Process mining system and process mining method

Abstract

To provide a process mining system and method that analyze not only the implementation order of tasks but also pre- and post-dependencies.SOLUTION: A calculation device of a computer 100 takes in from a memory a plurality of pieces of deliverable information created in the process of carrying out work and access logs to the plurality of pieces of deliverable information and configures event data on the basis of the plurality of pieces of deliverable information and the access logs. The event data includes a plurality of tasks that correspond to each of the plurality of pieces of deliverable information and the date and time of executing each of the plurality of tasks. The method includes visualizing a work process that connects the plurality of tasks in chronological order on the basis of the event data, linking a predetermined task belonging to the plurality of tasks with reference information accessed when the work is promoted, setting a dependence between the predetermined task and a task whose outcome is deliverable information to which the reference information corresponds when the reference information corresponds to any of the plurality of pieces of deliverable information, and adding the dependence to the work process so as to make it visible.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a system for performing process mining and a method thereof. Process mining is known as a technique for supporting business efficiency improvement by visualizing the processing pattern of a business process through the accumulation of event log data and specifically identifying improvement points.

Background Art

[0002] DX (Digital Transformation) represented by the introduction of "RPA (Robotic Process Automation)" has attracted attention as an effort to improve business efficiency and productivity. RPA is software that substitutes or replaces work that was previously handled only by humans or more advanced work by utilizing AI, machine learning, etc. DX is for a company to utilize digital technology and transform its organization and business model. In such DX efforts, in order to identify inefficient operations to be improved or transformed, for example, visualization of business processes as shown in Patent Document 1 has been carried out.

[0003] On the other hand, regarding the identification of this business process, it has required a great deal of time and effort, such as observing an organization where experts with business analysis skills are performing operations, conducting interviews with the operation staff, reading operation manuals and standards, manually writing them up, and further repeating the confirmation and correction of the written-up content with various stakeholders related to the operation.

[0004] Depending on the business understanding of the interviewees, exceptional operations with low execution frequency may be overlooked. Furthermore, the more advanced the operation, the more rapidly the way of conducting the operation continues to change. For this reason, even if the business process at a certain point in time is identified, the business reform is not over, and it must be continued following the changes.

[0005] Therefore, in recent years, process mining systems have attracted attention as technologies for automating the discovery of business processes. This is a technology or method for estimating the tasks being performed in a business and the order of their execution (business process) from the logs in the IT systems used when conducting business, that is, operation histories, event data, etc., as shown in, for example, Patent Documents 1 and 2 and Non-Patent Document 1. A task is an element that constitutes a business process. For example, in the procurement business process, there are tasks such as "create a purchase specification" and "check inventory", and in the design business process, there are tasks such as "design the shape of component XX", "evaluate the strength of component XX", and "create materials for design review". By this process mining technology, it is possible to automate or streamline the business analysis that conventionally had to be performed by experts, and by continuously using it, changes in the business process can also be captured.

Prior Art Documents

Patent Documents

[0006]

Patent Document 1

Patent Document 2

Non-Patent Documents

[0007]

Non-Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0008] However, the above technologies have the following problems. In Patent Document 1, the extraction of business processes needs to be carried out by experts. Although the technologies described in Patent Document 2 and Non-Patent Document 1 can estimate the execution order of tasks, they cannot clarify the precedence dependencies between tasks. The precedence dependency is a mandatory relationship regarding precedence / succession among multiple tasks. For example, in the procurement business, before the task of "conducting inspection and acceptance", the task of "delivering / accepting goods" must be carried out. In the design business, before the task of "analyzing the strength of the structure", the task of "designing the CAD shape of the structure" must be carried out.

[0009] Therefore, for business process reform involving swapping the order of multiple tasks and concurrent processing of multiple tasks, in addition to the conventional process mining technology, it is necessary to conduct business analysis such as interviews by experts.

[0010] Therefore, an object of the present invention is to provide a process mining system and a method thereof that can analyze not only the execution order of tasks but also the precedence dependencies between tasks.

