Artificial intelligence robot process mining system

Abstract

The invention discloses an artificial intelligence robot process mining system, and relates to the technical field of artificial intelligence, and the system comprises a data and model unit which is used for collecting data, carrying out the multiple analysis processing, and combining a process mining algorithm and a tool to construct a process model; the model middle table unit is used for classifying and combining the constructed process models and managing a combination result; and the application layer unit is used for docking various business systems of an enterprise and automatically executing a business process. According to the method, the automation technology and the artificial intelligence technology are fused and applied to process mining, so that the identification and execution efficiency of the process can be improved; rPA is used for replacing manual operation, and automatic processing of repeated tasks is achieved; and in combination with a non-frequent inductive algorithm, the flow chart is simplified, and the calculation complexity is reduced, so that the efficiency and the intelligent level of flow mining are further improved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and more specifically, to an artificial intelligence robot process mining system. Background Technology

[0002] In the process of enterprise digital transformation, lean analytics and advanced analytics are key stages for improving efficiency and reducing costs. After establishing a certain level of automation and information technology infrastructure, enterprises typically begin optimizing business processes to improve operational efficiency and reduce waste. In traditional lean analytics methods, industrial engineers or consultants identify problems in production and operations and guide continuous improvement through on-site diagnosis and analysis. However, traditional methods rely heavily on experience, and the diagnostic process has certain limitations and lags.

[0003] Most manufacturing enterprises are still lagging behind in implementing lean management, struggling to fully utilize data and information technology for systematic and intelligent analysis. The core task of the current lean analysis phase is to identify and optimize key processes in enterprise production management through process mining tools combined with real-time data, thereby achieving more accurate, timely, and comprehensive diagnostics. While traditional process mining techniques can achieve process reengineering and optimization, their efficiency is low, and the lack of deep application of artificial intelligence and automation technologies limits the efficiency of business process execution.

[0004] There are currently no effective solutions to the problems in the relevant technologies. Summary of the Invention

[0005] In response to the problems in related technologies, this invention proposes an artificial intelligence robot process mining system to overcome the aforementioned technical problems existing in the existing related technologies.

[0006] Therefore, the specific technical solution adopted by the present invention is as follows:

[0007] An artificial intelligence robot process mining system, the system comprising:

[0008] The data and model unit is used to collect data and perform multiple analysis and processing steps, combining process mining algorithms and tools to build process models;

[0009] The model platform unit is used to classify and combine the constructed process models, and manage the combination results.

[0010] Application layer units are used to interface with various enterprise business systems and automate business processes.

[0011] Furthermore, the data and model units include:

[0012] The data source module is used to store structured data, unstructured data, and external data.

[0013] The data acquisition module is used to collect data from the data source module and upload it to the data center module through extraction, transformation and loading operations;

[0014] The data center module is used to store the collected data, perform initial analysis and processing on the data, and upload the results of the initial analysis and processing to the data platform module.

[0015] The data middle platform module is used to receive the initial analysis and processing results uploaded by the data center module and perform secondary analysis and processing.

[0016] The model development module is used to model processes based on the results of secondary analysis and processing, using process mining algorithms and tools to build process models suitable for enterprise business.

[0017] Furthermore, the data center module includes:

[0018] The data warehouse submodule is used to receive and store data uploaded by the data acquisition module;

[0019] The database submodule is used for mining and analyzing the data stored in the data warehouse submodule;

[0020] The data processing submodule is used to perform initial analysis and processing of the mining and analysis results using big data processing tools.

[0021] Furthermore, the data platform module includes:

[0022] The wide table submodule is used to receive the initial analysis and processing results uploaded by the data center module, and to build a database table by integrating the multi-dimensional data fields in the initial analysis and processing results;

[0023] The Data Mart submodule is used to analyze target indicators based on data in the database tables using a pre-defined data analysis model.

[0024] The data element submodule is used to define the data elements of the process model based on the analysis results.

[0025] Furthermore, data elements include: model tag library, processing logic, data assets, and data structure.

