A Business Process Visualization Method and System

Abstract

The present invention belongs to the technical field of process management, and discloses a business process visualization method and system. The method includes the following steps: constructing a business process expert system, a process link analysis model, and a business process analysis model; using the process link analysis model to perform process link analysis on real-time business process data; using the business process analysis model to perform business process analysis on a real-time predicted business process relationship diagram; using the business process expert system to optimize the real-time predicted business process information to generate an optimized real-time predicted business process relationship diagram; performing chart conversion on several real-time process link data corresponding to the optimized real-time predicted business process information, adding several real-time data charts and the analysis results of the real-time process link data to the optimized real-time predicted business process relationship diagram, and performing visualization. The present invention solves the problems of low efficiency, poor accuracy, and simple functions existing in the prior art.

Description

Technical Field

[0001] The present invention belongs to the technical field of process management, and particularly relates to a business process visualization method and system. Background Art

[0002] With the expansion of enterprise scale and the increase in business complexity, business process management has become increasingly important. A complete and clear business process architecture can not only improve the work efficiency of employees, but also increase the profits of enterprises and achieve systematic and standardized management of enterprises. However, traditional business process management often relies on manual analysis and drawing, with low efficiency, poor accuracy, and easy to make mistakes, resulting in omissions in enterprise projects, and seriously affecting the normal operation of enterprises, bringing significant losses to enterprises. Moreover, the functions of traditional business process management are simple and cannot analyze business process data, thus unable to discover potential business process problems and defects. Therefore, how to use advanced technical means to achieve automatic identification, analysis, and visual display of business processes has become an urgent problem to be solved in the current business process management field. Summary of the Invention

[0003] In order to solve the problems of low efficiency, poor accuracy, and simple functions existing in the prior art, the purpose of the present invention is to provide a business process visualization method and system.

[0004] The technical solution adopted by the present invention is as follows:

[0005] A business process visualization method includes the following steps:

[0006] According to the historical business process information of several enterprises, use the reinforcement learning algorithm to construct a business process expert system, and according to the historical business process data of several enterprises, use the deep learning algorithm to construct corresponding process link analysis models and business process analysis models;

[0007] Collect the real-time business process data of an enterprise, use the process link analysis model to perform process link analysis on the real-time business process data of the enterprise, and obtain the real-time predicted business process information of the enterprise and the real-time process link analysis results corresponding to each real-time process link data in the real-time business process data;

[0008] Generate a real-time predicted business process relationship diagram of the enterprise according to the real-time predicted business process information, and use the business process analysis model to perform business process analysis on the real-time predicted business process relationship diagram to obtain corresponding real-time business process analysis results;

[0009] According to the real-time business process analysis results, use the business process expert system to optimize the real-time prediction business process information, and obtain the corresponding optimized real-time prediction business process information. According to the optimized real-time prediction business process information, generate the optimized real-time prediction business process relationship diagram of the enterprise;

[0010] Perform chart conversion on the data of several real-time process links corresponding to the optimized real-time prediction business process information to obtain the corresponding real-time data charts. Add the several real-time data charts and the analysis results of the real-time process link data to the optimized real-time prediction business process relationship diagram and perform visualization.

[0011] Furthermore, according to the historical business process information of several enterprises, use the reinforcement learning algorithm to construct a business process expert system, and according to the historical business process data of several enterprises, use the deep learning algorithm to construct a process link analysis model and a business process analysis model, including the following steps:

[0012] Collect several pieces of business process knowledge, as well as the historical business process information and historical business process data of several enterprises, and perform preprocessing to obtain several preprocessed business process knowledge, several preprocessed historical business process information, and several preprocessed historical business process data;

[0013] According to several preprocessed business process knowledge and several preprocessed historical business process information, use the natural language processing algorithm and the reinforcement learning algorithm to construct a business process expert system;

[0014] According to the historical business process data of several enterprises, use the deep learning algorithm to construct a process link analysis model and generate several historical prediction business process information;

[0015] According to several historical prediction business process information, use the deep learning algorithm to construct a business process analysis model.

