Business and financial fusion management method and system based on intelligent data analysis

Abstract

The invention relates to the technical field of business and financial fusion management, and discloses a business and financial fusion management method and system based on intelligent data analysis, and the method comprises the steps: capturing order state change event data and voucher generation event data from a key business system and a financial system through an automatic means; and the timeliness and accuracy of the original data are ensured. Then, the data are cleaned and standardized, format differences are eliminated, data standards are unified, and then a business and financial event association engine is adopted to process the cleaned and standardized data so as to identify and establish association between business events and financial vouchers. Marking the successfully associated data as associated data and generating a fusion event record; and marking the data which are not successfully associated as to-be-processed data. Thus, the accuracy and comprehensiveness of business and financial data integration are improved, the flexibility and adaptability of data processing are enhanced through an intelligent means, and more powerful decision support is provided for enterprises.

Description

Technical Field

[0001] This application relates to the field of business and finance integration management technology, and more specifically, to a business and finance integration management method and system based on intelligent data analysis. Background Technology

[0002] In today's rapidly developing information age, the efficient integration of business operations and financial management has become a key factor in enhancing corporate competitiveness. However, current technologies still face numerous challenges in achieving real-time correlation and mapping between business and financial data. Traditionally, a company's business systems (such as sales and procurement) and financial systems (such as accounting and cash management) typically operate independently. This often results in lags in data interaction between the two systems, hindering real-time data sharing and synchronization. This information silo phenomenon not only affects the timeliness and accuracy of decision-making but can also lead to issues such as duplicate data entry errors and data inconsistencies.

[0003] Specifically, various order status change events generated during business process execution, such as order creation, shipment, and returns, typically need to be promptly reflected in the financial system for corresponding accounting processing. However, due to the lack of an effective mechanism to ensure that these business events can be automatically and accurately mapped to financial vouchers, manual intervention is often required for verification and adjustment, which undoubtedly increases workload and the possibility of errors. Furthermore, as enterprises grow in size and business complexity increases, the massive amounts of business and financial data make manual matching increasingly difficult, urgently requiring an intelligent method to solve this problem.

[0004] Therefore, we look forward to an optimized business and finance integration management solution based on intelligent data analysis. Summary of the Invention

[0005] This application addresses the problems of low data processing efficiency, error-proneness, and information silos between existing business and financial systems. Embodiments of this application provide a business-finance integrated management method and system based on intelligent data analysis.

[0006] Firstly, a business-finance integration management method based on intelligent data analysis is provided, comprising: capturing order status change event data from a key business system; capturing voucher generation event data from a financial system; cleaning and standardizing the order status change event data and the voucher generation event data to obtain cleaned and standardized order status change event data and cleaned and standardized voucher generation event data; inputting the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data into a business-finance event association engine to obtain association analysis results; if the association analysis result indicates successful association, marking the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data as associated, and generating a fused event record containing the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data; if the association analysis result indicates unsuccessful association, marking the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data as pending processing.

[0007] In conjunction with the first aspect, in one possible implementation, the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are input into the business and finance event association engine to obtain association analysis results, including: inputting the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data into the strong key association analysis unit of the business and finance event association engine to obtain first-level association analysis results.

[0008] In conjunction with the first aspect, in one possible implementation, the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are input into the strong key association analysis unit of the business and finance event association engine to obtain a first-level association analysis result. This includes: the strong key association analysis unit checking whether a predefined strong association key exists simultaneously in the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data; if it exists, the first-level association analysis result is a successful association; if it does not exist, the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are input into the intelligent association analysis unit of the business and finance event association engine to obtain a second-level association analysis result.

[0009] In conjunction with the first aspect, in one possible implementation, the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are input into the intelligent correlation analysis unit of the business and finance event correlation engine to obtain secondary correlation analysis results, including: determining the time proximity between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain a time proximity score; determining the monetary similarity between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain a monetary similarity score; determining the monetary similarity between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data... The entity association between the cleaned and standardized voucher generation event data is determined to obtain an entity association score; the consistency of the units of measurement between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data is determined to obtain a unit of measurement consistency score; the text similarity between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data is determined to obtain a text similarity score; based on the time proximity score, the amount similarity score, the entity association score, the unit of measurement consistency score, and the text similarity score, the secondary association analysis result is generated.

[0010] In conjunction with the first aspect, in one possible implementation, the secondary association analysis result is generated based on the time proximity score, the amount similarity score, the entity relevance score, the unit consistency score, and the text similarity score. This includes: inputting the time proximity score, the amount similarity score, the entity relevance score, the unit consistency score, and the text similarity score into a scoring model to obtain a comprehensive data association score; if the comprehensive data association score is greater than or equal to a preset threshold, the secondary association analysis result is a successful association; if the comprehensive data association score is less than the preset threshold, the secondary association analysis result is a failed association.

[0011] In conjunction with the first aspect, one possible implementation further includes: before inputting the time proximity score, the amount similarity score, the entity association score, the unit consistency score, and the text similarity score into the scoring model, based on the text similarity score, performing weighted commutation conformance on the time proximity score, the amount similarity score, the entity association score, and the unit consistency score to obtain conformed time proximity score, conformed amount similarity score, conformed entity association score, and conformed unit consistency score.