[0026] Furthermore, the semantics of a data element are defined by three elements: object class, characteristics, and representation.

[0027] Furthermore, the model development module includes:

[0028] Tool sub-molds are used to deploy and execute process mining algorithms using programming languages ​​and machine learning tools;

[0029] The algorithm submodule is used to model the secondary analysis results using several process mining algorithms, and to build a process model suitable for enterprise business.

[0030] Furthermore, process mining algorithms include: infrequent inductive mining algorithms, alpha series algorithms, heuristic algorithms, genetic algorithms, and log classification algorithms.

[0031] Furthermore, the model platform unit includes:

[0032] The model marketplace module is used to classify process models according to different business domains of an enterprise and build corresponding domain models.

[0033] The model combination module is used to analyze the industry to which the models in each field belong, and to combine the field models within the same industry to construct an industry model;

[0034] The model management module is used to manage domain models and industry models.

[0035] Furthermore, the application layer unit includes:

[0036] Customized modules are used to connect with various business systems of enterprises and can be customized according to the needs of different industries;

[0037] The automation module is used to automate business processes using robotic process automation (RPA) technology and to obtain handles to desktop form controls and web page elements.

[0038] The beneficial effects of this invention are as follows:

[0039] This invention integrates automation and artificial intelligence technologies for process mining, thereby improving the efficiency of process identification and execution; it uses RPA to replace manual operations, realizing the automated processing of repetitive tasks; and it combines infrequent inductive algorithms to simplify flowcharts and reduce computational complexity, thus further improving the efficiency and intelligence level of process mining. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a schematic diagram of an artificial intelligence robot process mining system according to an embodiment of the present invention;

[0042] Figure 2 This is a flowchart of an artificial intelligence robot process mining system according to an embodiment of the present invention;

[0043] Figure 3 This is a schematic diagram of a direct follow-up activity in an artificial intelligence robot process mining system according to an embodiment of the present invention;

[0044] Figure 4 This is a schematic diagram of direct following graph segmentation in an artificial intelligence robot process mining system according to an embodiment of the present invention.

[0045] In the picture:

[0046] 1. Data and Model Unit; 2. Model Platform Unit; 3. Application Layer Unit. Detailed Implementation

[0047] To further illustrate the various embodiments, the present invention provides accompanying drawings, which are part of the disclosure of the present invention. These drawings are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these drawings, those skilled in the art should be able to understand other possible implementation methods and the advantages of the present invention.

[0048] According to an embodiment of the present invention, an artificial intelligence robot process mining system is provided.

[0049] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments, such as... Figure 1 As shown, according to an embodiment of the present invention, an artificial intelligence robot process mining system includes:

[0050] Data and Model Unit 1 is used to collect data and perform multiple analysis and processing operations, and to build process models by combining process mining algorithms and tools.

[0051] In this optional embodiment, the data and model unit 1 includes:

[0052] The data source module is used to store structured data, unstructured data, and external data.

[0053] It should be noted that structured data is compatible with IBM DB2 database, Microsoft SQL Server database, and Oracle Database; unstructured data includes audio, video, images, text, etc.; external data plays a certain auxiliary role in process mining. For example, for the financial market department of a bank, financial market data can be introduced as external data to optimize parameters in the system.

[0054] The data acquisition module is used to collect data from the data source module and upload it to the data center module through extraction, transformation and loading operations.

[0055] The data center module is used to store the collected data, perform initial analysis and processing on the data, and upload the results of the initial analysis and processing to the data platform module.

[0056] In this optional embodiment, the data center module includes:

[0057] The data warehouse submodule is used to receive and store data uploaded by the data acquisition module;

[0058] The database submodule is used for mining and analyzing the data stored in the data warehouse submodule;

[0059] The data processing submodule is used to perform initial analysis and processing of the mining and analysis results using big data processing tools.

