An AI-based enterprise business data processing system and method

Abstract

The present invention relates to the field of enterprise business management, and particularly to an AI-based enterprise business data processing system and method. The method includes: Step 1, obtaining source system data and storing the obtained data in one or more target systems; Step 2, constructing a data synchronization policy, and monitoring and managing data synchronization between target system databases based on the data synchronization policy, including: generating a data synchronization policy based on the data synchronization cost; quantifying the importance of data and data synchronization tasks, combining them with the data synchronization cost, and optimizing the data synchronization policy; adding adjustment rules to the optimized data synchronization policy to dynamically adjust the importance of data or tasks; Step 3, sending the data of one or more target systems to the cloud platform for storage.

Description

Technical Field

[0001] The present invention relates to the field of enterprise business management, and particularly relates to an AI-based enterprise business data processing system and method. Background Art

[0002] In today's digital age, enterprise business data, as the core asset of enterprise operation and development, covers a very wide range. Enterprise business data usually includes the basic data of the enterprise (such as the company's registered address, registered capital, business scope, organizational structure, etc.), asset information (such as detailed information of various assets such as the enterprise's fixed assets and current assets), business operation information (including customer information, customer behavior information), financial data (such as the enterprise's income, expenditure, profit, assets and liabilities, etc.), supply chain data, etc.

[0003] In enterprise business data processing, common methods include data collection and integration, data storage and management, data analysis and mining. Common enterprise business systems include enterprise resource planning (ERP) systems, customer relationship management (CRM) systems, and big data processing platforms such as Hadoop and Flink, etc.

[0004] In the process of enterprise business processing, data synchronization between different systems is an urgent problem to be solved. Since enterprises usually use multiple different business systems to support their respective business functions, these systems are often developed independently and use different data storage solutions. Summary of the Invention

[0005] The present invention realizes data synchronization between the source system (CRM) and different target systems (ERP, Hadoop) and data synchronization between different target systems by constructing a data synchronization strategy, and ensures data consistency during the data synchronization process.

[0006] The technical solution proposed by the present invention is: an AI-based enterprise business data processing method, the method comprising:

[0007] Step 1, obtain source system data, and store the obtained data in one or more target systems;

[0008] Step 2, construct a data synchronization strategy, and monitor and manage data synchronization between target system databases based on the data synchronization strategy, including:

[0009] Generate a data synchronization strategy based on the data synchronization cost;

[0010] After quantifying the importance of the data and the data synchronization task and combining it with the data synchronization cost, optimize the data synchronization strategy;

[0011] Add adjustment rules to the optimized data synchronization strategy to dynamically adjust the importance of data or tasks;

[0012] Step 3: Send the data of one or more target systems to the cloud platform for storage.

[0013] Preferably, the source system includes a customer relationship management system CRM and a business intelligence BI system, and the target system includes an enterprise resource planning ERP system and a big data processing Hadoop system;

[0014] Before obtaining the data of the source system and storing the obtained data in the target system database, it further includes:

[0015] Judge whether the target system database is connected to the source system;

[0016] Judge the consistency of the data synchronized from the source system to different target systems;

[0017] The judgment of whether the target system database is connected to the source system includes the following steps:

[0018] Modify one or more fields in the source system, and detect whether the corresponding fields in the target system change. If so, go to step 2; otherwise, go to the following steps:

[0019] Modify one or more fields in the target system, and check whether the corresponding fields in the source system are synchronized automatically. If so, go to step 2; otherwise, go to the following steps:

[0020] Sample and export the data of the target system and the source system for comparison, and judge whether they are consistent. If so, go to step 2; otherwise, go to the following steps:

[0021] View the API interface call logs of the target system to confirm whether there are exchange records in the source system. If so, go to step 2; otherwise, automatically connect the target system database to the source system, which specifically includes the following steps:

[0022] Determine the data fields to be synchronized;

[0023] Identify the sensitive fields in the synchronized data fields, encrypt or hash the sensitive fields, and create independent indexes for the frequently queried fields;

[0024] Call the API interface of the target system to dock with the source system, or use an ETL tool to extract data from the source system, and load it into the ERP system after cleaning;

[0025] According to the preset monitoring frequency, regularly audit the access rights of the API interface and remove redundant account information.

