Analysis device, analysis method, and analysis program

The analysis device predicts software function execution frequencies using spatial statistical models, addressing the inefficiencies of conventional methods by generating a spatial proximity matrix, thus efficiently assessing software usage and optimizing resource allocation.

WO2024231979A1PCT designated stage expired Publication Date: 2024-11-14NIPPON TELEGRAPH & TELEPHONE CORP
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
PCT/JP2023/017279
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-08
Publication Date
2024-11-14

AI Technical Summary

Technical Problem

Conventional methods for assessing software bloat, such as Splunk, face challenges in efficiently determining the usage status of software functions, leading to high costs and limited accuracy due to the need for extensive data observation over long periods.

Method used

An analysis device and method that generates information on the relationship between software functions and predicts their execution frequency using a spatial proximity matrix and spatial statistical models, allowing for efficient assessment of function usage without requiring comprehensive data collection.

Benefits of technology

Enables accurate and cost-effective prediction of software function execution frequencies, providing insights for optimizing software by identifying functions that can be eliminated, thereby reducing resource usage and improving system performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure JP2023017279_14112024_PF_FP_ABST
    Figure JP2023017279_14112024_PF_FP_ABST
Patent Text Reader

Abstract

An analysis device (10) comprises a generation unit (11) and a prediction unit (13). The generation unit (11) generates information indicating a relationship among a plurality of functions included in software. The prediction unit (13) predicts a frequency at which each of the plurality of functions is to be executed on the basis of an observation value of a frequency at which each of the plurality of functions has been executed and the information generated by the generation unit (11).
Need to check novelty before this filing date? Find Prior Art

Description

Analysis device, analysis method, and analysis program

[0001] The present invention relates to an analysis device, an analysis method, and an analysis program.

[0002] One of the problems surrounding software in recent years is software bloat. Software bloat is a common name for software that occupies a large amount of computer resources while not being utilized. For example, software is a program installed on a computer. Also, the computer on which software runs is sometimes called a system.

[0003] In the past, the idea was "don't touch a running program." Now, the idea is more likely to be "getting rid of unused software is a better investment than maintaining junk, or obsolete, systems."

[0004] The scope of what needs to be eliminated in a system can vary (granularity), from individual functions (function calls) in the software to the entire software stack. For example, when eliminating a function, software architects and programmers must identify the functions within the software that need to be eliminated (or, in a broader sense, the functions that need to be modified, including deletion). Therefore, in systems that are already in operation, it is important to understand the usage status of functions within the software in order to objectively determine what to eliminate.

[0005] Conventionally, in order to improve the observability of a system, a tool called Splunk has been provided that implements technology for "acquiring, aggregating, and utilizing logs, traces, and metrics" (see, for example, Non-Patent Document 1).

[0006] Splunk assumes that a mechanism for collecting data from the system being monitored and sending the collected data to a server is built in. Splunk is also intended to monitor the usage status of services or APIs (Application Programming Interfaces) provided by the system.

[0007] Splunk, Integrated Security and Observability Platform, [online], [Retrieved April 19, 2023], Internet (https: / / www.splunk.com / ja_jp?301= / ja_jp.html)

[0008] However, the conventional techniques have the problem that it is difficult to efficiently grasp the status of functions included in software.

[0009] For example, in order to understand the status of functions included in software, it is conceivable to observe the execution frequency of all functions over the entire period during which the software is in operation using the technology described in Non-Patent Document 1. However, such observation over the entire period requires a great deal of cost.

[0010] To reduce costs, it is possible to observe the execution frequency of only a certain period or a certain function. However, such limited observation results may make it difficult to accurately grasp the software status.

[0011] In order to solve the above-mentioned problems and achieve the objectives, the analysis device is characterized by having a generation unit that generates information indicating the relationships between multiple functions included in software, and a prediction unit that predicts the frequency with which each of the multiple functions will be executed based on observed values ​​of the frequency with which each of the multiple functions is executed and the information generated by the generation unit.

[0012] According to the present invention, the status of functions included in software can be grasped efficiently.