[0012] In conjunction with the first aspect, in one possible implementation, based on the text similarity score, weighted commutation conformal processing is performed on the temporal proximity score, the monetary similarity score, the entity association score, and the unit consistency score to obtain conformed temporal proximity scores, conformed monetary similarity scores, conformed entity association scores, and conformed unit consistency scores. This includes: determining a fractional-order guiding operator for the weighted difference parameterization representation; and performing single-modal distribution modulation on the temporal proximity score, the monetary similarity score, the entity association score, and the unit consistency score to obtain modulated temporal proximity scores. The modulated amount similarity score, modulated entity association score, and modulated unit consistency score are subjected to single-modal distribution modulation; a modulated weight co-action conformal space is defined; based on the modulated weight co-action conformal space and the text similarity score, the modulated time proximity score, modulated amount similarity score, modulated entity association score, and modulated unit consistency score are subjected to single-modal distribution modulation and conformal mapping at the single-modal fractional order to obtain the conformed time proximity score, the conformed amount similarity score, the conformed entity association score, and the conformed unit consistency score.

[0013] In conjunction with the first aspect, in one possible implementation, determining the text similarity between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain a text similarity score includes: performing semantic embedding encoding on the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain an order status change event semantic embedding encoding vector and a voucher generation event semantic embedding encoding vector; and calculating the cosine similarity between the order status change event semantic embedding encoding vector and the voucher generation event semantic embedding encoding vector as the text similarity score.

[0014] In conjunction with the first aspect, in one possible implementation, the order status change event data includes order ID, customer ID, product ID, amount, timestamp, and event type, and the voucher generation event data includes voucher ID, account code, amount, debit / credit direction, and timestamp.

[0015] Secondly, a business-finance integration management system based on intelligent data analysis is provided for executing the aforementioned business-finance integration management method based on intelligent data analysis. The system includes: an order status change event capture module for capturing order status change event data from a key business system; a voucher generation event capture module for capturing voucher generation event data from a financial system; a data cleaning module for cleaning and standardizing the order status change event data and the voucher generation event data to obtain cleaned and standardized order status change event data and cleaned and standardized voucher generation event data; and a correlation analysis result generation module for generating correlation analysis results from the cleaned and standardized order status change event data and the voucher generation event data. The cleaned and standardized voucher generation event data is input into the business and finance event association engine to obtain association analysis results; the fusion event record generation module is used to mark the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data as associated if the association analysis result is successful association, and generate a fusion event record containing the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data; the data marking module is used to mark the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data as pending processing if the association analysis result is unsuccessful association.

[0016] Compared to existing technologies, the business-finance integration management method and system based on intelligent data analysis provided in this application first captures order status change event data and voucher generation event data from key business systems and financial systems respectively through automated means, ensuring the timeliness and accuracy of the raw data. Then, this data is cleaned and standardized to eliminate format differences and unify data standards. Subsequently, a business-finance event correlation engine is used to process the cleaned and standardized data to identify and establish the correlation between business events and financial vouchers. Data that is successfully correlated is marked as correlated and a fusion event record is generated; data that fails to be correlated is marked as pending processing. This not only improves the accuracy and comprehensiveness of business-finance data integration but also enhances the flexibility and adaptability of data processing through intelligent means, providing enterprises with more powerful decision support. Attached Figure Description

[0017] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0018] Figure 1The illustration shows a schematic flowchart of a business-finance integration management method based on intelligent data analysis according to an embodiment of this application.

[0019] Figure 2 The illustration shows a schematic flowchart of S4 in the business and finance integration management method based on intelligent data analysis according to an embodiment of this application.

[0020] Figure 3 The illustration shows a schematic flowchart of step S43 in the business-finance integration management method based on intelligent data analysis according to an embodiment of this application.

[0021] Figure 4 The figure shows a schematic flowchart of step S435 in the business and finance integration management method based on intelligent data analysis according to an embodiment of this application.

[0022] Figure 5 The figure shows a schematic flowchart of step S436 in the business and finance integration management method based on intelligent data analysis according to an embodiment of this application.

[0023] Figure 6 The illustration shows a schematic structural diagram of a business-finance integration management system based on intelligent data analysis according to an embodiment of this application. Detailed Implementation

[0024] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0025] Figure 1 The illustration shows a schematic flowchart of a business-finance integration management method based on intelligent data analysis according to an embodiment of this application. Figure 1As shown, this application provides a business-finance integration management method based on intelligent data analysis, including: S1, capturing order status change event data from a key business system; S2, capturing voucher generation event data from a financial system; S3, cleaning and standardizing the order status change event data and the voucher generation event data to obtain cleaned and standardized order status change event data and cleaned and standardized voucher generation event data; S4, inputting the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data into a business-finance event association engine to obtain association analysis results; S5, if the association analysis result is a successful association, marking the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data as associated, and generating a fusion event record containing the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data; S6, if the association analysis result is a failed association, marking the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data as pending processing.