[0060] It should be noted that the data warehouse submodule uses GP and stores the data collected by the data acquisition module from the data source module; the database submodule is implemented using the Hadoop Base database for big data mining and analysis; and the data processing submodule uses Storm (a big data processing tool) for real-time analysis and processing of enterprise process data.

[0061] The data middle platform module is used to receive the initial analysis and processing results uploaded by the data center module and perform secondary analysis and processing.

[0062] In this optional embodiment, the data platform module includes:

[0063] The wide table submodule is used to receive the initial analysis and processing results uploaded by the data center module, and to build a database table by integrating the multi-dimensional data fields in the initial analysis and processing results;

[0064] The Data Mart submodule is used to analyze target indicators based on data in the database tables using a pre-defined data analysis model.

[0065] The data element submodule is used to define the data elements of the process model based on the analysis results.

[0066] In this optional embodiment, the data elements include: a model tag library, processing logic, data assets, and data structure.

[0067] In this optional embodiment, the semantics of a data element is defined by three elements: object class, characteristics, and representation.

[0068] It should be further explained that the wide table submodule is a database table type implemented by integrating multi-dimensional data fields, mainly used to improve data query efficiency and analysis capabilities; the data mart submodule includes process marts, customer marts, marketing marts, and behavior marts, which support in-depth analysis of key indicators through multi-dimensional storage and data analysis models; data elements include model tag libraries, processing logic, data assets, and data structures. Data elements are the basic units for building data models, and independent semantic concepts are defined through the three elements of object class, characteristics, and representation.

[0069] The model development module is used to model processes based on the results of secondary analysis and processing, using process mining algorithms and tools to build process models suitable for enterprise business.

[0070] In this optional embodiment, the model development module includes:

[0071] Tool sub-molds are used to deploy and execute process mining algorithms using programming languages ​​and machine learning tools.

[0072] In this optional embodiment, the process mining algorithm includes: infrequent inductive mining algorithm, alpha series algorithm, heuristic algorithm, genetic algorithm and log classification algorithm.

[0073] The algorithm submodule is used to model the secondary analysis results using several process mining algorithms, and to build a process model suitable for enterprise business.

[0074] It should be noted that the process is usually represented using various flowcharts, including Petri nets, process trees, BPMN, and direct follow-up graphs. Specific algorithms include inductive mining algorithms, alpha series algorithms, heuristic algorithms, genetic algorithms, and log classification algorithms. The main tools are the Python development language and machine learning tools.

[0075] Furthermore, the specific steps of the infrequent inductive mining algorithm are as follows:

[0076] I. The problem of discovering a log L is decomposed into discovering n sub-processes of n sub-logs obtained by splitting L; where four basic operators are used: × represents the exclusive operator; -> represents the sequential operator; Q represents the loop operator; ^ represents the concurrency operator. The rules for the four operators are as follows:

[0077] The rules of the exclusive operator are that there is no relationship between a subprocess A and another subprocess B. An activity in subprocess A cannot be a successor activity in another subprocess B, and an activity in subprocess B cannot be a successor activity in another subprocess A. The two are not related to each other.

[0078] The rule for sequential operators is that there are outgoing edges from one subprocess A to another subprocess B, but no incoming edges; in general, there is only one path from one subprocess to the other.

[0079] The rule for concurrency operators is that a subprocess A has both outgoing edges to another subprocess and incoming edges from another subprocess B to this subprocess A; the two intersect and exist in parallel.

[0080] The rule of the loop operator is that an activity starts from sub-process A, reaches another sub-process B, and then returns to A from B, which can be summarized as starting here and ending here. Assume there is an event log L={<a,b,c> ,<a,c,b> ,<a,d,e> ,<a,d,e,f,d,e>}

[0081] 1. Convert the event log into one that directly follows the activity graph, such as... Figure 3 As shown.