[0013] FIG. 1 is a diagram illustrating an example of the configuration of an analysis apparatus according to a first embodiment. FIG. 2 is a flowchart illustrating the processing flow of a generation unit. FIG. 3 is a diagram illustrating the relationship between functions. FIG. 4 is a diagram illustrating an example of a spatial proximity matrix when k = 1. FIG. 5 is a diagram illustrating an example of a spatial proximity matrix when k = 2. FIG. 6 is a diagram illustrating examples of explanatory variables. FIG. 7 is a flowchart illustrating the processing flow of a prediction unit. FIG. 8 is a diagram illustrating an example of an execution frequency prediction result. FIG. 9 is a diagram illustrating an example of output of an execution frequency prediction result. FIG. 10 is a flowchart illustrating the processing flow of a spatial distribution analysis unit. FIG. 11 is a diagram illustrating examples of deviation and spatial lag variables. FIG. 12 is a diagram illustrating an example of output of a spatial distribution analysis result. FIG. 13 is a diagram illustrating an example of a computer that executes a program.

[0014] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below with reference to the accompanying drawings, in which: FIG. 1 is a block diagram showing an analysis apparatus, an analysis method, and an analysis program according to an embodiment of the present invention;

[0015] [Configuration of First Embodiment] The configuration of an analysis device according to a first embodiment will be described with reference to Fig. 1. Fig. 1 is a diagram showing an example of the configuration of an analysis device according to the first embodiment.

[0016] The analysis device 10 receives input of software source code and software execution logs. For example, the source code is a text file. Furthermore, for example, the execution log is a file that records the execution date and time and the number of executions of each function included in the software. The analysis device 10 may receive input of the execution log for the entire period during which the software was running, or may receive input of the execution log for a partial period. Furthermore, the execution log may include the execution logs of all functions, or may not include some functions (functions may contain missing data).

[0017] The analysis device 10 predicts the execution frequency of each of a plurality of functions included in the software based on the input data, and outputs the execution frequency prediction results. The analysis device 10 also analyzes the spatial characteristics of the functions and outputs the spatial distribution analysis results.

[0018] The analysis device 10 includes a generation unit 11, a preprocessing unit 12, a spatial distribution analysis unit 14, and a prediction unit 13. The analysis device 10 also obtains a spatial proximity matrix, explanatory variables, and observed values ​​as intermediate data during processing.

[0019] The generator 11 generates information indicating the relationships between multiple functions included in the software. The spatial proximity matrix is ​​an example of information indicating the relationships between multiple functions. The generator 11 also acquires attribute information of each function as an explanatory variable.

[0020] The preprocessing unit 12 acquires the observed value of the execution frequency of each function from the execution log. For example, if the execution log includes the execution date and time of each function, the preprocessing unit 12 counts the number of times each function is executed based on the execution date and time. The preprocessing unit 12 may also acquire the observed value of the execution frequency for a specified period.

[0021] The prediction unit 13 predicts the frequency at which each of the plurality of functions will be executed, based on the observed value of the frequency at which each of the plurality of functions is executed and the information generated by the generation unit 11 .

[0022] The spatial distribution analysis unit 14 analyzes the spatial characteristics of each of the plurality of functions in a space having each of the plurality of functions as an element.

[0023] The processing of the analysis device 10 will be described in detail below using a flowchart.

[0024] 2 is a flowchart showing the flow of processing by the generation unit 11. As shown in Fig. 2, the generation unit 11 analyzes the call relationships of the input source code and calculates the distance of the shortest path between functions (step S101).

[0025] Here, as shown in Figure 3, there may be a call relationship between functions. Figure 3 is a diagram explaining the relationship between functions. If a second function is called when a first function is executed, it is assumed that there is a call relationship between the first function and the second function. For example, Figure 3 shows that there is a call relationship between function 1 and function 2, but there is no call relationship between function 2 and function 3.

[0026] The generation unit 11 calculates the distance of the shortest path between functions using equations (1-1), (1-2), and (1-3) based on the calling relationships.

[0027]

[0028] Here, f is a function, and i and j are identifiers that distinguish functions. As shown in equation (1-3), the maximum value of i and j, i.e., the number of functions, is N.