[0016] Furthermore, according to several preprocessed business process knowledge and several preprocessed historical business process information, use the natural language processing algorithm and the reinforcement learning algorithm to construct a business process expert system, including the following steps:

[0017] According to the preprocessed historical business process information, define the simulation environment, state space, action space, and reward function of the reinforcement learning algorithm;

[0018] Based on the simulation environment, state space, action space, and reward function, according to several preprocessed historical business process information, use the reinforcement learning algorithm to construct the inference engine of the business process expert system;

[0019] Based on the business process knowledge after several preprocessings, using natural language processing algorithms, construct a named entity extraction model and an entity relationship extraction model, generate the knowledge base of the business process expert system, and integrate the knowledge base and the inference engine to construct the business process expert system.

[0020] Furthermore, the reinforcement learning algorithm is the DQN algorithm.

[0021] Furthermore, according to the historical business process data of several enterprises, using deep learning algorithms, construct a process link analysis model, and generate several historical predicted business process information, including the following steps:

[0022] Perform data parsing on the preprocessed historical business process data to obtain several corresponding historical process link data;

[0023] Add labels to each piece of historical process link data to obtain several pieces of historical process link data with real process link information labels and real process link data analysis labels;

[0024] According to several pieces of historical process link data, use deep learning algorithms to construct a process link analysis model, and generate several historical predicted process link information labels;

[0025] Integrate several historical predicted process link information labels corresponding to the same preprocessed historical business process data to obtain the corresponding historical predicted business process information.

[0026] Furthermore, the process link analysis model is constructed based on the RF-BiLSTM algorithm.

[0027] Furthermore, according to several pieces of historical predicted business process information, use deep learning algorithms to construct a business process analysis model, including the following steps:

[0028] According to the historical predicted business process information, use the named entity extraction model and the entity relationship extraction model to extract several historical business process named entities and the corresponding several historical business process entity relationships;

[0029] According to several historical business process named entities and several historical business process entity relationships, generate the corresponding historical predicted business process relationship diagram, and add the corresponding real business process analysis label to each historical predicted business process relationship diagram;

[0030] According to several historical predicted business process relationship diagrams with real business process analysis labels, use deep learning algorithms to construct a business process analysis model.

[0031] Furthermore, the business process analysis model is constructed based on the GNN algorithm.

[0032] Further, according to the real-time business process analysis results, use the business process expert system to optimize the real-time prediction business process information, and obtain the corresponding optimized real-time prediction business process information. According to the optimized real-time prediction business process information, generate the optimized real-time prediction business process relationship diagram of the enterprise, including the following steps:

[0033] According to the real-time business process analysis results, retrieve the corresponding real-time business process optimization strategy from the knowledge base of the business process expert system;

[0034] According to the real-time business process optimization strategy, update the action space of the inference engine of the business process expert system to obtain the updated action space;

[0035] According to the real-time prediction business process information, update the state space of the inference engine of the business process expert system to obtain the updated state space;

[0036] Based on the updated action space, updated state space, and reward function, generate optimized business process information to obtain the corresponding optimized real-time prediction business process information;

[0037] According to the optimized real-time prediction business process information, generate the optimized real-time prediction business process relationship diagram of the enterprise.

[0038] A business process visualization system for implementing the business process visualization method, the system includes a model construction unit, a process link analysis unit, a business process analysis unit, a business process optimization unit, and a business process visualization unit that are connected in sequence.

[0039] The beneficial effects of the present invention are:

[0040] The present invention discloses a business process visualization method and system, which uses a deep learning algorithm to construct a process link analysis model to automatically identify and analyze each process link in the business process data, improving the efficiency and accuracy of business process management, reducing the input of labor costs, being able to accurately discover potential problems in each process link of the business process, and improving functionality and practicality; using a deep learning algorithm to construct a business process analysis model, being able to discover defects and problems in the business process, and using a business process expert system to optimize the business process, improving the intelligent level of business process management; visually displaying the business process relationship diagram, the analysis results of process link data, and the data charts of each process link, improving the convenience of business process management.

[0041] Other beneficial effects of the present invention will be further described in the specific implementation manner. Description of the Drawings

[0042] Figure 1It is a flowchart of the business process visualization method in the present invention.

[0043] Figure 2 It is a structural block diagram of the business process visualization system in the present invention. Detailed implementation manners

[0044] The present invention will be further explained below in conjunction with the accompanying drawings and specific embodiments.