Means for Solving the Problem

[0011] To achieve the above object, the present invention provides a system in which a computing device of a computer executes process mining of a business based on a program. The computing device fetches, from a memory, a plurality of work product information created in the process of advancing the business and access logs for each of the plurality of work product information, constructs event data based on the plurality of work product information and the access logs for each of the plurality of work product information, the event data including a plurality of tasks corresponding to each of the plurality of work product information and the execution date and time of each of the plurality of tasks, visualizes a business process in which the plurality of tasks are connected in chronological order based on the event data, links a predetermined task belonging to the plurality of tasks and reference information accessed when advancing the business, and when the reference information corresponds to any of the plurality of work product information, sets a dependency relationship between the predetermined task and a task that takes as a result the work product information corresponding to the reference information, and adds and visualizes the dependency relationship to the business process. Further, another invention is a method for process mining of a business.

Effects of the Invention

[0012] According to the present invention, it is possible to provide a process mining system and a method thereof that can analyze not only the execution order of tasks but also the dependency relationship between tasks.

Brief Description of the Drawings

[0013]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Embodiments for Carrying Out the Invention

[0014] Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a process mining system. This system includes a computer 100. The deliverable file 101, the reference file 102, and the business process data 105 are recorded in the memory. A controller or a processor as an arithmetic device of the computer hardware executes a program recorded in the memory to realize functional modules for process mining, such as an access date and time acquisition unit 103, an event data generation unit 104, a business process estimation unit 106, and a business process visualization unit 108. Various functional units may be rephrased as means.

[0015] The group of deliverable files 101 as deliverable information of the business process is created in the process of performing the business, and includes, for example, a plurality of document files. The access date and time acquisition unit 103 acquires the access date and time (access log) of various documents in the group of deliverable files 101 and the reference file (reference information) 102, which is a document or a website referred to during the business execution. The event data generation unit 104 generates event data based on metadata such as the creation and / or editing date and time, the accessed file name, and the file type at the time of accessing the group of deliverable files 101. An event refers to a task of the business process. The date and time of the access log to the file is used as the execution date and time of the task (log to the task), and both are included in the event data. Details will be described later. Note that a task is one element constituting the business process.

[0016] The business process estimation unit 106 estimates the business process based on the event data and generates business process data 105. Note that the estimation may be rephrased as analysis, analysis, determination, identification, or decision.

[0017] The task-reference information association unit 107 links the task and the reference data based on the reference date and time data of the deliverable file, the reference date and time data of the reference files other than the deliverable file, and the estimated business process data 105, and extracts the files referred to during the execution period of each task.

[0018] When the reference file is included in the deliverable file group, the task-reference information association unit 107 sets the task estimated based on the reference file as the preceding task of the target task. When the reference file is not included in the deliverable file group 101, it registers the business process data 105 as reference information for the target task. The business process visualization unit 108 visualizes the business process data 105.

[0019] FIG. 2 shows a block diagram according to a second embodiment of the process mining system. In this embodiment, an application log generation unit 201 for recording the elapsed date and time of operations such as the startup, termination, active, and inactive states of the application is added to the system of FIG. 1.

[0020] The task-reference information association unit 107 associates the reference files of each task according to the application log, for example, the time ratio and switching frequency of the active applications.

[0021] The deliverable file group 101 is a set of files storing document files created during the process of business execution. The storage destination may be a database such as a document management system, a file server, or a local PC. Although the name "file" is used for convenience, it can be any data created, defined, etc. during the process of business execution. For example, it may be a part of a data table in an arbitrary database.

[0022] The deliverables in the development and design stages are reports, specifications, studies, technical materials, drawings, etc. Specific examples include functional specifications, experiment reports, technical materials recording assumed issues and countermeasures, etc. In the case of calculation sheets for strength, thermal load, cost, etc., they are files such as spreadsheet software or simulation software created in past projects. It includes 2D drawings, 3D CAD data, component assembly information, various attribute information (material, component type, welding instructions, etc.) showing the shape and structure.

[0023] In the production stage, there are die data, manufacturing process sheets, assembly process sheets, quality plans, inspection records, cost estimates, etc. In addition, all things created during the process of a certain business, such as program codes, meeting minutes, test data, operation manuals, etc., are deliverables.