[0029] For example, the generation unit 11 regards the relationships between functions shown in FIG. 3 as a graph in which functions are nodes and call relationships are edges, and calculates the distance of the shortest path between nodes in the graph using equation (1-1). In this case, the distance between adjacent nodes is set to 1. The generation unit 11 can also calculate the distance using an existing algorithm (for example, Dijkstra's algorithm). Reference 1 (Dijkstra's algorithm): E. W. Dijkstra: A Note on Two Problems in Connexion with Graphs. Numerische Mathematik, 1:269.271. 1959.

[0030] As shown in equation (1-2), the function f i and function f j The distance of the shortest path between j and function f i is equal to the distance of the shortest path between

[0031] In the following description, functions 1, 2, 3, 4, and 5 in FIG. 3 are respectively functions f 1 , function f 2 , function f 3 , function f 4 , and the function f 5 As shown in FIG. 3, function 1 and function 2 are in a call relationship and are adjacent nodes, so d(f 1 , f 2 ) = 1. Also, function 1 can be reached from function 4 via function 2, so d(f 1 , f 4 ) = 2.

[0032] For example, the distance of the shortest path from function 1 to each function is as follows: d(f 1 , f 1 ) = 0 d(f 1 , f 2 ) = 1 d(f 1 , f 3 ) = 1 d(f 1 , f 4 ) = 2 d(f 1 , f 5 ) = 2

[0033] Next, the generator 11 specifies the degree of proximity k (step S102). The generator 11 also calculates the value of each cell of the spatial proximity matrix as shown in equation (2) (step S103). The generator 11 then outputs the calculated spatial proximity matrix (step S104). In this way, the generator 11 generates a spatial proximity matrix based on the call relationships between multiple functions as information indicating the relationships between multiple functions.

[0034]

[0035] Moreover, an example of the spatial proximity matrix C is shown in equation (3).

[0036]

[0037] As shown in equation (3), if the distance of the shortest path between the first function and the second function is equal to or less than the degree of proximity k and is not 0, the generation unit 11 sets the value of the cell to 1. Otherwise, the generation unit 11 sets the value of the cell to 0.

[0038] 4 is a diagram showing an example of a spatial proximity matrix when k=1. The spatial proximity matrix may be represented as table-format data as shown in FIG. 4. Since the distance between the shortest paths of Function 1 and Function 2 is 1, cell c 12 The value of is 1. On the other hand, the distance between the shortest path of function 1 and function 4 is 2, so the value of cell c 12 The value of is 0. However, since the spatial proximity matrix is ​​a symmetric matrix, c ij = c ji is.

[0039] For example, the values ​​of the cells representing the proximity relationship between Function 1 and each function are as follows: 11 = 0 ← d(f1 , f 1 ) = 0c 12 = 1 ← 1 ≦ d(f 1 , f 2 ) ≦ k c 13 = 1 ← 1 ≦ d(f 1 , f 3 ) ≦ k c 14 = 0 ← k < (f 1 , f 4 ) c 15 = 0 ← k < (f 1 , f 5 )

[0040] In this embodiment, k=1, but k may be a value other than 1. For reference, a spatial proximity matrix when k=2 is shown in FIG. 5. FIG. 5 is a diagram showing an example of a spatial proximity matrix when k=2. When k=1, adjacent functions are considered to be adjacent functions. When k=2, in addition to adjacent functions, functions adjacent to adjacent functions are also considered to be adjacent functions.

[0041] The generator 11 also acquires attribute values ​​of each function from the source code as explanatory variables (step S105). The attribute values ​​include, for example, the line of code (LoC) of the function and the number of methods in the function.

[0042] 6 is a diagram showing an example of explanatory variables. Furthermore, explanatory variable 1 (LoC) and explanatory variable 2 (number of methods) are treated as column vectors as in equations (4) and (5), respectively.

[0043]

[0044]

[0045] 7 is a flowchart showing the flow of processing by the prediction unit 13. As shown in Fig. 7, the prediction unit 13 creates a spatial statistical model based on the spatial proximity matrix and explanatory variables (step S201).