[0045] Embodiment 1:

[0046] As Figure 1 shown, this embodiment provides a business process visualization method, including the following steps:

[0047] S1: According to the historical business process information of several enterprises, use the reinforcement learning algorithm to construct a business process expert system, and according to the historical business process data of several enterprises, use the deep learning algorithm to construct the corresponding process link analysis model and business process analysis model, including the following steps:

[0048] S1-1: Collect a number of business process knowledge, as well as the historical business process information and historical business process data of several enterprises, and perform preprocessing to obtain a number of preprocessed business process knowledge, a number of preprocessed historical business process information, and a number of preprocessed historical business process data;

[0049] S1-2: According to a number of preprocessed business process knowledge and a number of preprocessed historical business process information, use the natural language processing algorithm and the reinforcement learning algorithm to construct a business process expert system, including the following steps:

[0050] S1-2-1: According to the preprocessed historical business process information, define the simulation environment, state space, action space, and reward function of the DQN algorithm;

[0051] S1-2-2: Based on the simulation environment, state space S = [s1,..., s i ,..., s I , action space A = [a1,..., a j ,..., a I' , and the reward function, where s i is the i-th state value, i is the state indicator, I is the size of the state space, a j is the j-th action value, j is the action indicator, I' is the size of the action space, according to a number of preprocessed historical business process information, use the Deep Q-network (DQN) algorithm to construct the inference engine of the business process expert system;

[0052] The DQN algorithm includes an agent, a deep Q-network, and an experience replay buffer. The agent is optimized and trained using preprocessed business process knowledge, and the generated experiences are stored in the experience replay buffer. The DQN algorithm enables the intelligent agent to learn the optimal strategy during the interaction with the environment to achieve a certain goal. Applying reinforcement learning to the inference mechanism construction of an expert system can make the expert system more intelligent in processing business process information and automatically adjust and optimize the inference process.

[0053] S1-2-3: According to a number of preprocessed business process knowledge, use natural language processing algorithms to construct a named entity extraction model and an entity relationship extraction model, generate the knowledge base of the business process expert system, and integrate the knowledge base and the inference engine to construct the business process expert system.

[0054] The named entity extraction model is constructed based on the BERT-LSTM-GAT algorithm, and the named entity extraction model includes a first input layer, a semantic feature extraction layer constructed based on the BERT pre-trained language sub-model and the Long Short-Term Memory (LSTM) algorithm, a graph feature extraction module layer constructed based on the Graph Attention Network (GAT) algorithm, a feature interaction and collaboration layer, a linear-chain Conditional Random Field (CRF) layer, and a first output layer.

[0055] The BERT pre-trained language sub-model converts the input data into word vectors. The sequence semantic feature extraction layer extracts the semantic features of the word vectors. The GAT network constructs a graph based on semantic similarity. The graph feature extraction layer extracts the graph features in the graph. The feature interaction and collaboration layer performs interaction and collaboration on the semantic features and graph features to obtain an interaction and collaboration feature sequence, and inputs the interaction and collaboration feature sequence into the CRF module to complete the annotation of named entities.

[0056] The entity relationship extraction model is constructed based on the BERT-LSTM-K-MAX algorithm, and the entity relationship extraction model includes a second input layer, a word representation layer constructed based on the BERT pre-trained language sub-model, an embedding layer, a masking layer, a capsule network layer, a named entity processing layer constructed based on the BILSTM algorithm, a dynamic pooling layer constructed based on the K-MAX pooling algorithm, a classification layer, and a second output layer.

[0057] The BERT pre-trained language sub-model in the word representation layer can extract the word representation of the input data more accurately. The word representation enters the embedding layer. The sparse self-attention mechanism in the embedding layer improves the calculation of effective weights by reducing the calculation of invalid weights to avoid the transmission of invalid attention weights, reducing the time complexity and space complexity. The capsule network layer is used to obtain the global position information and comprehensive feature information, that is, the position information of the input data. The word representation is put into the dynamic routing to obtain the comprehensive feature representation of the input data, and finally the core feature vector of the input data is obtained. Then, the LSTM network is used to process the named entities extracted by the named entity extraction model from the input data. The dynamic pooling layer uses the dynamic K-MAX pooling algorithm for the final screening work. Finally, the classification layer completes the conversion from the feature vector to the probability of the predicted preliminary answer, and the second output layer outputs the final entity relationship;