[0026] Preferably, determining the consistency of data synchronized from the source system to different target systems includes:

[0027] Obtain multiple pieces of data to be verified from different target systems, and divide them into blocks according to the same rule to obtain multiple corresponding data blocks one and data blocks two;

[0028] Calculate the hash value one of each data block one and the hash value two of data block two;

[0029] Construct a Merkle tree one based on the hash value one, and construct a Merkle tree two based on the hash value two;

[0030] After obtaining the root hash value of Merkle tree one and the root hash value of Merkle tree two, compare them. If there are differences, find the corresponding data blocks one and data blocks two, and use data block one as the standard to overwrite the corresponding data block two;

[0031] If data block two cannot be automatically overwritten, push the different data blocks one and data blocks two to the host computer for manual processing.

[0032] Preferably, generating a data synchronization strategy based on the data synchronization cost includes:

[0033] Construct a dynamic adjustment model to dynamically adjust the data volume ratio of full data and incremental data. Specifically:

[0034] Construct a state matrix , where represents the data change rate, represents the full data synchronization cost, represents the incremental data synchronization cost, represents the average time consumption of historical synchronization;

[0035] Construct a weight matrix , the first row elements of the weight matrix represent the decision vectors corresponding to full data synchronization; the second row elements of the weight matrix represent the decision vectors corresponding to incremental data synchronization;

[0036] Generate a synchronization strategy ; where and represent the data ratio of full synchronization and the data ratio of incremental synchronization;

[0037] Optimize the weight matrix by the gradient descent method to minimize the synchronization cost , where the gradient calculation process is:

[0038] ;

[0039] The updated weight matrix is ;

[0040] When happens, trigger full - volume data synchronization; otherwise, synchronize incremental data according to the ratio; where represents the full - volume synchronization threshold.

[0041] Preferably, after quantifying the importance of data and data synchronization tasks and combining them with data synchronization costs, optimizing the data synchronization strategy includes:

[0042] Extract the importance of data and data synchronization tasks from the metadata of a target system. The importance of data is the weight of the impact of the data on the system business defined based on the preset business rules of the target system, and the importance of the data synchronization task is the priority of the task defined based on the type of the synchronization task. The types of tasks include real - time report tasks and offline analysis tasks;

[0043] Specifically: Based on the rating of the customer corresponding to the data, associate the importance of the data with the rating, and quantify the importance of the data with the customer rating ; Based on the priority of the data synchronization task, associate the importance of the data synchronization task with the priority, and quantify the importance of the data synchronization task with the priority ;

[0044] Add and to to form a new state matrix ;

[0045] Construct a new synchronization cost , where represents the penalty coefficient, represents the delay time during incremental data synchronization;

[0046] Update the weight matrix by the gradient - descent method to minimize ; where the gradient calculation process is: ; where , ;

[0047] The optimized synchronization strategy: ;

[0048] When happens, force full - volume data synchronization, that is, ; where represents the task importance threshold one;

[0049] When happens, allow delay in incremental data synchronization, represents the task importance threshold two.

[0050] Preferably, adjustment rules are added to the optimized data synchronization strategy to dynamically adjust the importance of data or tasks, including:

[0051] Based on customer behavior, transaction data, and system business rules, quantify customer value and formulate dynamic adjustment rules for customer ratings, specifically including the following steps:

[0052] Obtain customer behavior, transaction data, and system business rule data from the target system database to form a customer data set;

[0053] Extract key features from the customer data set to form a customer key feature set , and the real-time key features include the customer's cumulative transaction amount, maximum transaction amount, activity, and repurchase rate;

[0054] After normalizing the customer key feature set, input it into the time series model ARIMA to predict the customer's consumption amount in the next 6 months;

[0055] Construct a customer value scoring model:

[0056] , where respectively represent the scoring weight coefficients; represents the customer's maximum transaction amount; represents the cumulative transaction amount; represents the repurchase rate; represents the predicted value of the customer's consumption amount in the next 6 months; represents the decay coefficient, ;

[0057] Set the classification interval for customer value scoring , When, judge that the customer is a high-value customer; when When, judge that the customer is a low-value customer;

[0058] Combine the type of data synchronization task, timeliness requirements, and the value of associated data to formulate dynamic adjustment rules for task priorities, specifically including the following steps:

[0059] According to the data usage, classify the types of data synchronization tasks into real-time risk control, real-time reports, batch analysis, and historical archiving;

[0060] Assign task weights to different types of data synchronization tasks ;

[0061] Construct a task importance scoring model:

[0062] ; where represents the task weight vector, ; indicates the task has been delayed; represents the adjustment coefficient; represents the data importance, ;

[0063] Set the classification interval for task importance scoring , when, it is judged as the highest priority task; when, it is judged that the synchronization task is a low priority task;

[0064] Set the adjustment rule:

[0065] When the customer's daily transaction amount is greater than 5 times the average value of the previous 6 months, trigger the upgrade of the customer value score; at the same time, upgrade the priority of the tasks associated with the customer;

[0066] When the customer has no transaction data for 3 consecutive months, and continuously decreases, then trigger the downgrade of the customer value score; at the same time, lower the priority of the tasks associated with the customer;

[0067] Specifically: transmit customer transaction data through Kafka, call the customer value scoring model through Flink, calculate the customer value score, and write the customer value score into Redis for real-time query by the system;

[0068] Periodically train the time series model ARIMA through Spark to update the Hive table.

[0069] Preferably, adding an adjustment rule in the optimized data synchronization strategy to dynamically adjust the importance of data or tasks, further includes:

[0070] Add a temporary importance adjustment mechanism to achieve temporary increase and decrease of the importance of data or tasks, specifically:

[0071] Add a temporary adjustment table in a target system database, the temporary adjustment table includes adjustment record ID, adjustment object type, customer ID, synchronization task ID, importance before adjustment, importance after adjustment, effective time, expiration time, operator ID and adjustment reason;

[0072] The adjustment mechanism is: the currently effective importance value is ; where, represents the quantization value of the currently effective importance, represents the importance after adjustment, represents the output of the task importance scoring model or the customer value scoring model, represents the current time, represents the effective time of the adjustment, Indicates the failure time of the adjustment;

[0073] Through the priority override middleware, use the adjusted importance to override the importance before adjustment;

[0074] According to the preset scanning frequency, automatically scan for expired adjustments and record them, and automatically stop the temporary importance adjustment mechanism after triggering the notification information.

[0075] Preferably, the step of using the adjusted importance to override the importance before adjustment through the priority override middleware further includes: integrating the customer importance score, task importance score, data importance, and temporary adjustment records to generate a comprehensive importance, and generating the finally adjusted importance based on the comprehensive importance and the adjusted importance in the temporary adjustment table. The specific steps are as follows:

[0076] Obtain the customer importance score based on the customer value scoring model ;

[0077] Obtain the task importance score based on the task importance scoring model ;

[0078] Obtain the data importance and the quantization value of the adjusted importance ;

[0079] Perform normalization on , , and ;

[0080] Comprehensive importance ;

[0081] Among them, represents the resource weight coefficient; among them, represents , the normalized value.

[0082] An AI-based enterprise business data processing system includes: a processor, a memory connected to the processor, and a communication module. The system is used to execute the AI-based enterprise business data processing method described above.

[0083] A computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the AI-based enterprise business data processing method described above.

[0084] Advantages of the present invention:

[0085] 1. For the present invention, first, it is determined whether the customer data of the source system (CRM system) is connected to the target system (ERP system), and in the case of non-connection, secure connection is achieved to ensure that customer data is in different data.

[0086] 2. For the present invention, after solving the connection between customer data and the target system, by designing a data synchronization strategy, different data storage schemes are made consistent. For example, while synchronizing data from the source system to the database of the ERP system, part of the data or analysis is stored in Hadoop, and the data consistency between the ERP system database and the Hadoop storage scheme is ensured through the data synchronization strategy.

[0087] 3. For the present invention, by dynamically adjusting the model, adjustment rules are added during the data synchronization process to adjust the importance of data or tasks, and the ratio of full data and incremental data during the data synchronization process is dynamically adjusted according to the importance of data and tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Figure 1 It is a flowchart of a method for processing enterprise business data based on AI according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0089] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments in the following description are only examples, and those skilled in the art can think of other obvious variations. The basic principles defined in the following description can be applied to other implementation schemes, variation schemes, improvement schemes, equivalent schemes, and other technical schemes that do not depart from the spirit and scope of the present invention.

[0090] It can be understood that the term "one" should be understood as "at least one" or "one or more". That is, in one embodiment, the number of an element can be one, and in another embodiment, the number of the element can be multiple. The term "one" cannot be understood as a limitation on the number.