[0046] The observed value Y obtained by the preprocessing unit 12 is expressed as a column vector as shown in equation (6). For example, equation (6) indicates that the execution frequency (number of executions) of function 1 was 100 times.

[0047]

[0048] The prediction unit 13 predicts the frequency at which each of the multiple functions is executed using a spatial statistical model that represents the frequency at which each of the multiple functions is executed, and that includes a term obtained by multiplying a spatial proximity matrix by an observation value and a term obtained by multiplying a regression coefficient by the attribute value of each of the multiple functions.

[0049] The spatial statistical model is shown in equation (7). The regression coefficient β 1 , regression coefficient β 2 , the error term ε is a parameter determined by spatial statistical analysis using a spatial statistical model. 1 and x 2 are vectors of explanatory variables shown in equations (4) and (5), respectively. However, the explanatory variable terms (β 1 x 1 and β 2 x 2 ) may be omitted.

[0050]

[0051] As shown in equation (8), W is a spatial weighting matrix obtained by normalizing the rows of the spatial proximity matrix C.

[0052]

[0053] In this embodiment, the ICAR model is used as the spatial statistical model. However, the spatial statistical model may also be BYM, Lenoux, etc. Reference 2: C. Brunsdon, L. Comber, Hiroaki Yutani (translator), Kazunao Kudo (translator), Taisuke Ichikawa (translator) (2018) Introduction to Geospatial Data Analysis with R. Kyoritsu Shuppan. Reference 3: R package "CARBayes"

[0054] The prediction unit 13 estimates parameters of the spatial statistical model from the observed values ​​and the values ​​of the explanatory variables (step S202). That is, the prediction unit 13 substitutes the observed values ​​and the values ​​of the explanatory variables into the spatial statistical model to obtain the parameters, as shown in equation (9).

[0055]

[0056] The error term ε is expressed as a column vector as in equation (9). In addition, the parameter (regression coefficient β 1 , regression coefficient β 2 ) is calculated as a probability distribution for which a median can be defined. That is, the prediction unit 13 calculates the probability distribution of the frequency at which each of the multiple functions is executed. However, the parameter may be calculated as a single value (for example, an average value) instead of a probability distribution.

[0057] The prediction unit 13 calculates the execution frequency based on the estimated parameters (step S203) and outputs the result (step S204).

[0058] The prediction unit 13 predicts the execution frequency by obtaining the probability distribution of the regression coefficients by Bayesian estimation from the spatial statistical model constructed by equation (9). In the example of equation (10), the median of the execution frequency is calculated from the median value of the regression coefficients. The regression coefficient β of the spatial statistical model 1 , regression coefficient β 2 The median (β 1 Med , β 2 Med ) is inserted. The last term is the weighted average of the execution frequency (median) of nearby functions. This allows predictions to be made that take into account not only the influence from explanatory variables but also the influence of nearby functions (spatial influence).

[0059]

[0060] As shown in equation (11), the prediction unit 13 predicts the execution frequency of each function based on the values ​​of explanatory variables and observed values ​​in the constructed spatial statistical model. The right-hand side of equation (11) is the median of the predicted values ​​of the execution frequency.

[0061]

[0062] 8 is a diagram showing an example of the execution frequency prediction result. In this way, the prediction unit 13 outputs the execution frequency using a probability distribution as a result of Bayesian estimation that takes uncertainty into account. Therefore, as shown in FIG. 8, the prediction unit 13 can calculate not only the median but also the value of the 95% credible interval.

[0063] Furthermore, the prediction unit 13 can visualize the prediction result and output it to a screen or the like as shown in Fig. 9. Fig. 9 is a diagram showing an example of output of the execution frequency prediction result.

[0064] 9, the prediction unit 13 displays the predicted value (median) of the execution frequency of each function as a bar graph, and also displays the 95% credible interval as an error bar.