[0024] The reference file 102 is a set of files referenced during the process of business execution. The reference files include information on the files and data in the deliverable file group 101, as well as other files. For example, in the task of "analyzing the strength of a structure", the "CAD data of the structure", which is the deliverable of the task of "designing the CAD shape of the structure", is referenced. And in the task of "analyzing the strength of a structure", files not included in the deliverable file group 101, such as "collection of analysis know-how" and "design standards", are also referenced.

[0025] The access date and time acquisition unit 103 acquires the access dates and times of the various files in the deliverable file group 101 and the reference file 102, and collectively records them in the memory as a table. The access date and time may be a time stamp of access such as reference, creation, or editing. An example of a table of the access dates and times of the various files shown in the deliverable file group 101 is shown in FIG. 3.

[0026] The event data generation unit 104 generates event data based on, for example, the creation or editing date and time of the file and the file type at the access date and time of the deliverable file group 101. The file type indicates, for example, which task in the business process the deliverable is. Specifically, it is something like "CAD data of a structure" or "intensity analysis result data".

[0027] For the deliverable file group 101, means for interactively specifying the file type may be applied to the system of FIG. 1, or means for automatically estimating the file type may be applied. An example of the latter will be described.

[0028] The latter means vectorizes each file in the deliverable file group 101, calculates the similarity between the vectors, clusters based on the similarity, and assigns the same type to the files within the same cluster. The name of the type may be the representative file name of the files within the cluster.

[0029] Although various methods have been proposed as the method for vectorizing each file and the method for calculating the similarity, the present invention does not need to be limited to a specific one. Any method that can simply calculate the similarity between files may be used. Note that the files include, for example, text documents, image data, spreadsheet data, 3D shape data, and the like.

[0030] An example of a method for vectorizing a text document and calculating similarity will be described. The text document is segmented into words using morphological analysis. This morphological analysis is not necessary when words in English or the like are separated by spaces. Next, the number of occurrences of each word is counted and represented as a vector. The cosine similarity is calculated for the vectors created for two documents (Reference: Eiji Nanba, Measurement of Similarity between Texts, Information Science and Technology, 2020, Vol. 70, No. 7, pp. 373-375).

[0031] The similarity is calculated based on the word frequency of occurrence. However, other methods for evaluating the importance of words, such as TF-IDF, may also be used (Reference: Akinobu Kageyama, Hiroshi Tsuji, Method for Extracting Features of Research Institutions Using the TF / IDF Algorithm, Transactions of the Institute of Electrical Engineers of Japan, Part C (Journal of Electronics, Information and Systems), 2005, Vol. 125, No. 5, pp. 713-719).

[0032] An example of a method for calculating similarity for 3D shape data will be described. In this calculation method, the target shape is represented in a voxel structure, blocks are defined by several adjacent voxels, the spectrum is calculated for each block, and it is represented as a vector (Reference: Patent No. 5024767). This calculation method calculates the cosine similarity for the vectors created for two 3D shape data in the same way as in the case of documents.

[0033] Similarly, with regard to the clustering method, it is not limited in the present invention. For example, the K-means method, the shortest distance method, the multidimensional scaling method, the Ward method, etc. are possible.

[0034] The event data generation unit 104 generates, as event data, for example, the table structure and its data shown in FIG. 4, and records this in the memory. This table has six fields: an ID, a case name, a task name based on the file type, a timestamp which is the date and time of the task, and the person in charge. FIG. 4 is an example of event data generated from the access date and time of the work product file group 101 shown in FIG. 3. Note that the case name is automatically identified from, for example, the storage location of the work product (such as a folder name) or a project plan associated with the person in charge (which clearly shows the cases and periods that the person in charge is engaged in). The system of FIG. 1 is not limited with respect to the method of obtaining the case name, and may have, for example, means for designating the case name.

[0035] The business process estimation unit 106 estimates the business process from the event data and generates business process data 105. The business process is a sequence of a plurality of tasks arranged in time series. Although various methods have been proposed as methods for estimating the business process from event data, the types thereof are not limited in the present invention. An example of a typical method will be described.