[0026] Specifically, in step S1, order status change event data is captured from the key business system. This data includes order ID, customer ID, product ID, amount, timestamp, and event type. It should be understood that in traditional management models, business activities and financial accounting often suffer from time lags and information barriers. Changes in order status recorded in the business system (such as order creation, approval, shipment, and completion) and the corresponding voucher generation events in the financial system (such as payment confirmation, revenue recognition, and cost transfer) often require manual verification and reconciliation. This process is not only inefficient and error-prone but also fails to provide a real-time view of the business situation, hindering management from making rapid decisions based on the latest data. Each change in order status may trigger or be associated with a financial impact event. For example, order confirmation may indicate future revenue, order shipment is directly related to a decrease in inventory and an increase in outstanding receivables, and the final order completion and payment signifies the actual achievement of revenue and a change in cash flow. Therefore, capturing order status change event data in real-time and accurately constitutes a bridge connecting business activities and their financial consequences, and is a prerequisite for achieving refined and automated integrated business and financial management.

[0027] In a specific embodiment, capturing order status change event data from critical business systems includes: First, identifying and accessing critical business systems, which include: Enterprise Resource Planning (ERP) systems, Customer Relationship Management (CRM) systems, Supply Chain Management (SCM) systems, e-commerce platforms, or other core applications that record the order lifecycle status. The data capture mechanism can be flexibly selected based on the technical architecture of the business system. Common implementation methods include using the system's application programming interface (API) for polling or subscribing to event notifications. For example, if the business system supports webhooks, relevant event data can be proactively pushed to the data receiving endpoint of the business-finance integration system when the order status changes. Another approach is through database-level technologies, such as setting database triggers, which automatically write the changed information (including order ID, customer ID, product ID, changed status, timestamp, amount involved, and other key fields) to an intermediate table or message queue when the status field in the order-related table is updated. For legacy systems that do not provide APIs or do not allow modification of the database structure, change data capture technology can be used to capture data changes by listening to the database transaction logs, or, in specific cases, periodic extraction tasks can be used, but this will sacrifice some real-time performance.

[0028] Specifically, in step S2, voucher generation event data is captured from the financial system. This voucher generation event data includes voucher ID, account code, amount, debit / credit direction, and timestamp. It should be understood that financial vouchers, as the basic unit of accounting, are the legal carriers for classifying, recording, and reporting economic transactions of an enterprise. The generation of each voucher signifies that the financial impact of an economic activity has been formally recognized and measured in accordance with accounting standards. For example, when a sales order is completed and the revenue recognition conditions are met, the financial system generates corresponding accounting vouchers, debiting "Accounts Receivable" or "Bank Deposits," and crediting accounts such as "Main Business Revenue" and "Taxes Payable—Value-Added Tax (Output Tax)," accurately recording the revenue amount, time of occurrence, related taxes, and impact on assets and liabilities. Therefore, while capturing order status change event data from the critical business system, it is necessary to capture voucher generation event data from the financial system in real time and accurately.

[0029] In one specific embodiment, capturing voucher generation event data from the financial system includes: First, accurately locating and accessing the enterprise's financial system, which includes the financial module within an Enterprise Resource Planning (ERP) system or specialized financial software. Next, selecting a data capture technology compatible with the financial system. In one example, standard APIs or extended development interfaces provided by the financial system can be utilized. It should be understood that current ERP systems support event-driven architectures, allowing external systems to subscribe to specific events, such as the "voucher posting" event. When the financial system successfully generates and saves a new voucher, a data packet containing key voucher information can be pushed to the data receiving service of the business-finance integration system in real time via a pre-defined webhook mechanism. Another API approach is polling, where the business-finance integration system periodically calls the financial system's query interface to obtain newly added or changed voucher data within a specified time period.

[0030] In other examples of this application, database-level integration techniques can be considered when APIs cannot be directly utilized or when real-time requirements are extremely high. If the environment permits and the risks are manageable, database triggers can be set on relevant voucher tables (such as voucher header tables and voucher line item tables) in the financial system database. When a new voucher is generated, the trigger automatically executes predefined logic to extract the core data of the new voucher (such as voucher ID, voucher date, posting date, voucher type, accounting subject code, summary, debit amount, credit amount, currency, auxiliary accounting item information, etc.) and write it to an intermediate data table or send it directly to a message queue.

[0031] Specifically, in step S3, the order status change event data and the voucher generation event data are cleaned and standardized to obtain cleaned and standardized order status change event data and cleaned and standardized voucher generation event data. It should be understood that critical business systems and financial systems are often built and operated independently, each following different data models, data entry standards, and historical evolution paths. This directly results in the raw order status change event data and voucher generation event data captured directly from these sources inevitably containing various noises or inaccurate information, exhibiting significant heterogeneity.