[0058] S1-3: According to the historical business process data of several enterprises, use deep learning algorithms to construct a process link analysis model and generate several historical predicted business process information, including the following steps:

[0059] S1-3-1: Parse the preprocessed historical business process data to obtain corresponding historical process link data;

[0060] S1-3-2: Add labels to each piece of historical process link data to obtain several pieces of historical process link data with true process link information labels and true process link data analysis labels;

[0061] S1-3-3: According to several pieces of historical process link data, use the Random Forest (RF)-Bidirectional Long Short-Term Memory (BiLSTM) algorithm to construct a process link analysis model and generate several historical predicted process link information labels;

[0062] S1-3-4: Integrate several historical predicted process link information labels corresponding to the same preprocessed historical business process data to obtain the corresponding historical predicted business process information;

[0063] S1-4: According to several pieces of historical predicted business process information, use deep learning algorithms to construct a business process analysis model, including the following steps:

[0064] S1-4-1: According to the historical predicted business process information, use a named entity extraction model and an entity relationship extraction model to extract several historical business process named entities and corresponding historical business process entity relationships;

[0065] S1-4-2: Generate corresponding historical predicted business process relationship diagrams based on a number of historical business process named entities and a number of historical business process entity relationships, and add corresponding real business process analysis labels to each historical predicted business process relationship diagram;

[0066] S1-4-3: Based on a number of historical predicted business process relationship diagrams with real business process analysis labels set, use the Graph Neural Network (GNN) algorithm to construct a business process analysis model;

[0067] The GNN algorithm is a powerful machine learning method specifically designed to process graph-structured data and can effectively capture complex dependencies and structural information in flowcharts, which is crucial for understanding the interactions between nodes in the graph;

[0068] S2: Collect the real-time business process data of the enterprise, use the process link analysis model to conduct process link analysis on the real-time business process data of the enterprise, and obtain the real-time predicted business process information of the enterprise and the real-time process link data analysis results corresponding to each real-time process link data in the real-time business process data, including the following steps:

[0069] S2-1: Collect the real-time business process data of the enterprise, perform data parsing on the real-time business process data, and obtain the corresponding number of real-time process link data;

[0070] The real-time business process data includes various data during the enterprise's business activities and all links of the internal circulation and processing of these data. The process links include customer information management, marketing and sales, product design and production, supply chain management, financial management, human resource management, internal communication and collaboration within the enterprise, and research and development, etc., covering the business process data in all stages from R & D, sales, management to after-sales of business projects;

[0071] S2-2: Convert the real-time process link data into a corresponding real-time process link data matrix, and input the real-time process link data matrix into the process link analysis model;

[0072] S2-3: Use the process link analysis model to conduct process link analysis on the real-time process link data matrix, and obtain the corresponding real-time predicted process link information and the corresponding real-time process link data analysis results, including the following steps:

[0073] S2-3-1: Use the trained RF structure in the process link analysis model to extract the feature contribution degrees of several real-time alternative key features in the real-time process link data matrix. The formula is:

[0074]

[0075] In the formula, is the feature contribution degree of the j'-th real-time alternative key feature; is the feature contribution degree of the j'-th real-time alternative key feature in the i'-th Classification And Regression Tree (CART); i' is the CART tree indicator; j' is the real-time alternative key feature indicator; n' is the total number of CART trees;

[0076]

[0077] In the formula, GI m , GI l' , GI r are the Gini indices of CART tree nodes m, l, and r; p mk' is the proportion of class k' in CART tree node m; K' is the total number of classes; m, l, r are node indicators; k' is the class indicator; M is the total number of CART tree nodes in the random forest;

[0078] S2-3-2: Normalize the feature contribution degrees of several real-time alternative key features to obtain the corresponding several normalized feature contribution degrees;

[0079] The formula is:

[0080]

[0081] In the formula, VIM j' is the normalized feature contribution degree; J is the total number of real-time alternative key features;

[0082] S2-3-3: Generate the feature selection standard values of several real-time alternative key features according to the normalized feature contribution degrees;

[0083] The formula is:

[0084]

[0085] In the formula, CFC j' is the feature selection standard value of the j'-th real-time alternative key feature; VIM j" is the normalized feature contribution degree of the j''-th real-time alternative key feature; j'' is the real-time alternative key feature indicator;

[0086] S2-3-4: Rank several real-time alternative key features in descending order according to the feature selection standard values, and select the first M real-time alternative key features as the real-time key features of the real-time process link data matrix;

[0087] S2-3-5: Based on M real-time key features, conduct process link analysis and prediction to obtain the corresponding real-time predicted process link information and the corresponding real-time process link data analysis results;

[0088] The real-time process link data analysis results include defects, problems, classification levels, evaluation conclusions, or other preset analysis situations existing in the real-time process link data;

[0089] S2-4: Traverse all real-time process link data, and integrate all the obtained real-time predicted process link information to obtain the enterprise's real-time predicted business process information;

[0090] S3: Based on the real-time predicted business process information, generate a real-time predicted business process relationship diagram of the enterprise, and use a business process analysis model to conduct business process analysis on the real-time predicted business process relationship diagram to obtain the corresponding real-time business process analysis results, including the following steps:

[0091] S3-1: Based on the real-time predicted business process information, use a named entity extraction model and an entity relationship extraction model to extract several real-time business process named entities and the corresponding several real-time business process entity relationships;

[0092] S3-2: Use the real-time business process named entities as nodes, and based on the real-time business process entity relationships as the connection lines between the nodes, generate a real-time predicted business process relationship diagram of the enterprise;

[0093] S3-3: Input the real-time predicted business process relationship diagram into the business process analysis model, and use the business process analysis model to extract node features, connection line features, and graph features;

[0094] S3-4: Based on the node features, connection line features, and graph features, conduct business process analysis to obtain the corresponding real-time business process analysis results;

[0095] S4: Based on the real-time business process analysis results, use a business process expert system to optimize the real-time predicted business process information to obtain the corresponding optimized real-time predicted business process information, and based on the optimized real-time predicted business process information, generate an optimized real-time predicted business process relationship diagram of the enterprise, including the following steps:

[0096] S4-1: Based on the real-time business process analysis results, retrieve the corresponding real-time business process optimization strategy from the knowledge base of the business process expert system;

[0097] S4-2: Based on the real-time business process optimization strategy, update the action space of the inference engine of the business process expert system to obtain the updated action space;

[0098] S4-3: Update the state space of the inference engine of the business process expert system according to the real-time prediction business process information to obtain an updated state space;

[0099] S4-4: Based on the updated action space A' = [a'1,..., a' j" ,..., a' I' , the updated state space S' = [s'1,..., s' i" ,..., s' I and the reward function, where s' i" is the i-th updated state value, i is the state indicator, and a' j" is the j-th updated action value, j is the action indicator, perform optimization of the business process information generation to obtain the corresponding optimized real-time prediction business process information, including the following steps:

[0100] S4-4-1: Based on the updated action space A' = [a'1,..., a' j" ,..., a' I' , the updated state space S' = [s'1,..., s' i" ,..., s' I and the reward function, use the Q function of the deep Q network to obtain the predicted Q values of possible actions;

[0101] S4-4-2: Use the agency to take the possible action corresponding to the highest predicted Q value as the execution action, and according to the reward function, obtain the reward value R'(s p' , a p' , s' p' );

[0102] S4-4-3: Update the predicted Q values of possible actions according to the reward value of the execution action to obtain the updated Q values of possible actions, and repeat the previous step until the iteration number threshold is reached;

[0103] The formula is:

[0104] Q(s' p' , a' p' ) = (1 - α)·Q(s p' , a p' ) + α·(R(s p' , a p' , s' p' ) + γ·Q max (s p' , a p' ))

[0105] In the formula, Q(s' p' , a' p') is the updated state value s' p' and the updated action value a' p' corresponding to the updated Q value; Q(s p' , a p' ) is the state value s p' and the action value a p' corresponding to the predicted Q value; α is the learning rate; Q max (s p' , a p' ) is the highest predicted Q value;

[0106] S4-4-4: According to the updated Q value, using the greedy strategy, output the execution action corresponding to the highest updated Q value as the real-time optimization strategy;