[0082] 2. Use four splitting operators to split the directly following graph. The splitting process is as follows:

[0083] Figure 4 In the diagram, the partitioning operations performed on a, b, c, and d in sequence are sequential partitioning, exclusive partitioning, concurrent partitioning, and iterative partitioning, which can be represented in procedural language as follows:

[0084] L={<a,b,c> ,<a,c,b> ,<a,d,e> ,<a,d,e,f,d,e> The operations performed in sequence are as follows:

[0085] SEQUENCESPLIT(L,({a},{b,c,d,e,f}))=[L1={ },L2={<b,c> ,<c,b> ,<d,e> ,<d,e,f,d,e>}];

[0086] EXCLUSIVECHOICESPLIT(L2,({b,c},{d,e,f}))={L3=<b,c> ,<c,b>},{L4=<d,e> ,<d,e,f,d,e>}];

[0087] PARALLELSPLIT(L3,({b},{c}))={ },{ <c>};

[0088] LOOPSPLIT(L4,({d,e},{f}))={<d,e>},{ <f>}

[0089] 3. Based on the above process, the process can be converted into a procedure tree as follows:

[0090] Discovery model: M=→(a,×(∧(b,c),Q(→(d,e),f))).

[0091] 4. Convert the process tree into a Petri net:

[0092] Infrequent inductive mining is introduced by adding behavior filters in all steps of inductive mining; the frequency of trajectories and events is ignored by inductive mining but is taken into account by infrequent inductive mining to distinguish between frequent and infrequent behaviors; the parameter K represents a user-defined threshold between 0 and 1 used to distinguish between frequent and infrequent behaviors.

[0093] II. Filtering and Cutting Selection Steps on Operators:

[0094] 1. Heuristic filtering:

[0095] L1=[<a,b,c,a,b,e,> 50,<a,b,f,e> 100,<d,e,f> 100,<d,f,e> 100,<d,e,d,f> 1).

[0096] Infrequent inductive mining only includes the direct follower graph of the most frequent edges; a node that is relatively infrequent compared to other output edges will be filtered out; if the frequency of an output edge of a node is less than k times the frequency of the strongest output edge of that node, then the output edge of that node is too infrequent; all infrequent edges are filtered out before cutting ×, → and looping.

[0097] 2. Final Follower Relationship Diagram:

[0098] L2=[<a,c,d,e,b> ,<a,b,a,e,d,c> ,<a,e,c,b,d> ,<a,d,b,c,e> ].

[0099] If all output edges of a node have a frequency of 1, no value of k can filter the edges; if the final follow graph is used, the edges can be effectively filtered out; similar to weak order relations, infrequent inductive mining uses the final follow graph, which is the transitive closure of the direct follow relation: the edge (a, b) exists if and only if a is followed by b somewhere in the log.

[0100] III. Filter Implementation Steps in Basic Cases:

[0101] 1. Single activity:

[0102] L1=[〈ε〉 100 ,〈a〉 100 ,〈a,a〉 100 ,〈a,a,a〉 100 ];

[0103] L2=[〈ε〉 1 ,〈a〉 100 ,〈a,a〉 10 ,〈a,a,a〉 1 ];

[0104] The segmented subprocesses L1 and L2 reenact a process model. In L1, all trajectories are frequent, and a flower pattern model is used. Infrequent inductive mining will only discover 'a' when the average occurrence of 'a' in each trajectory of the log is close enough to 1 (depending on the relative threshold k).

[0105] 2. Empty trajectory:

[0106] L=[〈a,b,c〉 100 <a,c,d> 100 [,〈a,d〉];

[0107] L1=[〈a〉 201 ];

[0108] L2=[〈ε〉 1 ,〈b〉 100 ,〈c〉 100 ];

[0109] L3=[〈d〉 201 ];

[0110] For L2:×(τ,...);

[0111] Event log L is divided into three sub-logs L1, L2, and L3 by the splitting operator. There is an empty trajectory in sub-log L2, which has a much lower frequency than the other trajectories. To improve the accuracy of the model, filtering is required.

[0112] Model platform unit 2 is used to classify and combine the constructed process models and manage the combination results.

[0113] In this optional embodiment, the model platform unit 2 includes:

[0114] The model marketplace module is used to classify process models according to different business domains of an enterprise and build corresponding domain models.