[0032] Specifically, the data may contain missing values; for example, some order records may lack crucial product IDs or customer information, or voucher data may omit necessary auxiliary accounting items. Data formats may vary widely, timestamp representations may be inconsistent, and amount fields may contain a mix of currency symbols, thousands separators, and even text descriptions. The data content may also contain errors, such as spelling mistakes during entry, illogical values ​​(e.g., negative amounts), and obsolete or inconsistent coding (e.g., old customer codes coexisting with new system codes). Furthermore, due to issues with system interfaces or data synchronization mechanisms, duplicate data records may occur. If this raw and unprocessed data is directly fed into the subsequent business and financial event correlation engine without processing, the consequences will be disastrous.

[0033] In a specific embodiment, data cleaning includes: filling in missing key information, handling inconsistent formatting, cleaning text data, correcting erroneous values, and identifying and processing duplicate data. Specifically, for missing values, strategies can be configured according to business logic. For non-critical information, the record or field can be ignored. Records missing key identifiers (such as order IDs) may be directly deemed invalid and removed. For numerical data (such as amounts) or timestamps, default values, mean / median / mode values ​​may be used for filling, or logical calculations based on relevant fields may be used, such as calculating the total amount based on the product unit price and quantity. For inconsistent formatting, such as date and time, parsing logic is written to recognize various common date formats and convert them uniformly to the standard ISO 8601 format. Processing the amount field requires removing currency symbols and thousands separators and converting it to a standard numeric type, while handling possible differences in decimal places. Text data cleaning includes removing leading and trailing spaces, converting full-width and half-width characters, handling special escape characters, and even performing basic spelling correction. Here, predefined dictionaries or simple edit distance algorithms can be used. Error correction relies on preset validation rules, such as the amount not being negative and specific codes existing in the master data list. Data that does not conform to these rules can be flagged, isolated for review, or automatically corrected according to the rules. Duplicate data identification and processing are based on unique identifiers or key field combinations of events for detection, and deduplication is performed according to a strategy. This deduplication strategy includes retaining the most recent record.

[0034] In a specific implementation, data standardization transforms cleaned data into a unified structure and representation, conforming to the input format expected by subsequent processing steps. This first requires defining a target data model, or standard schema, for order status change event data and voucher generation event data respectively. For example, a standardized order status change event data model should include clearly defined fields such as `order_id` (string type, unique identifier), `customer_id` (string or integer type), `product_id` (string or integer type), `amount` (numeric type), `timestamp` (standard date and time format), and `event_type` (enumeration or string type, such as CREATED, SHIPPED, COMPLETED, etc.). Similarly, a standardized voucher generation event data model should include `voucher_id` (string type, unique identifier), `account_code` (string type), `amount` (numeric type), `direction` (enumeration or character type, such as 'D' for debit, 'C' for credit), and `timestamp` (standard date and time format). The standardization process involves mapping and populating raw data from different source systems, with different field names, data types, or encoding schemes, into two standard models. For example, the order amount field in source system A might be named `TotalValue`, while in source system B it might be named `OrderAmt`; both need to be mapped to the `amount` field in the standard model. If multiple currencies are involved, the standardization process should include currency conversion logic, uniformly converting all amounts to a specified base currency, and potentially adding `original_currency` and `exchange_rate` fields to the model to preserve the original information. For coded information, such as customer IDs or product IDs, if there are inconsistencies in encoding across systems, the standardization process needs to call the Master Data Management (MDM) service or maintain an internal mapping table to convert the local IDs of each system into a globally unified master data ID. Similarly, for summary text in business event types or financial documents, normalization may be required, such as standardizing various texts indicating shipment ("Shipped", "Dispatched", "Out for Delivery") to "SHIPPED". It is worth mentioning that in other implementation examples of this application, data standardization can also be performed in other ways, such as Z-score standardization, and this is not limited to this application.

[0035] Specifically, in step S4, the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are input into the business and finance event correlation engine to obtain correlation analysis results. It should be understood that after capturing and completing the initial data cleaning and standardization, this application will obtain two relatively clean and uniformly formatted data streams: one reflecting order status changes at the business operation level, and the other recording voucher generation at the financial accounting level. However, these two sets of data are still independent at this point, and they are not correlated. Without this correlation, the data is merely a collection of data, unable to form a coherent business and financial view, let alone in-depth fusion analysis, risk monitoring, or intelligent decision support. Therefore, this application inputs the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data into the business and finance event correlation engine to obtain correlation analysis results.

[0036] Specifically, the business and financial event association engine can identify and utilize potential correlation clues existing in both types of data, thereby automatically establishing connections between business events and financial records. Furthermore, considering the complexity of real-world application scenarios, such as the lack of explicit strong correlation keys in some cases, the business and financial event association engine is also equipped with intelligent correlation analysis capabilities. This allows for deep data correlation based on multiple dimensions, ensuring that the most likely match is found even in the absence of direct identifiers.

[0037] In one embodiment, inputting the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data into the business and finance event association engine to obtain association analysis results includes: inputting the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data into the strong key association analysis unit of the business and finance event association engine to obtain first-level association analysis results.

[0038] In one embodiment, such as Figure 2 As shown, the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are input into the strong key association analysis unit of the business and finance event association engine to obtain the first-level association analysis result, including: S41, the strong key association analysis unit checks whether a predefined strong association key exists simultaneously in the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data; S42, if it exists, the first-level association analysis result is a successful association; S43, if it does not exist, the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are input into the intelligent association analysis unit of the business and finance event association engine to obtain the second-level association analysis result.

