metric value measurement step of measuring a metric value for each of at least one or more types of metric measurement items, and each metric measurement item. Having a quality evaluation coordinate composed of coordinate axes corresponding to, specifying the coordinate position in the quality evaluation coordinates for each of the quality estimation target program and the comparison target program based on the measurement result by the metrics value measuring step, Performing a geometric operation based on each of the identified coordinate positions, estimating the quality of the quality estimation target program based on the result of the geometric operation, a quality estimation step of calculating a quality estimation value of the quality estimation target program, And a quality estimated value presenting step of presenting a quality estimated value of the quality estimation target program calculated in the quality estimating step. 31. A computer-readable recording medium having recorded therein a program for causing a computer to execute a program quality control support method for estimating a program quality in program development and supporting the program quality control. The program of the first edition and the program of at least one or more revisions are associated with each other as a revision program group, and a program management step of managing a plurality of versions of the program by the program revision for each revision program group, and management by the program management step A revised version of any of the revised program groups specified as a quality estimation target program is specified, and a program of a version earlier than the quality estimation target program belonging to the same revised program group as the quality estimation target program is specified. Program specifying step for specifying a program as a comparison target program, the quality estimation target program and the comparison target program, a metric value measuring step for measuring a metric value for each of at least one or more metric measurement items, Based on the measurement result of the metric value by the metric value measurement step, the quality estimation step of estimating the quality of the quality estimation target program and calculating the quality estimation value of the quality estimation target program; and the quality estimation step calculated by the quality estimation step. A computer-readable recording medium recording a program for causing a computer to execute a program quality control support method, which comprises a quality estimation value presentation step of presenting a quality estimation value of a quality estimation target program. 32. A program for causing a computer to execute a program quality control support method for estimating the quality of a program in the program development and supporting the quality control of the program, and at least one or more of the first version program in the program development. The program management step of managing a plurality of versions of the program by the program revision for each revision program group by associating with the program of the revised version of A program that designates one of the revised versions of the program as a quality estimation target program, and designates a program of a version earlier than the quality estimation target program as a comparison target program that belongs to the same revised program group as the quality estimation target program. Specifying step, with respect to the quality estimation target program and the comparison target program, a metric value measuring step for measuring a metric value for each of at least one or more metric measuring items, and a metric value by the metric value measuring step. A quality estimation step of estimating the quality of the quality estimation target program based on the measurement result and calculating a quality estimation value of the quality estimation target program, and a quality estimation value of the quality estimation target program calculated by the quality estimation step. A program for causing a computer to execute a program quality control support method, the method including: