Cross-validation processing method and apparatus for internal transactions

Abstract

The application discloses a kind of internal transaction cross-validation processing method and device, it is related to big data field, which comprises: extracting multiple types of transaction feature vectors from the internal transaction information of unfinished cross-validation to form a judgment condition matrix aggregation;When cross-validation is passed between the internal transactions of at least two transaction records according to the aggregation, the internal transaction of preliminary cross-validation is passed;According to the identification of preliminary cross-validation passing transaction, the data source image of preliminary verification passing internal transaction is retrieved from the source system database;After extracting key elements from the image, a cloud image measurement transaction element item set is formed, and the cloud image measurement transaction element item set of at least two transaction records of internal transaction is matched.If the matching is passed, the cloud map of the internal transaction of preliminary cross-validation passing is checked, and the internal transaction of preliminary cross-validation passing is marked and included in the marked internal transaction database after marking.The application can improve the accuracy and efficiency of cross-validation processing of internal transaction.

Description

Technical Field

[0001] The present invention relates to the field of big data technology, and in particular to a cross-verification processing method and device for internal transactions. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the invention that are recited in the claims. No statement herein is admitted to be prior art by virtue of its inclusion in this section.

[0003] Insider transactions refer to the transfer of resources, services, obligations, or other benefits between corporate entities at all levels and their subsidiaries, or between subsidiaries themselves. Internal transactions involve entities involved in interdependent relationships involving resources, services, obligations, or other benefits: natural persons, legal entities, or unincorporated organizations. Internal transactions can identify risky transactions involving interdependent relationships involving resources, services, obligations, or other benefits.

[0004] Regulatory requirements require all corporate entities and subsidiaries at all levels to identify, report, and manage their internal transactions, and to disclose information uniformly in accordance with consolidated financial statements. Therefore, efficient and accurate verification of internal transactions is crucial. Existing technologies primarily employ a combination of automated data capture and manual verification, as follows:

[0005] After the system automatically extracts transaction data, it determines whether it is an internal transaction through the counterparty. If the system does not automatically extract the transaction, it manually identifies and re-enters the internal transaction. Due to defects and issues in system data extraction and manual identification, incomplete transaction data recognition can occur, leading to errors and omissions in the identification and reporting of internal transactions by corporate entities and subsidiaries. Therefore, to ensure the completeness and accuracy of internal transaction data, corporate entities and subsidiaries will forward their reported internal transactions to the counterparty. The counterparty will then determine whether a specific internal transaction has been fully and accurately reported by the counterparty and whether the reported content is consistent. If the counterparty determines that a specific internal transaction has been fully and accurately reported by the counterparty, the counterparty can use the system's cross-verification function to check a match, indicating that the reported content of both parties is consistent. If the counterparty is unable to check a match, it indicates that at least one of the parties to the transaction has made an error or omission in the identification and reporting of the internal transaction, requiring verification and re-reporting.

[0006] Currently, the work of pushing internal transactions to corresponding institutions or subsidiaries relies on manual processing every six months. Not only is the workload of sorting and issuing large amounts of data, but it can also lead to human errors, and even the timeliness of reporting cannot be guaranteed.

[0007] Therefore, the existing internal transaction cross-verification method is highly dependent on manual processing. Since manual cross-verification is labor-intensive, inefficient, and highly dependent on the professionalism and sense of responsibility of the handling personnel themselves, it is easy to have missed matches, matching errors, etc. Summary of the Invention

[0008] An embodiment of the present invention provides a cross-validation processing method for internal transactions, for improving the accuracy and efficiency of cross-validation processing for internal transactions. The method includes:

[0009] obtaining internal transaction information that has not completed cross-validation from an internal transaction database; the internal transaction database is an internal transaction database pre-established based on internal transaction information obtained from multiple source system databases;

[0010] Extracting various types of transaction feature vectors from the internal transaction information that has not completed cross-validation to form a judgment condition matrix aggregation;

[0011] When it is determined that the cross-validation between the internal transactions recorded by at least two transaction parties passes according to the judgment condition matrix aggregation, the internal transactions that have passed the preliminary cross-validation are obtained;

[0012] According to the identifier of the internal transaction that has passed the preliminary cross-verification, the data source image of the internal transaction that has passed the preliminary cross-verification is retrieved from the source system database;

[0013] After extracting key elements from the data source image, a cloud image measurement transaction element item set is formed. The cloud image measurement transaction element item sets corresponding to the internal transactions recorded by at least two transaction parties are matched. If the match is successful, the internal transaction that has passed the preliminary cross-validation of the cloud image verification is obtained;

[0014] The internal transactions that have passed the preliminary cross-validation will be marked as internal transactions that have completed cross-validation and included in the marked internal transaction database.

[0015] An embodiment of the present invention further provides a cross-validation processing device for internal transactions, for improving the accuracy and efficiency of cross-validation processing for internal transactions. The device includes:

[0016] an acquiring unit, configured to acquire internal transaction information that has not completed cross-validation from an internal transaction database; the internal transaction database is a pre-established internal transaction database based on internal transaction information acquired from multiple source system databases;

[0017] A matrix aggregation construction unit is used to extract various types of transaction feature vectors from internal transaction information that has not completed cross-validation to form a judgment condition matrix aggregation;

[0018] A preliminary cross-validation unit, configured to obtain an internal transaction that has passed the preliminary cross-validation when it is determined that cross-validation between internal transactions recorded by at least two transaction parties has passed according to the judgment condition matrix aggregation;

[0019] The backtracking image unit is used to retrieve the data source image of the internal transaction that has passed the preliminary cross-verification from the source system database according to the identifier of the internal transaction that has passed the preliminary cross-verification;

[0020] A cloud image metrology processing unit is configured to extract key elements from the data source image to form a cloud image metrology transaction element item set, match the cloud image metrology transaction element item sets corresponding to internal transactions recorded by at least two transaction parties, and obtain an internal transaction that has passed preliminary cross-validation of the cloud image verification if a match is successful;

[0021] The cross-validation unit is optimized to mark internal transactions that have passed the preliminary cross-validation as completed cross-validation internal transactions and include them in the marked internal transaction database.

[0022] An embodiment of the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned cross-verification processing method for internal transactions when executing the computer program.

[0023] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the computer program implements the cross-verification processing method for internal transactions.

[0024] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the cross-verification processing method for internal transactions.

[0025] Compared with the technical solutions in the prior art that combine automatic data acquisition and manual identification to perform internal transaction cross-validation, which have low efficiency and accuracy, the cross-validation processing solution for internal transactions provided by the embodiments of the present invention has the following beneficial technical effects: an internal transaction database is pre-established from internal transaction information obtained from multiple source system databases, and internal transaction information that has not completed cross-validation is obtained from the internal transaction database; multiple types of transaction feature vectors are extracted from the internal transaction information that has not completed cross-validation to form a judgment condition matrix aggregation; based on the judgment condition matrix aggregation and the extraction of key elements from the data source image, a cloud image metering transaction element item set is formed to perform cross-validation of internal transactions, which can achieve efficient and accurate cross-validation of internal transactions and improve the accuracy and efficiency of cross-validation processing of internal transactions. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0027] Figure 1 Schematic diagram of the process of cross-verification processing of internal transactions in an embodiment of the present invention;

[0028] Figure 2 Schematic diagram of a flow chart of a cross-verification processing method for internal transactions according to another embodiment of the present invention;

[0029] Figure 3 Schematic diagram of the pretreatment method according to an embodiment of the present invention;

[0030] Figure 4 Schematic diagram of the process of marking judgment and table output in an embodiment of the present invention;

[0031] Figure 5 Schematic diagram of a flow chart of a cross-verification processing method for internal transactions according to another embodiment of the present invention;

[0032] Figure 6 Schematic diagram of a flow chart of a cross-verification processing method for internal transactions according to another embodiment of the present invention;

[0033] Figure 7 Schematic diagram of the process of preliminary cross-validation and cloud graph verification processing method in an embodiment of the present invention;

[0034] Figure 8 Schematic diagram of the process of cloud graph restoration processing method for internal transactions in an embodiment of the present invention;

[0035] Figure 9 Schematic diagram of the structure of the cross-verification processing device for internal transactions in an embodiment of the present invention. DETAILED DESCRIPTION

[0036] To make the purpose, technical solutions and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below with reference to the accompanying drawings. Here, the exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0037] The acquisition, transmission, storage, use, and processing of data in the technical solution of this application comply with the relevant provisions of national laws and regulations.

[0038] It should be noted that in the embodiments of the present application, certain software, components, models and other existing solutions in the industry may be mentioned. They should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of the present application, but it does not mean that the applicant has or will necessarily use the solution.

[0039] Figure 1 FIG. 1 is a flow chart of a cross-verification processing method for internal transactions according to an embodiment of the present invention. Figure 1 As shown, the method includes the following steps:

[0040] Step 101: Obtaining internal transaction information that has not completed cross-validation from an internal transaction database; the internal transaction database is an internal transaction database pre-established based on internal transaction information obtained from multiple source system databases;

[0041] Step 102: extracting multiple types of transaction feature vectors from the internal transaction information that has not completed cross-validation to form a judgment condition matrix aggregation;

[0042] Step 103: When it is determined that the cross-validation between the internal transactions recorded by at least two transaction parties has passed according to the judgment condition matrix aggregation, the internal transactions that have passed the preliminary cross-validation are obtained;

[0043] Step 104: According to the identifier of the internal transaction that has passed the preliminary cross-verification, retrieve the data source image of the internal transaction that has passed the preliminary cross-verification from the source system database;

[0044] Step 105: After extracting key elements from the data source image, a cloud image measurement transaction element item set is formed. The cloud image measurement transaction element item sets corresponding to the internal transactions recorded by at least two transaction parties are matched. If the match succeeds, the internal transaction that has passed the preliminary cross-validation of the cloud image verification is obtained.

[0045] Step 106: Mark the internal transactions that have passed the preliminary cross-validation as internal transactions that have completed cross-validation and then add them into the marked internal transaction database.