[0065] 10 is a flowchart showing the processing flow of the spatial distribution analysis unit 14. As shown in FIG. 10, the spatial distribution analysis unit 14 creates a spatial weighting matrix from the spatial proximity matrix (step S301). The spatial distribution analysis unit 14 may acquire the spatial weighting matrix created by the prediction unit 13.

[0066] Next, the spatial distribution analysis unit 14 calculates the overall average value from the median of the predicted values ​​of each function, as in equation (12) (step S302).

[0067]

[0068] Specifically, the spatial distribution analysis unit 14 calculates the average value of the medians shown in FIG. 8 and the like, as shown in equation (13).

[0069]

[0070] Next, the spatial distribution analysis unit 14 calculates the deviation of each function as in equation (14) (step S303). The deviation is the difference between the average value and each predicted value, and is an index of the deviation from the average value.

[0071]

[0072] Specifically, the spatial distribution analysis unit 14 calculates the deviation as shown in equation (15) from the median value shown in FIG. 8 and the average value calculated by equation (13).

[0073]

[0074] Next, the spatial distribution analysis unit 14 calculates the spatial lag variable as shown in equation (16) (step S304). The spatial lag variable is a weighted average of the deviations of the proximity set and is an index of the deviation from the average value in the proximity set. A proximity set is a set of functions that are close to each other. Functions whose cell value in the spatial proximity matrix is ​​1 are included in the same proximity set.

[0075]

[0076] Specifically, the spatial distribution analysis unit 14 calculates the spatial lag variable as shown in equation (17) from the spatial weighting matrix and the deviation obtained by equation (15).

[0077]

[0078] Fig. 11 is a diagram showing examples of deviations and spatial lag variables. The spatial distribution analysis unit 14 visualizes the deviations and spatial lag variables shown in Fig. 11 and outputs them as spatial distribution analysis results on a screen or the like (step S305). Fig. 12 is a diagram showing an example of output of spatial distribution analysis results.

[0079] As shown in FIG. 12, the spatial distribution analysis unit 14 plots the coordinates of each function on a coordinate plane with the deviation on the horizontal axis and the spatial lag variable on the vertical axis.

[0080] The region (first quadrant) where the deviation is large (greater than or equal to 0) and the spatial lag variable is large (greater than or equal to 0) is a hotspot, which, along with its surrounding functions (neighboring functions), is frequently executed. Functions plotted in the hotspot are unlikely to be candidates for elimination.

[0081] On the other hand, the area (third quadrant) where the deviation is small (less than 0) and the spatial lag variable is small (less than 0) is a Coolspot, which, along with its surroundings (neighboring functions), is executed infrequently, and functions plotted in the Coolspot are likely candidates for elimination.

[0082] For example, users (software architects, programmers, etc.) can efficiently select functions to eliminate by preferentially examining functions plotted in Coolspot.

[0083] Note that functions plotted in the second and fourth quadrants are more likely to be candidates for elimination than functions plotted in the hotspot, but less likely to be candidates for elimination than functions plotted in the coolspot. For example, a function plotted in the second quadrant tends to be executed less frequently than its neighboring functions, but more frequently than its neighboring functions. Similarly, a function plotted in the fourth quadrant tends to be executed more frequently than its neighboring functions, but more frequently than its neighboring functions.

[0084] [Effects of the First Embodiment] As described above, the generation unit 11 generates information indicating the relationships between multiple functions included in software. The prediction unit 13 predicts the frequency with which each of the multiple functions will be executed, based on the observed values ​​of the execution frequency of each of the multiple functions and the information generated by the generation unit 11.

[0085] This makes it possible to efficiently grasp the status of the functions included in the software.

[0086] For example, it would be extremely costly to incorporate a tool for data acquisition and transmission into all functions within software that is already running and measure the frequency of calls using the technology described in Non-Patent Document 1. Furthermore, having the tool running constantly while the software is in operation would lead to a decrease in system performance and response, thereby reducing usability and service levels.

[0087] On the other hand, the analysis device 10 can predict the execution frequency of a function based on data obtained from a limited observation period or a local number of observations, without measuring the call frequency.

[0088] The prediction unit 13 calculates the probability distribution of the frequency at which each of the multiple functions is executed, allowing the analysis device 10 to provide not only the execution frequency value but also additional information such as a credible interval.