[0107] S4-4-5: Optimize the real-time predicted business process information according to the real-time optimization strategy to obtain the corresponding optimized real-time predicted business process information;

[0108] S4-5: Generate an optimized real-time predicted business process relationship diagram for the enterprise according to the optimized real-time predicted business process information;

[0109] S5: Perform chart conversion on the data of several real-time process links corresponding to the optimized real-time predicted business process information to obtain the corresponding real-time data charts, add the several real-time data charts and the analysis results of the real-time process link data to the optimized real-time predicted business process relationship diagram, and perform visualization;

[0110] The chart conversion provides diversified data visualization methods, including line charts, bar charts, pie charts, etc., as well as interactive data chart displays. Users can customize the data display methods and dimensions according to their needs to deeply understand the operation of business process data.

[0111] Embodiment 2:

[0112] As Figure 2 shown, this embodiment provides a business process visualization system for implementing the business process visualization method. The system includes a model construction unit, a process link analysis unit, a business process analysis unit, a business process optimization unit, and a business process visualization unit that are connected in sequence;

[0113] The model construction unit is used to construct a business process expert system using the reinforcement learning algorithm according to the historical business process information of several enterprises, and construct a corresponding process link analysis model and business process analysis model using the deep learning algorithm according to the historical business process data of several enterprises;

[0114] A process step analysis unit, configured to collect real-time business process data of an enterprise, and use a process step analysis model to perform process step analysis on the real-time business process data of the enterprise, so as to obtain real-time predicted business process information of the enterprise and real-time process step analysis results corresponding to each real-time process step data in the real-time business process data;

[0115] A business process analysis unit, configured to generate a real-time predicted business process relationship diagram of the enterprise according to the real-time predicted business process information, and use a business process analysis model to perform business process analysis on the real-time predicted business process relationship diagram, so as to obtain corresponding real-time business process analysis results;

[0116] A business process optimization unit, configured to optimize the real-time predicted business process information according to the real-time business process analysis results by using a business process expert system, so as to obtain corresponding optimized real-time predicted business process information, and generate an optimized real-time predicted business process relationship diagram of the enterprise according to the optimized real-time predicted business process information;

[0117] A business process visualization unit, configured to perform chart conversion on several real-time process step data corresponding to the optimized real-time predicted business process information to obtain corresponding real-time data charts, add the several real-time data charts and the real-time process step analysis results to the optimized real-time predicted business process relationship diagram, and perform visualization.

[0118] The present invention discloses a business process visualization method and system. By using a deep learning algorithm to construct a process step analysis model, automatic identification and analysis are performed on each process step in the business process data, which improves the efficiency and accuracy of business process management, reduces the input of labor costs, can accurately discover potential problems in each process step of the business process, and improves the functionality and practicability; by using a deep learning algorithm to construct a business process analysis model, defects and problems in the business process can be discovered, and a business process expert system is used for business process optimization, which improves the intelligent level of business process management; the business process relationship diagram, the real-time process step analysis results, and the data charts of each process step are visually displayed, which improves the convenience of business process management.

[0119] The present invention is not limited to the above optional implementation manners, and any person can obtain other various forms of products under the inspiration of the present invention. The above specific implementation manners should not be construed as limiting the protection scope of the present invention. The protection scope of the present invention should be defined by the claims, and the description can be used to interpret the claims.