[0036] First, the business process estimation unit 106 arranges the tasks in the order of execution, that is, in time series, for each case. For example, assume that the following business process flows for each case are obtained. The alphabets are task names, and (A, B, C) indicates that they were executed in the order of A → B → C. Case 1: (A, B, C, D) Case 2: (A, C, D) Case 3: (A, E, C, D) Case 4: (A, B, C, E, C, D)

[0037] Next, the business process estimation unit 106 extracts the transitions of the tasks for all cases. For example, looking only at Case 1, the task transition can be expressed as in 501 of FIG. 5. When the business process estimation unit 106 then adds the transition pattern of Case 2 to this transition, it becomes as in 502 of FIG. 5. Similarly, when the transition patterns of Case 3 and Case 4 are added, they become as in 503 and 504 respectively.

[0038] The business process estimation unit 106 thus sets a model representing the task transition as business process data 105 and registers it in the memory. As models representing task transitions, BPMN (Business Process Model and Notification), Petri nets, UML activity diagrams, etc. have been proposed, and they can be appropriately selected and their types are not limited. For example, for the event data shown in FIG. 4, the business process data 105 in which the business process is estimated is as shown in FIG. 6, and the business process visualization unit 108 displays this on the output device of the user's computer 100.

[0039] Based on the reference date and time data of the deliverable file 101 (FIG. 3), the reference date and time data of the reference files 102 other than the deliverable file, and the business process data (FIG. 6) estimated by the business process estimation unit, the task-reference information association unit 107 extracts the files referred to during the execution period of each task. When the reference file is included in the group of deliverable files 101, the task estimated based on the reference file is set as the preceding task of the target task. When the reference file is not included in the group of deliverable files, it is registered in the business process data 105 as reference information for the target task.

[0040] Each task constituting the business process is often performed while referring to some information. For example, the task of "analyzing the strength of a structure" refers to the "CAD data of the structure". Since this "CAD data of the structure" is the deliverable of the task of "designing the CAD shape of the structure", "designing the CAD shape of the structure" must be performed prior to "analyzing the strength of the structure". That is, the "CAD data of the structure" is referred to during the execution period of the task of "analyzing the strength of the structure", and the dependency relationship between tasks is estimated based on the access date and time to this.

[0041] To explain more specifically, the access date and time acquisition unit 103 records the reference date and time data of the work product file and the reference date and time data of reference files other than the work product file in the memory as the storage table shown in FIG. 7. The business process is as shown in FIG. 6, and the event data that is the basis for the business process estimation is as shown in FIG. 4.

[0042] The task-reference information association unit 107 determines that the "Product Specification.docx" shown in the row with ID = 1 in FIG. 7 was referenced on June 10, 2021. Then, the task-reference information association unit 107 determines that in the event data shown in FIG. 4, the task of "Determining operating conditions" was also carried out on June 10, 2021. That is, the task of "Determining operating conditions" refers to the "Product Specification.docx". Since the "Product Specification.docx" is not included in the work product file group 101, this is reference information.

[0043] The "Company X\Operating Condition Definition_Tanaka_210110.docx" shown in the row with ID = 3 in FIG. 7 was referenced on June 11, 2021. Also, in the event data shown in FIG. 4, the task of "Selecting equipment configuration" was carried out on June 12, 2021, and the task of "Determining operating conditions" that was carried out before that was carried out on June 10, 2021. Therefore, the task-reference information association unit 107 determines that "Selecting equipment configuration" was carried out during the period from June 11, 2021 to June 12, 2021. For this reason, the task of "Selecting equipment configuration" refers to the "Company X\Operating Condition Definition_Tanaka_210110.docx".

[0044] Since the "Company X\Operating Condition Definition_Tanaka_210110.docx" is the work product file of the task of "Determining operating conditions", the preceding task of "Selecting equipment configuration" is "Determining operating conditions".