[0089] The generation unit 11 generates a spatial proximity matrix based on the call relationships between the multiple functions as information indicating the relationships between the multiple functions. The prediction unit 13 predicts the execution frequency of each of the multiple functions using a spatial statistical model that represents the execution frequency of each of the multiple functions and includes a term obtained by multiplying the spatial proximity matrix by an observation value. This allows the analysis device 10 to predict the execution frequency of a function from the source code and a minimum observation value.

[0090] The prediction unit 13 predicts the frequency of execution of each of the plurality of functions using a spatial statistical model that further includes a term obtained by multiplying the regression coefficient by the attribute value of each of the plurality of functions, thereby enabling the analysis device 10 to predict the execution frequency with even higher accuracy.

[0091] The spatial distribution analysis unit 14 analyzes the spatial characteristics of each of the multiple functions in a space having each of the multiple functions as an element. As a result, the analysis device 10 can provide analysis results from multiple perspectives, as shown in Fig. 12, and assist the user in determining which functions to eliminate.

[0092] (When a Function Has Missing Values) The analysis device 10 of this embodiment can perform prediction using a spatial statistical model even when a function has missing values.

[0093] When a function is missing, that is, when an observed value of the execution frequency of the function cannot be obtained, the analysis device 10 sets the value of formula (6) to "NA". "NA" is a value indicating a missing value, not a numerical value. For example, when function 1 is missing, the analysis device 10 sets y 1 The value of is set to "NA".

[0094] The analysis device 10 then estimates the parameters of the spatial statistical model with the values ​​including "NA." At this time, the analysis device 10 estimates the parameters of all functions, including the missing function (step S202 in FIG. 7). The analysis device 10 then calculates the execution frequency using the estimated parameters (step S203 in FIG. 7).

[0095] "NA" is different from the observed value of the execution frequency being 0. If the observed value of the execution frequency of a certain function is 0, it means that the execution frequency of function 1 has been measured, and if it has not been executed even once, the analysis device 10 sets the value of the function in equation (6) to the numeric value "0" rather than "NA."

[0096] As a result, the analysis device 10 can estimate the execution status within the software even if it is not possible to observe the execution frequency of all functions over the entire period and there are functions for which data is missing.

[0097] [System Configuration, etc.] The components of each device shown in the figures are conceptual functional units and do not necessarily have to be physically configured as shown. In other words, the specific form of distribution and integration of each device is not limited to that shown, and all or part of the devices can be functionally or physically distributed or integrated in any unit depending on various loads, usage conditions, etc. Furthermore, all or any part of the processing functions performed by each device can be realized by a CPU and a program analyzed and executed by the CPU, or can be realized as hardware using wired logic.

[0098] Furthermore, among the processes described in this embodiment, all or part of the processes described as being performed automatically can be performed manually, or all or part of the processes described as being performed manually can be performed automatically using a known method.In addition, the information including the processing procedures, control procedures, specific names, various data and parameters shown in the above documents and drawings can be changed as desired unless otherwise specified.

[0099] [Program] In one embodiment, the analysis device 10 can be implemented by installing a program that executes the above-described learning process as package software or online software on a desired computer. For example, by executing the above-described program on an information processing device, the information processing device can function as the analysis device 10. The information processing device referred to here includes desktop and notebook personal computers. Other information processing devices also include mobile communication terminals such as smartphones, mobile phones, and PHS (Personal Handyphone Systems), as well as slate terminals such as PDAs (Personal Digital Assistants).

[0100] The analysis device 10 may also be implemented as a server device that provides services related to the above processing to a client terminal device used by a user. For example, the server device may be implemented as a server device that receives source code and an execution log as input and outputs a predicted execution frequency of each function. In this case, the server device may be implemented as a web server or as a cloud that provides services related to the above processing through outsourcing.

[0101] 13 is a diagram showing an example of a computer that executes a program. The computer 1000 includes, for example, a memory 1010 and a CPU 1020. The computer 1000 also includes a hard disk drive interface 1030, a disk drive interface 1040, a serial port interface 1050, a video adapter 1060, and a network interface 1070. These components are connected by a bus 1080.