Claims

1. A business process visualization method, characterized in that: The steps include: Based on the historical business process information of several companies, a reinforcement learning algorithm is used to build a business process expert system. Based on the historical business process data of several companies, a deep learning algorithm is used to build corresponding process link analysis models and business process analysis models, including the following steps: Collecting some business process knowledge, as well as some historical business process information and historical business process data of some enterprises, and preprocessing them to obtain some preprocessed business process knowledge, some preprocessed historical business process information, and some preprocessed historical business process data; Based on some pre-processed business process knowledge and some pre-processed historical business process information, a business process expert system is constructed using a natural language processing algorithm and a reinforcement learning algorithm, including the following steps: Based on the preprocessed historical business process information, define the simulation environment, state space, action space and reward function of the reinforcement learning algorithm; Based on the simulation environment, state space, action space and reward function, and according to some pre-processed historical business process information, a reinforcement learning algorithm is used to build the inference engine of the business process expert system; Based on some pre-processed business process knowledge, natural language processing algorithms are used to build named entity extraction models and entity relationship extraction models, generate the knowledge base of the business process expert system, and integrate the knowledge base and reasoning engine to build a business process expert system; Based on the historical business process data of several companies, we use deep learning algorithms to build a process analysis model and generate some historical forecast business process information; Based on some historical forecast business process information, a business process analysis model is constructed using deep learning algorithms; Collect the real-time business process data of the enterprise, use the process link analysis model to perform process link analysis on the real-time business process data of the enterprise, and obtain the real-time predicted business process information of the enterprise and the real-time process link data analysis results corresponding to each real-time process link data in the real-time business process data; Generate a real-time prediction business process relationship diagram for the enterprise based on the real-time prediction business process information, and use the business process analysis model to perform business process analysis on the real-time prediction business process relationship diagram to obtain the corresponding real-time business process analysis results; According to the real-time business process analysis results, the business process expert system is used to optimize the real-time predicted business process information to obtain the corresponding optimized real-time predicted business process information. According to the optimized real-time predicted business process information, the optimized real-time predicted business process relationship diagram of the enterprise is generated; The real-time process link data corresponding to the optimized real-time prediction business process information are converted into graphs to obtain corresponding real-time data graphs, and the real-time data graphs and real-time process link data analysis results are added to the optimized real-time prediction business process relationship diagram and visualized.

2. A business process visualization method according to claim 1, characterized in that: The reinforcement learning algorithm is the DQN algorithm.

3. A business process visualization method according to claim 1, characterized in that: Based on the historical business process data of several companies, a deep learning algorithm is used to build a process link analysis model and generate some historical prediction business process information, including the following steps: Perform data analysis on the pre-processed historical business process data to obtain corresponding historical process link data; Label each historical process link data to obtain a number of historical process link data with real process link information labels and real process link data analysis labels; Based on several historical process link data, a deep learning algorithm is used to build a process link analysis model and generate several historical prediction process link information labels; Several historical prediction process link information labels corresponding to the same pre-processed historical business process data are integrated to obtain the corresponding historical prediction business process information.

4. A business process visualization method according to claim 3, characterized in that: The process link analysis model is constructed based on the RF-BiLSTM algorithm.

5. A business process visualization method according to claim 3, characterized in that: Based on some historical prediction business process information, a business process analysis model is constructed using a deep learning algorithm, including the following steps: Based on the historical predicted business process information, a named entity extraction model and an entity relationship extraction model are used to extract a number of historical business process named entities and a number of corresponding historical business process entity relationships; Generate a corresponding historical prediction business process relationship diagram based on a number of historical business process named entities and a number of historical business process entity relationships, and add a corresponding real business process analysis label to each historical prediction business process relationship diagram; Based on several historical prediction business process relationship diagrams with real business process analysis labels, a business process analysis model is constructed using a deep learning algorithm.

6. A business process visualization method according to claim 5, characterized in that: The business process analysis model is built based on the GNN algorithm.

7. A business process visualization method according to claim 5, characterized in that: According to the real-time business process analysis results, the business process expert system is used to optimize the real-time prediction business process information to obtain the corresponding optimized real-time prediction business process information. According to the optimized real-time prediction business process information, the optimized real-time prediction business process relationship diagram of the enterprise is generated, including the following steps: According to the real-time business process analysis results, the corresponding real-time business process optimization strategy is retrieved from the knowledge base of the business process expert system; According to the real-time business process optimization strategy, the action space of the inference engine of the business process expert system is updated to obtain an updated action space; According to the real-time predicted business process information, the state space of the inference engine of the business process expert system is updated to obtain an updated state space; Based on the updated action space, updated state space and reward function, optimize the business process information generation to obtain the corresponding optimized real-time predicted business process information; Based on the optimized real-time prediction business process information, generate the enterprise's optimized real-time prediction business process relationship diagram.

8. A business process visualization system, used to implement the business process visualization method according to any one of claims 1 to 7, characterized in that: The system comprises a model building unit, a process link analysis unit, a business process analysis unit, a business process optimization unit and a business process visualization unit which are connected in sequence.

Images