[0115] The model combination module is used to analyze the industry to which the models in each field belong, and to combine the field models within the same industry to construct an industry model;

[0116] The model management module is used to manage domain models and industry models.

[0117] It should be further explained that the Model Mart module contains process models for specific sub-fields, such as financial process models; the Model Combination module combines domain models in the Model Mart to form industry models, analyzes the industry to which the domain models belong, and combines models that belong to a certain industry, such as combining bank deposit business processes, loan business processes, and intermediary business processes into a bank process model; the Model Management module manages the various models, including model updates, model combinations, model decomposition, and model deletion.

[0118] Application layer unit 3 is used to connect with various business systems of the enterprise and automate the execution of business processes.

[0119] In this optional embodiment, the application layer unit 3 includes:

[0120] Customized modules are used to connect with various business systems of enterprises and can be customized according to the needs of different industries;

[0121] The automation module is used to automate business processes using robotic process automation (RPA) technology and to obtain handles to desktop form controls and web page elements.

[0122] It should be further explained that Application Layer Unit 3 interfaces with Enterprise Resource Planning (ERP) systems, financial systems, and Customer Relationship Management (CRM) systems; it also provides customized development for different industries, such as workshop management systems for manufacturing, counter systems and core systems for banking, and market information systems for the securities industry; the interface operations of each system are changed from manual operation to Robotic Process Automation (RPA) operation, with RPA obtaining handles of application controls on the operating system desktop form and web page elements for automated operation.

[0123] Furthermore, this invention improves the efficiency of enterprise business process execution by applying automation and AI technologies to process mining.

[0124] 1. Apply automation technology during process mining and optimization:

[0125] Robotic Process Automation (RPA) is a business process automation technology based on software robots. During process discovery, some tedious processes are repetitive and require a lot of time to handle manually. RPA can be applied to complete a series of processes that previously required manual processing based on pre-configured automatic execution scripts, thereby automating business processes.

[0126] RPA automates business processes by simulating user workflow actions. RPA robots require that the business processes they execute must have clear business process rules. By configuring clear and fixed rules into preset execution scripts, repetitive human business operations, such as report data entry, system data auditing, invoice verification, and supplier auditing, can all be completed by configuring RPA to accomplish these repetitive and low-value tasks.

[0127] Functionally, RPA is a program that handles repetitive tasks and simulates manual operations, and can achieve the following five main functions:

[0128] (1) Data retrieval and recording: RPA can perform data retrieval, data migration and data entry across systems;

[0129] (2) Image recognition and processing: Information is recognized through optical character recognition (OCR) technology, and text can be reviewed and analyzed on this basis;

[0130] (3) The platform can upload and download data according to the pre-designed path, and complete the automatic reception and output of data streams;

[0131] (4) Data processing and analysis, including data inspection, data screening, data calculation, data organization, and data verification;

[0132] (5) Information monitoring and output: RPA can simulate human judgment to achieve functions such as workflow allocation, standard report generation, decision-making based on clear rules, and automatic information notification.

[0133] 2. AI learning ability:

[0134] Utilizing artificial intelligence (AI) technology during process mining can improve process efficiency; for example, OCR technology can be used to recognize contract texts and invoices, converting unstructured data into structured data, thereby improving mining efficiency; in the telephone customer service process of banks, intelligent voice recognition technology can be used to identify the questions or requests expressed by customers over the phone, and robot customer service can automatically process and respond according to the question classification, saving manpower.

[0135] 3. Infrequent inductive mining algorithm:

[0136] By introducing infrequent factor analysis into the inductive mining algorithm, focusing on frequent processes in enterprise business, assigning higher path weights, simplifying flowcharts, and reducing algorithm time, the efficiency of process mining is improved.