[0039] Specifically, after the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are input into the business event association engine, they are first processed by the strong key association analysis unit. The strong key association analysis unit checks whether a predefined strong association key exists simultaneously in the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data, based on a pre-configured rule base. The rule base specifies which field combinations appear simultaneously in both events and have completely equal values, and these combinations can be considered strong association keys. For example, a rule might be defined as: if the summary field of a voucher event contains an order number in a specific format, and this order number is exactly the same as the order ID field value of an order event, then a strong association is formed. Of course, this is just an example; those skilled in the art can define rules in the rule base according to actual circumstances. The strong key association analysis unit strictly checks the input data pairs according to these rules. If any strong association rule is found to be satisfied, in a specific example, the ID of order event O123 is O123, and the summary field of voucher event V456 happens to record "Payment: Order O123", then the strong key match is successful. At this point, the first-level association analysis result will immediately output a successful association as the first-level association analysis result, and this result will be passed to the subsequent processing module. The association analysis process for this pair of event data is now complete.

[0040] If the strong key association analysis unit fails to find any strong association keys that meet the conditions after traversing all rules, it will determine that the first-level association has failed. At this time, the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data will not be discarded, but will be input into the intelligent association analysis unit of the business and finance event association engine to start the second-level association analysis.

[0041] In one embodiment, such as Figure 3As shown, the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are input into the intelligent correlation analysis unit of the business and finance event correlation engine to obtain secondary correlation analysis results, including: S431, determining the time proximity between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain a time proximity score; S432, determining the amount similarity between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain an amount similarity score; S433, determining the amount similarity between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data. S434. Determine the consistency of measurement units between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain a measurement unit consistency score; S435. Determine the text similarity between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain a text similarity score; S436. Generate the secondary association analysis result based on the time proximity score, the amount similarity score, the entity association score, the measurement unit consistency score, and the text similarity score.

[0042] Specifically, after receiving the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data, the intelligent association analysis unit of the business and finance event association engine begins to calculate the similarity / consistency scores of five dimensions in parallel or serially.

[0043] The time proximity score is calculated by determining the time difference between the timestamp of the order event and the timestamp of the voucher event, and then mapping the time difference to a score. For example, the smaller the time difference, the higher the score. An exponential decay function or a piecewise function can be used for mapping.

[0044] The amount similarity score is calculated by comparing the relevant amounts in the order event with the amounts in the voucher event. Here, the relevant amounts include: the total order amount and the amount involved in this status change. Considering factors such as possible taxes, discounts, currency conversions, installment payments, etc., the comparison logic may be quite complex. For example, it may determine whether the two are absolutely equal, whether the relative error is within a threshold, and whether there is a multiple relationship, etc., and finally outputs an amount similarity score.

[0045] The entity relevance score is calculated by checking whether the core business entities involved in two events are consistent. These core business entities include customer ID, supplier ID, product ID, and project number. The relationship between entities can be determined using a master data management system or knowledge graph, such as the correspondence between a subsidiary and its parent company, or between product codes and material codes, and the entity relevance score is then output.

[0046] The consistency score for units of measurement is determined by verifying whether the units of measurement for quantities, service durations, etc., involved in the event are the same. For example, if an order records the sale of 10 units of a certain product and the voucher records the corresponding revenue, a higher score is given if the units are consistent, and a lower score is given if they are inconsistent, such as one using "piece" and the other "box," and there is no conversion relationship.

[0047] Text similarity scoring is achieved by performing semantic embedding encoding, converting the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data into semantic embedding encoding vectors for order status change events and voucher generation events, respectively. Then, the cosine similarity between the semantic embedding encoding vectors for order status change events and voucher generation events is calculated to obtain the text similarity score. That is, in one embodiment, as... Figure 4 As shown, determining the text similarity between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain a text similarity score includes: S4351, performing semantic embedding encoding on the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain a semantic embedding encoding vector for the order status change event and a semantic embedding encoding vector for the voucher generation event; S4352, calculating the cosine similarity between the semantic embedding encoding vector for the order status change event and the semantic embedding encoding vector for the voucher generation event as the text similarity score.

[0048] Finally, considering that the scores of a single dimension cannot independently determine the final association result, as they each reflect the strength of a certain aspect of the association probability, in actual business, there may be situations where the score of one dimension is low but other dimensions strongly corroborate the association, and vice versa. Therefore, a scoring model is introduced, using the scores of these five dimensions as input features. Through the model's internal calculation logic, a comprehensive data association score value is generated that can comprehensively reflect the confidence of the association between the event data. The calculated comprehensive data association score value is then compared with a preset threshold. If the comprehensive data association score value reaches or exceeds the preset threshold, the secondary association analysis result is output as a successful association; conversely, if the score value is lower than the preset threshold, the secondary association analysis result is output as a failed association. For example, the preset threshold is set to 0.85. Of course, the above is only an example, and the preset threshold can be set and adjusted according to actual conditions or experience.