[0045] In the above example, the files that have been referenced even once are associated with the tasks. However, the association (link) may be set according to the superiority or inferiority of the references, for example, the reference probability (frequency). For example, the "http: / / ×××.com / news1234.html" shown in the line with ID = 26 in Figure 7 is referenced during the execution of the task of "performing a design review" (Figure 4: ID = 17). "Performing a design review" has been executed three times (Figure 4: ID = 14, 15, 17), but in other cases (Figure 4: ID = 14, 15), this file has not been referenced. As a result, the reference probability is only 33.3%, and since the frequency of reference is low, it may not be associated with the task as a referenced file. Thus, it is desirable to be able to set a threshold for the reference probability. The superiority or inferiority may be appropriately set according to other factors such as the number of reference times, reference time, and number of reference persons in addition to the reference probability. The task-reference information association unit 107 may display a menu for setting superiority or inferiority information, such as item selection of superiority or inferiority and threshold of superiority or inferiority, on the screen of Figure 8 so that the user can set the superiority or inferiority information.

[0046] In this way, the task-reference information association unit 107 associates the preceding tasks and reference information of each task with the business process. The task-reference information association unit 107 records the information in which the preceding tasks and reference information are linked to the business process, that is, the information in Figure 8, in the memory. The business process visualization unit 108 displays this information to the user via the input device 100 of the computer.

[0047] In Figure 8, the execution order between multiple tasks is defined as a solid-line arrow, and the dependency relationship between multiple tasks is defined as a dashed-line arrow (an arrow pointing from a subsequent task to a preceding task). The reference information is shown as a parallelogram.

[0048] In FIG. 8, for example, the task of "evaluating heat resistance" is performed after the task of "evaluating performance reliability". However, there is no broken line from the task of "evaluating heat resistance" to the task of "evaluating performance reliability", and the latter task is not a preceding task for the former task. That is, since there is no dependency relationship between the tasks of "evaluating heat resistance" and "evaluating performance reliability", either task can be performed first, or if there is spare capacity in the personnel, they can be performed in parallel. The fact that both tasks can be performed in parallel means that the business period can be shortened, which is an important piece of advice for business process reform. Conventional process mining techniques could not clarify the existence of such a preceding and subsequent relationship between multiple tasks.

[0049] The application log generation unit 201 in FIG. 2 records the start time, end time, active time, and non-active time of the application in the memory as the execution log of the application. For example, while the task of "analyzing the strength of a structure" is being performed, the analysis application is started, and the analysis application refers to the "CAD data of the structure". During that time, the CAD application is also started in parallel. The task-reference information association unit 107 associates a reference file with the task based on the execution log of the application.

[0050] While performing business, users often check emails using a mail application or check the weather and news using a web browser application, for example. In that case, it is not desirable for the email or news site to be linked as reference information for "analyzing the strength of a structure".

[0051] The operation times of the analysis application and the CAD application are long. On the contrary, the operation times of the mail application and the web browser application are short. And the switching frequency between the analysis application and the CAD application is high, while the switching frequency to the mail application and the web browser application is low.

[0052] The task-reference information association unit 107 links a task and a reference file based on the time ratio of the active application using the application execution log. For example, in the case of a file referenced during the execution of a certain task, when the time ratio B / A (where A is the execution period of the task and B is the referenced time (the time when the application was active)) exceeds a threshold value (for example, 10%), the task-reference information association unit 107 links the reference file to the task.

[0053] Instead of, or in addition to, the time ratio, the switching frequency may be used. For example, assume that the active application is switched N times during the execution of a certain task. Let the number of switching times for each application be n i be. When this n i / N exceeds a threshold value (for example, 30%), the task-reference information association unit 107 associates the reference file with the task.

[0054] In the example shown in FIG. 8, in the task of "performing a design review", the task of "evaluating heat resistance" is not a preceding task. This may be due to chance or a problem in process mining. Therefore, there may be cases where a user or administrator wants to associate another task as a preceding task with a specific task. Therefore, input / output means for adding, deleting, and changing the dependency relationship of tasks may be added to the system. FIG. 9 shows a flowchart in which the dependency relationship of tasks is added to the business process. A dashed line from the task of "performing a design review" to the task of "evaluating heat resistance" is added as the dependency relationship, indicating that the latter task is the preceding task of the former task.