[0137] like< / f> < / c> Figure 2 As shown, this invention adopts a layered, loosely coupled architecture design. In a specific embodiment, the AI ​​robot process mining system architecture is divided into 7 layers, from top to bottom: application system layer, model platform, model development, data platform, data center, data acquisition, and data source. The data source is the enterprise data source. The data acquisition layer collects data from the data source and uploads it to the data center layer. The data platform performs preliminary data analysis and processing before uploading it to the data platform. The data platform performs further analysis and processing on the data. The model development uses algorithms to perform process mining on the data. The model platform builds model combinations and model marts. The application system layer connects to the enterprise application system through the model platform.

[0138] In summary, by utilizing the above-mentioned technical solutions of this invention, the integration of automation technology and artificial intelligence technology into process mining can improve the efficiency of process identification and execution; RPA can replace manual operations to automate repetitive tasks; and by combining infrequent inductive algorithms, flowcharts can be simplified and computational complexity reduced, thereby further improving the efficiency and intelligence level of process mining.

[0139] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An artificial intelligence robot process mining system, characterized in that, The system includes: The data and model unit is used to collect data and perform multiple analysis and processing steps, combining process mining algorithms and tools to build process models; The model platform unit is used to classify and combine the constructed process models, and manage the combination results. Application layer units are used to interface with various enterprise business systems and automate business processes.

2. The artificial intelligence robot process mining system according to claim 1, characterized in that, The data and model unit includes: The data source module is used to store structured data, unstructured data, and external data. The data acquisition module is used to collect data from the data source module and upload it to the data center module through extraction, transformation and loading operations; The data center module is used to store the collected data, perform initial analysis and processing on the data, and upload the results of the initial analysis and processing to the data platform module. The data middle platform module is used to receive the initial analysis and processing results uploaded by the data center module and perform secondary analysis and processing. The model development module is used to model processes based on the results of secondary analysis and processing, using process mining algorithms and tools to build process models suitable for enterprise business.

3. The artificial intelligence robot process mining system according to claim 2, characterized in that, The data center module includes: The data warehouse submodule is used to receive and store data uploaded by the data acquisition module; The database submodule is used for mining and analyzing the data stored in the data warehouse submodule; The data processing submodule is used to perform initial analysis and processing of the mining and analysis results using big data processing tools.

4. The artificial intelligence robot process mining system according to claim 3, characterized in that, The data platform module includes: The wide table submodule is used to receive the initial analysis and processing results uploaded by the data center module, and to build a database table by integrating the multi-dimensional data fields in the initial analysis and processing results; The Data Mart submodule is used to analyze target indicators based on data in the database tables using a pre-defined data analysis model. The data element submodule is used to define the data elements of the process model based on the analysis results.

5. The artificial intelligence robot process mining system according to claim 4, characterized in that, The data elements include: model tag library, processing logic, data assets, and data structure.

6. The artificial intelligence robot process mining system according to claim 5, characterized in that, The semantics of the data elements are defined by three elements: object class, characteristics, and representation.

7. The artificial intelligence robot process mining system according to claim 6, characterized in that, The model development module includes: Tool sub-molds are used to deploy and execute process mining algorithms using programming languages ​​and machine learning tools; The algorithm submodule is used to model the secondary analysis results using several process mining algorithms, and to build a process model suitable for enterprise business.

8. The artificial intelligence robot process mining system according to claim 7, characterized in that, The process mining algorithms include: infrequent inductive mining algorithm, α-series algorithm, heuristic algorithm, genetic algorithm and log classification algorithm.

9. The artificial intelligence robot process mining system according to claim 1, characterized in that, The model platform unit includes: The Model Mart module is used to classify process models according to different business domains of an enterprise and build corresponding domain models. The model combination module is used to analyze the industry to which the models in each field belong, and to combine the field models within the same industry to construct an industry model; The model management module is used to manage domain models and industry models.

10. The artificial intelligence robot process mining system according to claim 1, characterized in that, The application layer unit includes: Customized modules are used to connect with various business systems of enterprises and can be customized according to the needs of different industries; The automation module is used to automate business processes using robotic process automation (RPA) technology and to obtain handles to desktop form controls and web page elements.

Images