[0091] Embodiment 1:

[0092] Refer to Figure 1 , the technical solution provided by the present invention is: a method for processing enterprise business data based on AI, including the following steps:

[0093] Step 1: Obtain source system data and store the obtained data in one or more target systems; the source system includes a customer relationship management system CRM and a business intelligence BI system, and the target system includes an enterprise resource planning ERP system and a big data processing Hadoop system;

[0094] Step 2: Construct a data synchronization policy, and monitor and manage data synchronization between target system databases based on the data synchronization policy, including the following sub-steps:

[0095] Step 2.1: Generate a data synchronization policy based on the data synchronization cost. It includes the following sub-steps:

[0096] Construct a dynamic adjustment model to dynamically adjust the data volume ratio of full data and incremental data. Specifically:

[0097] Construct a state matrix , where represents the data change rate, represents the full data synchronization cost, represents the incremental data synchronization cost, represents the average time consumption of historical synchronization.

[0098] The data change rate is the percentage of the number of newly added or modified records in the total number of records per unit time, and the number of changed records can be counted from the database log; the full data synchronization cost is the time, bandwidth, and computing resources (CPU and memory usage) consumed during full synchronization; the incremental data synchronization cost is the resource consumption (CPU and memory usage) of the CDC tool; the average time consumption of historical synchronization is the average time consumption of the past N synchronization tasks.

[0099] Construct a weight matrix , where the first row elements of the weight matrix represent the decision vector corresponding to full data synchronization; the second row elements of the weight matrix represent the decision vector corresponding to incremental data synchronization;

[0100] Generate a synchronization policy ; where and represent the data ratio of full synchronization and the data ratio of incremental synchronization;

[0101] Optimize the weight matrix by the gradient descent method to minimize the synchronization cost , where the gradient calculation process is:

[0102] ;

[0103] The updated weight matrix is ;

[0104] When , trigger full data synchronization; otherwise, synchronize incremental data according to the ratio; where represents the full synchronization threshold.

[0105] For example, in this embodiment, the state vector in the current state is ;

[0106] The weight matrix is , ;

[0107] Obtained after Softmax normalization , ;

[0108] The synchronization policy is that 52% of the synchronized data volume uses full synchronization and 48% uses incremental synchronization.

[0109] Step 2.2. After quantifying the importance of the data and the data synchronization task, combine it with the data synchronization cost to optimize the data synchronization policy. It includes the following sub-steps:

[0110] Extract the data importance and the importance of the data synchronization task from the metadata of a target system. The data importance is the impact weight of the data on the system business extracted based on the predefined business rules of the target system, and the importance of the data synchronization task is the priority of the task defined based on the type of the synchronization task. The types of the tasks include real-time report tasks and offline analysis tasks;

[0111] Specifically: Based on the rating of the customer corresponding to the data, associate the importance of the data with the rating, and quantify the importance of the data with the customer rating ; Based on the priority of the data synchronization task, associate the importance of the data synchronization task with the priority, and quantify the importance of the data synchronization task with the priority ;

[0112] Add and to to form a new state matrix ;

[0113] Construct a new synchronization cost , where represents the penalty coefficient, represents the delay time during incremental data synchronization;

[0114] Update the weight matrix by the gradient descent method to minimize ; The gradient calculation process is: ;

[0115] Among them, , ;

[0116] The optimized synchronization policy: ;

[0117] When , force full synchronization of the data, that is, ; where, represents the first task importance threshold;

[0118] When it is the case, incremental data synchronization delay is allowed, represents the second task importance threshold.

[0119] For example, in this embodiment, the two target systems are the ERP system and the Hadoop system; after the data of the CRP system (source system) is accessed to the ERP system, it is necessary to synchronize the data to Hadoop for backup.

[0120] Currently, it is necessary to synchronize the order data of VIP customers. The of VIP customers, the type of the synchronization task is order data. Let the priority of the order data task be 3, that is ;

[0121] When , , , it is the case,

[0122] ;

[0123] Optimal solution: When it is the case, ; when it is the case, . Because the data of VIP customers is highly important, even if the cost of full synchronization is slightly higher, the system still chooses full synchronization (to avoid the penalty risk of 75 units).