[0049] In one embodiment, such as Figure 5 As shown, the secondary association analysis result is generated based on the time proximity score, the amount similarity score, the entity relevance score, the unit consistency score, and the text similarity score, including: S4361, inputting the time proximity score, the amount similarity score, the entity relevance score, the unit consistency score, and the text similarity score into the scoring model to obtain a comprehensive data association score; S4362, if the comprehensive data association score is greater than or equal to a preset threshold, the secondary association analysis result is a successful association; S4363, if the comprehensive data association score is less than the preset threshold, the secondary association analysis result is a failed association.

[0050] In one specific embodiment, the scoring model uses a weighted summation model. That is, a preset weight is assigned to each of the time proximity score, amount similarity score, entity relevance score, unit consistency score, and text similarity score. These weights represent the relative importance of each dimension in judging overall relevance. In one possible approach, the preset weights can be based on the experience of business experts. For example, in some business scenarios, precise matching of amount and entity is considered most critical, so the amount similarity score and entity relevance score can be given higher weights; while in other scenarios, time proximity and text matching may be more important, so the weights of the time proximity score and text similarity score can be increased accordingly. In a specific example, the weights for time proximity can be set to 0.15, amount similarity to 0.30, entity relevance to 0.25, unit consistency to 0.10, and text similarity to 0.20. Of course, the above is only an example, and the specific weights can be set and adjusted according to actual conditions or experience. In another possible approach, one could analyze historical data samples that have been confirmed to be related or not, and use statistical methods (such as correlation analysis, feature importance assessment) or simple optimization algorithms to determine a set of weight combinations that maximize the distinction between the two classes of samples.

[0051] In particular, when the text similarity score is obtained by calculating the cosine similarity of the embedded encoding vector, the time proximity score, the amount similarity score, the entity association score, and the unit of measurement consistency score will have weight differences from the text similarity score. That is, there will be inconsistencies in the coefficient mode in the weight allocation pattern, which will affect the calculation accuracy of the scoring model.

[0052] Based on this, if the time proximity score is... The amount similarity score The entity relevance score The consistency score of the units of measurement and the text similarity score If each of these is used as a single-model score, weighted commutation conformal processing is required to improve the accuracy of the weighted scoring. Specifically, in a preferred embodiment, the method further includes: before inputting the time proximity score, the monetary similarity score, the entity association score, the unit of measurement consistency score, and the text similarity score into the scoring model, weighted commutation conformal processing is performed on the time proximity score, the monetary similarity score, the entity association score, and the unit of measurement consistency score based on the text similarity score to obtain conformed time proximity scores, conformed monetary similarity scores, conformed entity association scores, and conformed unit of measurement consistency scores.

[0053] Specifically, based on the text similarity score, the temporal proximity score, the monetary similarity score, the entity association score, and the unit of measurement consistency score are weighted and commutated to obtain conformed temporal proximity score, conformed monetary similarity score, conformed entity association score, and conformed unit of measurement consistency score, including: First, a fractional-order guided operator is determined for the parameterization representation of weighted differences to enhance the text similarity score as a conformal objective. The specificity of the weight state is expressed as: ;in, This represents the time proximity score. This represents the similarity score of the amounts. This represents the entity relevance score. This indicates the consistency score of the unit of measurement. This represents the text similarity score. This represents a fractional guided operator.

[0054] Then, the time proximity score value The amount similarity score The entity relevance score Consistency score with the unit of measurement Single-modal distributed modulation is performed to obtain modulated time proximity scores, modulated amount similarity scores, modulated entity association scores, and modulated unit of measurement consistency scores. This single-modal distributed modulation is represented as follows: ;in, This represents the time proximity score. The amount similarity score The entity relevance score Consistency score with the unit of measurement The first in Each score value This represents the score among the following: modulated time proximity score, modulated amount similarity score, modulated entity association score, and modulated unit of measurement consistency score. Each modulated score value.

[0055] That is, to ensure the value The distribution guided by weighted differentiation satisfies the unimodal distribution property of a single mode.

[0056] Then, the conformal space of the interaction of the modulated weight states is defined as follows: ;in, This represents the conformal space of the combined effect of the modulated weight states.

[0057] Finally, based on the conformal space of the modulated weighted interaction and the text similarity score, the modulated temporal proximity score, the modulated monetary similarity score, the modulated entity association score, and the modulated unit consistency score are subjected to single-modal distribution modulation for conformal mapping at the single-modal fractional order to obtain the conformed temporal proximity score, the conformed monetary similarity score, the conformed entity association score, and the conformed unit consistency score. This facilitates the suppression of coefficient modal inconsistency through weighted conformal projection. ;in, This represents the first of the following: the conformed temporal proximity score, the conformed monetary similarity score, the conformed entity association score, and the conformed unit of measurement consistency score. Each conformal score value.

[0058] In this way, by controlling the target specificity sensitivity through fractional-order guiding operators, and by utilizing the unimodality of the single-mode distribution to maintain the mapping invariance of the coefficients themselves, and finally by using conformal space mapping, a similarity score metric with joint invariance of coefficient modes can be constructed under the constraint of weighted conformal projection, thereby improving the accuracy of weighted scoring.