[0055] The task-reference information association unit 107 may add the occurrence probability of the preceding task to the display of the dependency relationship (dashed line). "There is a 20% probability of a preceding task" in Fig. 9 is an example of this additional display. For example, assuming that data for 10 projects has been prepared as work product files and reference files for process mining, in 2 projects, "Evaluate heat resistance" is the preceding task of "Conduct design review", but in 8 projects, it was not the preceding task. Therefore, it may be statistically determined not to set a preceding task. Thus, it is beneficial to attach related displays such as the probability of a preceding task to the business process flow diagram for the dependency relationship.

[0056] The present invention is not limited to the above-described embodiments and includes various modifications. For example, the above-described embodiments have been described in detail for easy understanding of the present invention and are not necessarily limited to those having all the configurations described. Also, a part of the configuration of one embodiment can be replaced with the configuration of another embodiment, and the configuration of another embodiment can be added to the configuration of one embodiment. Further, for a part of the configuration of each embodiment, addition, deletion, and replacement with other configurations are possible. Also, each of the above configurations, functions, processing units, processing means, etc. may be realized in hardware, for example, by designing a part or all of them with an integrated circuit. Also, each of the above configurations, functions, etc. may be realized in software by a processor interpreting and executing a program for realizing each function. Information such as a program, table, file, etc. for realizing each function can be placed in a memory, a recording device such as a hard disk, SSD (Solid State Drive), or a recording medium such as an IC card, SD card, DVD.

Explanation of Reference Numerals

[0057] 100: Computer, 101: Deliverable file, 102: Reference file, 103: Access date and time acquisition unit, 104: Event data generation unit, 105 Business process data, 106: Business process estimation unit, 107: Task-reference information linking unit, 108: Business process visualization unit, 201: Application log generation unit

Claims

1. A system in which an arithmetic unit of a computer executes process mining of a business based on a program, The arithmetic unit, takes in, from a memory, a plurality of work product information created in the process of advancing the business and access logs for each of the plurality of work product information, configures event data based on the plurality of work product information and the access logs for each of the plurality of work product information, and the event data includes a plurality of tasks corresponding to each of the plurality of work product information and the execution date and time of each of the plurality of tasks, visualizes a business process in which the plurality of tasks are connected in chronological order based on the event data, links a predetermined task belonging to the plurality of tasks with reference information accessed when advancing the business, and when the reference information corresponds to any of the plurality of work product information, a dependency is set between the predetermined task and a task that takes as its result the work product information corresponding to the reference information, adds the dependency to the business process and visualizes it, determines the reference information to be linked to the predetermined task based on the execution date and time of the predetermined task and the access log of the reference, determines whether or not to link the predetermined task and the reference information according to the superiority or inferiority of the reference, A process mining system.

2. The arithmetic unit, determines the task based on the metadata of the work product file as the work product information, links the work product file as the result of the task, The process mining system according to Claim 1.

3. The arithmetic unit, sets the dependency such that a task that takes as its result the work product information corresponding to the reference information is a task that should precede the predetermined task, The process mining system according to claim 1.

4. The arithmetic unit When the reference information does not correspond to any of the plurality of work product information, add a link between the predetermined task and the reference information to the business process and visualize it. The process mining system according to claim 1.

5. The arithmetic unit Link the reference information to the predetermined task based on the execution log of the application used in the business. The process mining system according to claim 1.

6. The arithmetic unit Can add a dependency between multiple tasks in the business process. The process mining system according to claim 1.

7. A process mining method for a business executed by a computer, Capture a plurality of work product information created in the process of proceeding with the business and access logs for each of the plurality of work product information, Based on the plurality of work product information and the access logs for each of the plurality of work product information, configure event data, and the event data includes a plurality of tasks corresponding to each of the plurality of work product information and the execution date and time of each of the plurality of tasks. Visualize the business process in which the plurality of tasks are connected in chronological order based on the event data, Link a predetermined task belonging to the plurality of tasks and reference information accessed when proceeding with the business. When the reference information corresponds to any of the plurality of work product information, set a dependency between the predetermined task and the task that takes the work product information corresponding to the reference information as a result. Add the dependency to the business process and visualize it. Based on the execution date and time of the predetermined task and the access log of the reference, determine the reference information to be linked to the predetermined task. Determine whether it is necessary to link the predetermined task and the reference information according to the superiority or inferiority of the reference. Process mining method.

Images