[0124] Step 2.3. Add adjustment rules to the optimized data synchronization strategy to dynamically adjust the importance of data or tasks. It includes the following sub-steps:

[0125] Based on customer behavior, transaction data, and system business rules, quantify customer value and formulate dynamic adjustment rules for customer ratings, which specifically include the following steps:

[0126] Obtain customer behavior, transaction data, and system business rule data from the target system database to form a customer data set;

[0127] Extract key features from the customer data set to form a customer key feature set , and the real-time key features include the cumulative transaction amount, maximum transaction amount, activity level, and repurchase rate of customers;

[0128] After normalizing the customer key feature set, input it into the time series model ARIMA to predict the customer's consumption amount in the next 6 months;

[0129] Construct a customer value scoring model:

[0130] , where respectively represent the scoring weight coefficients; represents the maximum transaction amount of the customer; represents the cumulative transaction amount; represents the repurchase rate; represents the predicted value of the customer's consumption amount in the next 6 months; represents the attenuation coefficient, ;

[0131] Set the classification interval of customer value score , when, it is determined that the customer is a high-value customer; when when, it is determined that the customer is a low-value customer;

[0132] Combined with the type, timeliness requirements of the data synchronization task and the value of the associated data, formulate the dynamic adjustment rules for the priority of the task, specifically including the following steps:

[0133] According to the data usage, classify the types of data synchronization tasks into real-time risk control, real-time reports, batch analysis, and historical archiving;

[0134] Assign task weights to different types of data synchronization tasks ;

[0135] Construct a task importance scoring model:

[0136] ; where represents the task weight vector, ; represents the time the task has been delayed, represents the adjustment coefficient; represents the data importance, ;

[0137] Set the classification interval of task importance score , when, it is determined as the highest priority task; when, it is determined that the synchronization task is a low priority task;

[0138] Set the adjustment rule:

[0139] When the customer's daily transaction amount is greater than 5 times the average of the previous 6 months, trigger the upgrade of the customer value score; at the same time, upgrade the priority of the task associated with the customer;

[0140] When the customer has no transaction data for 3 consecutive months, and continuously decreases, then trigger the downgrade of the customer value score; at the same time, lower the priority of the task associated with the customer;

[0141] Specifically: transmit customer transaction data through Kafka, call the customer value scoring model through Flink to calculate the customer value score, and write the customer value score into Redis for real-time query by the system;

[0142] Periodically train the time series model ARIMA through Spark to update the Hive table.

[0143] Step 3: Send the data of one or more target systems to the cloud platform for storage.

[0144] Before performing Step 1, the following steps are also included:

[0145] Judge whether the target system database is connected to the source system and judge the consistency of the data synchronized from the source system to different target systems.

[0146] Among them, judging whether the target system database is connected to the source system includes the following steps:

[0147] Modify one or more fields in the source system, detect whether the corresponding fields in the target system change. If so, enter Step 2; otherwise, enter the following steps:

[0148] Modify one or more fields in the target system and check whether the corresponding fields in the source system are synchronized. If so, enter Step 2; otherwise, enter the following steps:

[0149] Sample and export the data of the target system and the source system for comparison to judge whether they are consistent. If so, enter Step 2; otherwise, enter the following steps:

[0150] View the API interface call logs of the target system to confirm whether there are exchange records with the source system. If so, enter Step 2; otherwise, automatically connect the target system database to the source system, which specifically includes the following steps:

[0151] Determine the data fields to be synchronized;

[0152] Identify the sensitive fields in the synchronized data fields, encrypt or hash the sensitive fields, and create independent indexes for the frequently queried fields;

[0153] Call the API interface of the target system to dock with the source system, or use an ETL tool to extract data from the source system, and load it into the ERP system after cleaning;

[0154] According to the preset monitoring frequency, regularly audit the access rights of the API interface and remove redundant account information.

[0155] Among them, judging the consistency of data synchronized from the source system to different target systems includes the following steps:

[0156] Obtain multiple data to be verified from different target systems, and perform chunking according to the same rules to obtain multiple corresponding data chunks one and data chunks two;

[0157] Calculate and obtain the hash value one of each data chunk one and the hash value two of data chunk two;

[0158] Construct Merkle tree one based on hash value one, and construct Merkle tree two based on hash value two;

[0159] After obtaining the root hash value of Merkle tree one and the root hash value of Merkle tree two, compare them. If there are differences, find the corresponding data chunks one and data chunks two, and use data chunk one as the standard to overwrite the corresponding data chunk two;

[0160] If data chunk two cannot be automatically overwritten, push the different data chunks one and data chunks two to the host computer for manual processing.