[0059] Specifically, in step S5, if the association analysis result is a successful association, the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are marked as associated, and a fused event record containing the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data is generated. It should be understood that when the business and finance event association engine, after passing through the strong key association analysis unit or the intelligent association analysis unit, ultimately outputs a successful association result, this signifies that the system has a high degree of confidence, confirming that the examined pair of order events and voucher events indeed constitutes an entity interconnected at the business and financial levels. The marking and fused record generation operations are performed to achieve a closed loop of business and finance integration value and ensure the effectiveness of data management.

[0060] In one specific embodiment, the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are marked as associated, including: pre-setting a field specifically for indicating the association status. When the association is successful, a database update instruction is executed to update the association status field value of the two corresponding event records from the initial state (such as pending or null) to the specific associated status identifier.

[0061] In one specific embodiment, generating a fused event record containing the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data includes: First, predefining the data structure or schema of the fused event record. This structure should be designed to accommodate key information from both business and financial perspectives. It is not simply a matter of splicing two original records, but rather a selective and structured integration. Its fields may include: a unique identifier generated for the fused record; a reference to the original order event; a reference to the original voucher event; key business information extracted from the order status change event data, such as order number, customer information, product / service details, order amount, specific type of order status change, and timestamp of the transaction; and key financial information extracted from the voucher generation event data, such as voucher number, accounting subject, debit / credit direction, voucher amount, posting date, and summary information. Upon receiving a successful association signal and the corresponding order status change event data and voucher generation event data, the system instantiates a new record. Then, according to predefined mapping rules, the values ​​of relevant fields are read from the order status change event data and populated into the business attribute fields corresponding to the new record; similarly, the values ​​of relevant fields are read from the voucher generation event data and populated into the financial attribute fields corresponding to the new record. This newly generated record will be persistently stored in a specific area of ​​a database table, a fact table of a data warehouse, or a data lake specifically used to store integrated data.

[0062] It is worth mentioning that in other implementation examples of this application, a fused event record containing the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data can also be generated by simple splicing or other data fusion methods. This is not limited to this application.

[0063] Specifically, in step S6, if the association analysis result indicates that association cannot be successfully established, the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are marked as pending processing. It should be understood that although the business and financial event association engine is well-designed, integrating the determinism of strong key matching with the multi-dimensional evaluation capabilities of intelligent analysis, it may still fail to establish associations between certain event pairs for various reasons. These reasons may include: data quality issues; exceptionally complex business scenarios exceeding the scope covered by preset rules or models; and intelligent association scores below a preset threshold, where although weak associations may exist, the system chooses not to confirm them to ensure accuracy. Marking these unassociated data as pending processing separates them from the regular process and guides them to subsequent processing stages, such as manual review, matching with other data, or other processing methods.

[0064] In summary, the business-finance integration management method based on intelligent data analysis provided in this application first captures order status change event data and voucher generation event data from key business systems and financial systems respectively through automated means, ensuring the timeliness and accuracy of the raw data. Then, this data is cleaned and standardized to eliminate format differences and unify data standards. Subsequently, a business-finance event correlation engine is used to process the cleaned and standardized data to identify and establish the correlation between business events and financial vouchers. Data that is successfully correlated is marked as correlated and a fusion event record is generated; data that cannot be successfully correlated is marked as pending processing. This not only improves the accuracy and comprehensiveness of business-finance data integration but also enhances the flexibility and adaptability of data processing through intelligent means, providing enterprises with more powerful decision support.

[0065] This application also provides a business and finance integration management system based on intelligent data analysis, such as... Figure 6As shown, the business-finance integration management system 600 based on intelligent data analysis includes: an order status change event capture module 610, used to capture order status change event data from key business systems; a voucher generation event capture module 620, used to capture voucher generation event data from the financial system; a data cleaning module 630, used to clean and standardize the order status change event data and the voucher generation event data to obtain cleaned and standardized order status change event data and cleaned and standardized voucher generation event data; and a correlation analysis result generation module 640, used to generate correlation analysis results from the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data. The event data is input into the business and finance event association engine to obtain the association analysis results; the fusion event record generation module 650 is used to mark the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data as associated if the association analysis result is successful, and generate a fusion event record containing the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data; the data marking module 660 is used to mark the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data as pending processing if the association analysis result is unsuccessful.

[0066] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0067] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0068] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A business-finance integrated management method based on intelligent data analysis, characterized in that, include: Capture order status change event data from critical business systems; Capture vouchers from the financial system to generate event data; The order status change event data and the voucher generation event data are cleaned and standardized to obtain cleaned and standardized order status change event data and cleaned and standardized voucher generation event data. Input the standardized order status change event data and the standardized voucher generation event data into the business and finance event association engine to obtain the association analysis results; If the association analysis result is a successful association, the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are marked as associated, and a fused event record containing the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data is generated; If the association analysis result is that the association cannot be successfully linked, the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data will be marked as pending processing.