[0102] The memory 1010 includes a read-only memory (ROM) 1011 and a random access memory (RAM) 1012. The ROM 1011 stores a boot program such as a basic input / output system (BIOS). The hard disk drive interface 1030 is connected to a hard disk drive 1090. The disk drive interface 1040 is connected to a disk drive 1100. A removable storage medium such as a magnetic disk or optical disk is inserted into the disk drive 1100. The serial port interface 1050 is connected to a mouse 1110 and a keyboard 1120, for example. The video adapter 1060 is connected to a display 1130, for example.

[0103] The hard disk drive 1090 stores, for example, an OS 1091, an application program 1092, a program module 1093, and program data 1094. That is, the programs that define each process of the analysis device 10 are implemented as program modules 1093 in which computer-executable code is written. The program modules 1093 are stored, for example, in the hard disk drive 1090. For example, the program modules 1093 for executing processes similar to those of the functional configuration of the analysis device 10 are stored in the hard disk drive 1090. The hard disk drive 1090 may be replaced by an SSD (Solid State Drive).

[0104] Furthermore, setting data used in the processing of the above-described embodiment is stored as program data 1094, for example, in the memory 1010 or the hard disk drive 1090. Then, the CPU 1020 reads the program module 1093 or the program data 1094 stored in the memory 1010 or the hard disk drive 1090 into the RAM 1012 as necessary, and executes the processing of the above-described embodiment.

[0105] The program module 1093 and program data 1094 may not necessarily be stored in the hard disk drive 1090, but may also be stored in a removable storage medium and read by the CPU 1020 via the disk drive 1100 or the like. Alternatively, the program module 1093 and program data 1094 may be stored in another computer connected via a network (such as a local area network (LAN) or a wide area network (WAN)). The program module 1093 and program data 1094 may then be read by the CPU 1020 from the other computer via the network interface 1070.

[0106] 10 Analysis device 11 Generation unit 12 Preprocessing unit 13 Prediction unit 14 Spatial distribution analysis unit

Claims

1. An analysis device comprising: a generation unit that generates information indicating the relationships between multiple functions included in software; and a prediction unit that predicts the frequency with which each of the multiple functions will be executed based on observed values ​​of the frequency with which each of the multiple functions is executed and the information generated by the generation unit.

2. The analysis device according to claim 1, wherein said prediction unit calculates a probability distribution of the frequency with which each of said plurality of functions is executed.

3. The analysis device according to claim 1, characterized in that the generation unit generates a spatial proximity matrix based on the call relationships between the plurality of functions as information indicating the relationships between the plurality of functions.

4. The analysis device described in claim 3, characterized in that the prediction unit predicts the frequency with which each of the multiple functions is executed using a spatial statistical model that represents the frequency with which each of the multiple functions is executed and that includes a term obtained by multiplying the spatial proximity matrix by the observed value.

5. The analysis device described in claim 4, characterized in that the prediction unit predicts the frequency with which each of the multiple functions is executed using the spatial statistical model, which further includes a term obtained by multiplying a regression coefficient by an attribute value of each of the multiple functions.

6. The analysis device according to claim 1, further comprising a spatial distribution analysis section for analyzing spatial characteristics of each of said plurality of functions in a space having each of said plurality of functions as an element.

7. An analysis method executed by an analysis device, comprising: a generation step of generating information indicating relationships between a plurality of functions included in software; and a prediction step of predicting the frequency with which each of the plurality of functions will be executed based on observed values ​​of the frequency with which each of the plurality of functions is executed and the information generated by the generation step.

8. An analysis program that causes a computer to execute the following steps: a generation step of generating information indicating the relationships between multiple functions included in software; and a prediction step of predicting the frequency with which each of the multiple functions will be executed based on the observed values ​​of the frequency with which each of the multiple functions is executed and the information generated by the generation step.

Citation Information

Patent Citations

  • Editor for specification description

    JP1989116729A

  • Program for predicting source code problem

    JP2007179488A