[0161] Embodiment two:

[0162] Based on Embodiment one, when dynamically adjusting the importance of data or tasks, this embodiment adds an adjustment rule. Specifically, it includes the following steps:

[0163] Add a temporary importance adjustment mechanism to achieve temporary promotion and reduction of the importance of data or tasks. Specifically:

[0164] Add a temporary adjustment table to a target system database. The temporary adjustment table includes adjustment record ID, adjustment object type, customer ID, synchronization task ID, importance before adjustment, importance after adjustment, effective time, expiration time, operator ID, and adjustment reason;

[0165] The adjustment mechanism is that the currently effective importance value is ; among them, represents the quantization value of the currently effective importance, represents the importance after adjustment, represents the output of the task importance scoring model or the customer value scoring model, represents the current time, represents the effective time of the adjustment, represents the expiration time of the adjustment;

[0166] Through the priority override middleware, use the adjusted importance to override the importance before adjustment;

[0167] According to the preset scanning frequency, automatically scan for expired adjustments and record them, and automatically stop the temporary importance adjustment mechanism after triggering the notification information.

[0168] In this embodiment, the temporary importance adjustment mechanism collaborates with the dynamic adjustment model, enabling business personnel of the enterprise to quickly respond to duration changes and temporarily adjust the importance of data or tasks.

[0169] Embodiment Three:

[0170] Based on Embodiment Two, this embodiment first integrates the customer importance score, task importance score, data importance, and temporary adjustment records to generate a comprehensive importance, and then compares it with the adjusted importance in the temporary adjustment table to generate the finally adjusted importance. This is to achieve the effect of managing the customer importance score, task importance score, data importance, and temporary adjustment records through the priority overlay middleware. The specific steps are as follows:

[0171] Obtain the customer importance score based on the customer value scoring model ;

[0172] Obtain the task importance score based on the task importance scoring model ;

[0173] Obtain the data importance and the quantization value of the adjusted importance ;

[0174] Perform normalization on , , and ;

[0175] Comprehensive importance ;

[0176] Among them, represents the resource weight coefficient; among them, represents , the normalized values,

[0177] If , then use to force coverage , otherwise use .

[0178] For example, taking the slice task of the 5G network as an example:

[0179] The input variables are , , , the effective ; They are respectively valued at 0.3, 0.3, 0.2, 0.2;

[0180] Comprehensive importance ;

[0181] Since , use Forced overwrite , , the finally adjusted importance is 0.9.

[0182] If the adjusted importance is the importance of the data, use the adjusted importance of the data to replace the importance of the data quantified based on the customer rating , form a new state matrix, and then finally obtain the full - volume synchronization data ratio and the incremental synchronization data ratio. Control the data synchronization process according to the full - volume synchronization data ratio and the incremental synchronization data ratio.

[0183] The present invention also provides an AI - based enterprise business data processing system, including: a processor, a memory and a communication module connected to the processor. The system is used to execute the described AI - based enterprise business data processing method.

[0184] The present invention also provides a computer - readable storage medium. The computer - readable storage medium stores a computer program, and the computer program is executed by the processor to implement the described AI - based enterprise business data processing method.

[0185] Embodiments disclosed in the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. Embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes program codes for executing the methods shown in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part, and / or installed from a removable medium. When the computer program is executed by a central processing unit (CPU), the above-mentioned functions defined in the methods of the present application are executed. It should be noted that the computer-readable medium described above in the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the two. A computer-readable storage medium can, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of a computer-readable storage medium can include, but are not limited to: an electrical connection having one or more wire segments, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium can be any tangible medium that contains or stores a program, and the program can be used by or combined with an instruction execution system, apparatus, or device. And in the present application, a computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program codes. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, and the computer-readable medium can send, propagate, or transmit a program for use by or combined with an instruction execution system, apparatus, or device. The program codes contained on the computer-readable medium can be transmitted by any suitable medium, including but not limited to: a wireless segment, a wire segment, an optical cable, RF, etc., or any suitable combination of the above.

[0186] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a part of code that contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions noted in the blocks may occur in a different order than that noted in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0187] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The objectives of the present invention have been fully and effectively achieved. The functions and structural principles of the present invention have been demonstrated and illustrated in the embodiments. Without departing from the said principles, any changes or modifications can be made to the embodiments of the present invention.