2. The business-finance integration management method based on intelligent data analysis according to claim 1, characterized in that, The standardized order status change event data and the standardized voucher generation event data are input into the business and finance event correlation engine to obtain correlation analysis results, including: The standardized order status change event data and the standardized voucher generation event data are input into the strong key association analysis unit of the business and finance event association engine to obtain the first-level association analysis results.

3. The business-finance integration management method based on intelligent data analysis according to claim 2, characterized in that, The cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are input into the strong key association analysis unit of the business and finance event association engine to obtain the first-level association analysis results, including: The strong key association analysis unit checks whether a predefined strong association key exists simultaneously in the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data. If it exists, then the result of the first-level association analysis is a successful association; If not, the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are input into the intelligent association analysis unit of the business and finance event association engine to obtain the secondary association analysis results.

4. The business-finance integration management method based on intelligent data analysis according to claim 3, characterized in that, The cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data are input into the intelligent correlation analysis unit of the business and finance event correlation engine to obtain secondary correlation analysis results, including: Determine the temporal proximity between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain a temporal proximity score. Determine the monetary similarity between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain a monetary similarity score. Determine the entity association between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain an entity association score. Determine the consistency of measurement units between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain a measurement unit consistency score. Determine the text similarity between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain a text similarity score; The secondary association analysis results are generated based on the time proximity score, the amount similarity score, the entity relevance score, the unit of measurement consistency score, and the text similarity score.

5. The business-finance integration management method based on intelligent data analysis according to claim 4, characterized in that, Based on the time proximity score, the monetary similarity score, the entity association score, the unit of measurement consistency score, and the text similarity score, the secondary association analysis results are generated, including: The time proximity score, the amount similarity score, the entity association score, the unit of measurement consistency score, and the text similarity score are input into the scoring model to obtain a comprehensive score for data association. If the comprehensive score of the data correlation is greater than or equal to the preset threshold, the secondary correlation analysis result is a successful correlation. If the comprehensive score of the data correlation is less than the preset threshold, the result of the secondary correlation analysis is that the correlation cannot be successfully established.

6. The business-finance integration management method based on intelligent data analysis according to claim 5, characterized in that, Also includes: Before inputting the time proximity score, the amount similarity score, the entity association score, the unit consistency score, and the text similarity score into the scoring model, a weighted commutation conformal process is performed on the time proximity score, the amount similarity score, the entity association score, and the unit consistency score based on the text similarity score to obtain the conformed time proximity score, the conformed amount similarity score, the conformed entity association score, and the conformed unit consistency score.

7. The business-finance integration management method based on intelligent data analysis according to claim 6, characterized in that, Based on the text similarity score, the temporal proximity score, the monetary similarity score, the entity association score, and the unit of measurement consistency score are weighted and commutated to obtain conformed temporal proximity score, conformed monetary similarity score, conformed entity association score, and conformed unit of measurement consistency score, including: Determine the fractional-order guiding operator for the parameterized representation of weighted differences; The time proximity score, the amount similarity score, the entity association score, and the unit of measurement consistency score are subjected to single-modal distribution modulation to obtain modulated time proximity score, modulated amount similarity score, modulated entity association score, and modulated unit of measurement consistency score. Define the conformal space for the interaction of modulated weight states; Based on the conformal space of the modulated weighted interaction and the text similarity score, the modulated temporal proximity score, the modulated monetary similarity score, the modulated entity association score, and the modulated unit consistency score are subjected to single-modal distribution modulation for conformal mapping at the single-modal fractional order to obtain the conformed temporal proximity score, the conformed monetary similarity score, the conformed entity association score, and the conformed unit consistency score.

8. The business-finance integration management method based on intelligent data analysis according to claim 4, characterized in that, Determining the text similarity between the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain a text similarity score includes: Semantic embedding encoding is performed on the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data to obtain the semantic embedding encoding vectors for order status change events and voucher generation events. The cosine similarity between the semantic embedding encoding vector of the order status change event and the semantic embedding encoding vector of the voucher generation event is calculated as the text similarity score.

9. The business-finance integration management method based on intelligent data analysis according to claim 1, characterized in that, The order status change event data includes order ID, customer ID, product ID, amount, timestamp, and event type. The voucher generation event data includes voucher ID, account code, amount, debit / credit direction, and timestamp.

10. A business-finance integration management system based on intelligent data analysis, used to execute the business-finance integration management method based on intelligent data analysis as described in any one of claims 1-9, characterized in that, include: The order status change event capture module is used to capture order status change event data from critical business systems; The voucher generation event capture module is used to capture voucher generation event data from the financial system; The data cleaning module is used to clean and standardize the order status change event data and the voucher generation event data to obtain cleaned and standardized order status change event data and cleaned and standardized voucher generation event data. The correlation analysis result generation module is used to input the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data into the business and finance event correlation engine to obtain correlation analysis results; The fusion event record generation module is used to mark the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data as associated if the association analysis result is a successful association, and generate a fusion event record containing the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data. The data tagging module is used to mark the cleaned and standardized order status change event data and the cleaned and standardized voucher generation event data as pending processing if the association analysis result is that the association cannot be successfully linked.

Images