Claims

1. An enterprise business data processing method based on AI, characterized in that: The method comprises: Step 1: Obtain source system data and store the acquired data in one or more target systems; Step 2: Build a data synchronization strategy to monitor and manage data synchronization between target system databases based on the data synchronization strategy, including: Generate a data synchronization strategy based on data synchronization costs; including: Build a dynamic adjustment model to dynamically adjust the data volume ratio between full data and incremental data. Specifically: Constructing the state matrix ,in, Indicates the data change rate, Indicates the full data synchronization cost, represents the incremental data synchronization cost, Indicates the average time taken for historical synchronization; Constructing the weight matrix , the first row elements of the weight matrix represent the decision vectors corresponding to the full data synchronization; the second row elements of the weight matrix represent the decision vectors corresponding to the incremental data synchronization; Generate synchronization strategy ;in and Indicates the proportion of data synchronized in full and incremen- tally in sync. Optimize the weight matrix by gradient descent to minimize the synchronization cost , where the gradient calculation process is: ； The updated weight matrix is ; when When the full data synchronization is triggered; otherwise, Proportional synchronous incremental data; where, Indicates the full synchronization threshold; After quantifying the importance of data and data synchronization tasks, we combine them with the data synchronization cost and optimize the data synchronization strategy. This includes: Extracting data importance and data synchronization task importance from metadata of a target system, wherein the data importance is the weight of the impact of the extracted data on the system business based on the preset business rules of the target system, and the data synchronization task importance is the priority of the task defined based on the type of synchronization task, and the task type includes real-time report task and offline analysis task; Specifically: Based on the rating of the customer corresponding to the data, the importance of the data is associated with the rating, and the importance of the data is quantified using the customer rating. ; Based on the priority of data synchronization tasks, the importance of data synchronization tasks is associated with the priority, and the importance of data synchronization tasks is quantified by the priority ; Will and Add to In the new state matrix ; Building a new synchronization cost ,in, represents the penalty coefficient, Indicates the delay time when synchronizing incremental data; Update the weight matrix by gradient descent to minimize ; The gradient calculation process is: ;in, , ; Optimized synchronization strategy: ; when When the data is fully synchronized, ;in, represents the task importance threshold of one; when When , incremental data synchronization delay is allowed. represents the task importance threshold 2; Add adjustment rules to the optimized data synchronization strategy to dynamically adjust the importance of data or tasks; Step 3: Store the data of one or more target systems on the cloud platform.

2. The method for processing enterprise business data based on AI according to claim 1, characterized in that: The source system includes a customer relationship management system (CRM) and a business intelligence (BI) system, and the target system includes an enterprise resource planning (ERP) system and a big data processing (Hadoop) system. Before obtaining the source system data and storing the obtained data in the target system database, it also includes: Determine whether the target system database is connected to the source system; Determine the consistency of data synchronized from the source system to different target systems; The step of determining whether the target system database is connected to the source system includes the following steps: Modify one or more fields in the source system and check whether the corresponding fields in the target system have changed. If yes, proceed to step 2. Otherwise, proceed to the following steps: Modify one or more fields in the target system and check whether the corresponding fields in the source system are automatically synchronized. If yes, proceed to step 2. Otherwise, proceed to the following steps: Sample and export the target system and source system data for comparison to determine whether they are consistent. If yes, proceed to step 2. Otherwise, proceed to the following steps: Check the API call log of the target system to confirm whether there is an exchange record in the source system. If yes, proceed to step 2. Otherwise, automatically connect the target system database with the source system. Specifically, the following steps are included: Determine the data fields that need to be synchronized; Identify sensitive fields in synchronized data fields, encrypt or hash sensitive fields, and create independent indexes for high-frequency query fields; Call the API interface of the target system to connect with the source system, or use ETL tools to extract data from the source system and load it into the ERP system after cleaning; According to the preset monitoring frequency, regularly audit the access rights of the API interface and remove redundant account information.

3. The enterprise business data processing method based on AI according to claim 2 is characterized in that: Determining the consistency of data synchronized from the source system to different target systems includes: Acquire multiple data to be verified from different target systems, and divide them into blocks according to the same rule to obtain multiple corresponding data blocks 1 and data blocks 2; Calculate and obtain the hash value 1 of each data block 1 and the hash value 2 of each data block 2; Construct Merkle tree one based on hash value one, and construct Merkle tree two based on hash value two; Obtain the root hash value of Merkle tree 1 and the root hash value of Merkle tree 2 and compare them. If there is a difference, find the corresponding data block 1 and data block 2, and use data block 1 as the basis to overwrite the corresponding data block 2. If data block 2 cannot be automatically overwritten, the difference data block 1 and data block 2 are pushed to the host computer for manual processing.

4. The method for processing enterprise business data based on AI according to claim 3, characterized in that: Add adjustment rules to the optimized data synchronization strategy to dynamically adjust the importance of data or tasks, including: Based on customer behavior, transaction data and system business rules, quantify customer value and formulate dynamic adjustment rules for customer ratings, which specifically include the following steps: Obtain customer behavior, transaction data, and system business rule data from the target system database to form a customer data set; Extract key features from customer data sets to form customer key feature sets ,Real-time key features include customers’ cumulative transaction amount, maximum transaction amount, activity, and repurchase rate; After normalizing the customer's key feature set, input it into the time series model ARIMA to predict the customer's consumption amount in the next 6 months; Build a customer value scoring model: ,in, Respectively represent the scoring weight coefficient; Indicates the maximum transaction amount of the customer; Indicates the cumulative transaction amount; Represents the repurchase rate; Indicates the predicted value of the customer's consumption in the next 6 months; represents the attenuation coefficient, ; Set customer value rating category range , When When the customer is judged as a low-value customer; Based on the type of data synchronization task, timeliness requirements, and the value of associated data, a dynamic adjustment rule for the task priority is formulated, which specifically includes the following steps: According to the purpose of data, the types of data synchronization tasks are divided into real-time risk control, real-time reporting, batch analysis and historical archiving; Assign task weights to different types of data synchronization tasks ; Build a task importance scoring model: ;in, represents the task weight vector, ; Indicates that the task has been delayed. represents the adjustment coefficient; Indicates the importance of data. ; Set the task importance rating classification interval , When , it is judged as the highest priority task; When , the synchronization task is judged as a low priority task; Set adjustment rules: When a customer's daily transaction volume is more than five times the average of the previous six months, the customer value score is upgraded; at the same time, the priority of the task associated with the customer is upgraded; When the customer has no transaction data for 3 consecutive months, and Continuous decline will trigger a downgrade of the customer value score and reduce the priority of tasks associated with the customer; Specifically: transmit customer transaction data through Kafka, call the customer value scoring model through Flink, calculate the customer value score, and write the customer value score to Redis for real-time query by the system; Periodically train the ARIMA time series model through Spark to update the Hive table.

5. The method for processing enterprise business data based on AI according to claim 4, characterized in that: The step of adding adjustment rules to the optimized data synchronization strategy to dynamically adjust the importance of data or tasks also includes: Add a temporary importance adjustment mechanism to temporarily increase or decrease the importance of data or tasks. Specifically: Adding a temporary adjustment table in a target system database, wherein the temporary adjustment table includes an adjustment record ID, an adjustment object type, a customer ID, a synchronization task ID, importance before adjustment, importance after adjustment, effective time, expiration time, operator ID, and adjustment reason; The adjustment mechanism is: the current effective importance value is ;in, A quantitative value indicating the importance of the current effect. represents the adjusted importance, represents the output of the task importance scoring model or the customer value scoring model, Indicates the current time. Indicates the effective time of the adjustment. represents the adjusted expiration time; Use the priority override middleware to override the importance before adjustment with the adjusted importance; According to the preset scanning frequency, the overdue adjustments are automatically scanned and recorded, and the temporary importance adjustment mechanism is automatically stopped after the notification information is triggered.

6. The method for processing enterprise business data based on AI according to claim 5, characterized in that: The method of using the priority overwriting middleware to overwrite the importance before adjustment with the adjusted importance also includes: integrating the customer importance score, the task importance score, the data importance and the temporary adjustment record to generate a comprehensive importance, and generating a final adjusted importance based on the comprehensive importance and the adjusted importance in the temporary adjustment table. The specific steps are as follows: Obtain customer importance scores based on the customer value scoring model ; Obtaining task importance scores based on the task importance scoring model ; Obtaining data importance and the quantitative value of adjusted importance ; Will , , and Perform normalization processing; Overall importance ; in, represents the resource weight coefficient; where, express , The normalized value, if , then use Forced coverage , otherwise use .

7. An enterprise business data processing system based on AI, comprising: A processor and a memory and a communication module connected to the processor, characterized in that the system is used to execute an AI-based enterprise business data processing method as described in any one of claims 1-6 above.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement an AI-based enterprise business data processing method as described in any one of claims 1 to 6.

Images