[0046] Compared with the existing technical solutions that combine automatic data acquisition and manual identification to perform internal transaction cross-verification, which have low efficiency and accuracy, the beneficial technical effects of the internal transaction cross-verification processing method provided by the embodiment of the present invention are: pre-establishing an internal transaction database from internal transaction information obtained from multiple source system databases, obtaining internal transaction information that has not completed cross-verification from the internal transaction database; extracting multiple types of transaction feature vectors from the internal transaction information that has not completed cross-verification to form a judgment condition matrix aggregation; performing cross-verification of internal transactions based on the judgment condition matrix aggregation and the extraction of key elements from the data source image to form a cloud image measurement transaction element item set, which can achieve efficient and accurate cross-verification of internal transactions and improve the accuracy and efficiency of cross-verification processing of internal transactions. Figures 2 to 9 The cross-verification processing method for internal transactions is introduced in detail.

[0047] In order to facilitate understanding of the overall concept of the present invention, first, the overall solution of the embodiment of the present invention is introduced.

[0048] In the pre-processing step of the embodiment of the present invention, it can be as follows Figure 2 The steps implemented in the "pre-processing module" are: Figure 2 FIG. 1 is a flow chart of a cross-verification processing method for internal transactions in another embodiment of the present invention. Figure 2 After obtaining relatively complete internal transactions, the "accounting system, business system and other source system databases of Institution A" and "accounting system, business and other source system databases of subsidiary company B" shown in the table are stored in layers and databases according to different transaction types, and then aggregated into an internal transaction database, such as Figure 2 The "internal transaction database" shown in the figure is a pre-established internal transaction database based on internal transaction information obtained from multiple source system databases, and internal transaction identification is regularly rechecked to improve the efficiency of subsequent cross-validation work and the accuracy of matching. This pre-processing step can be done as follows: Figure 3 The preprocessing module process shown is implemented as follows, Figure 3 Schematic diagram of the pretreatment method in an embodiment of the present invention.

[0049] In the above step 101, when obtaining the internal transaction information that has not completed cross-validation from the internal transaction database, first, Figure 2 The "marking judgment module" shown in FIG. 1 judges whether the internal transaction obtained is an unmarked transaction, that is, whether it is an internal transaction that has not completed cross-validation. This step 101 can be performed as follows: Figure 4 The marking judgment module process shown is implemented. Figure 4 Schematic diagram of the process of marking judgment and table output in an embodiment of the present invention.

[0050] The above steps 102 and 103 can be performed by Figure 2 The cross-validation module shown in the figure is used to realize this. That is, through the judgment condition matrix aggregation algorithm, the system automatically determines whether the cross-validation between internal transactions has passed. This can greatly reduce the workload of the existing system combining data acquisition and manual cross-validation of internal transactions, and improve matching efficiency.

[0051] The above steps 104 and 105 can be performed as follows Figure 2 The data cloud graph verification module shown is used to achieve this, that is, by tracing back the image information of the source system, extracting key elements and matching them again to verify whether the cross-validation results are correct. This can improve matching accuracy, reduce the workload of manual inspection, and improve the efficiency of internal transaction cross-validation.

[0052] The above step 106 is as follows Figure 2 "Verification passed, marking" is shown.

[0053] In order to facilitate understanding of the present invention, it is described in detail below with reference to examples.

[0054] 1) Each level of the enterprise and its subsidiaries capture internal transaction data from multiple source system databases, such as financial accounting systems, front-end business systems, and other source systems (public credit systems, financial market systems, interbank asset management systems, financial systems, deposit systems, and centralized procurement systems), and aggregate them into a source database. Figure 3 As shown, the source database runs a pre-configured preprocessing module. This module uses algorithms such as cleaning non-internal transactions, radar calculations, and neural network modeling to supplement key elements. After obtaining relatively complete internal transactions, these are stored in separate databases based on transaction types and aggregated to form an internal transaction database. The purpose of preprocessing is to improve the efficiency of subsequent cross-validation.

[0055] In addition, the pre-processing module can also be used to regularly re-check transaction data in the source database that has not been identified as internal transactions to determine whether the previous identification results are incorrect, thereby improving the accuracy of internal transaction cross-matching.

[0056] 2) If Figure 4 As shown, the marking judgment module screens, sorts, and marks the internal transactions in the internal transaction database on a daily basis. If the screening finds that an internal transaction has been marked, that is, the internal transaction has completed cross-validation, then the transaction will be entered into the marked internal transaction database and await further processing by the subsequent consolidated data output and disclosure module. If the screening finds that an internal transaction has not been marked, that is, the internal transaction has not completed cross-validation, then the cross-validation module will be executed for the transaction.

[0057] 3) the connotation of the cross-validation module is that 9 types of transaction feature vectors are extracted from the unmarked internal transaction information, which can be respectively: vector 1: amount vector (including transaction contract amount, transaction occurrence amount, transaction balance), vector 2: price vector (transaction price-amount, transaction price-non-amount), vector 3: transaction borrower direction vector, vector 4: transaction date difference vector, vector 5: transaction contract number vector, vector 6: transaction abstract vector, vector 7: COA subject number vector, vector 8: transaction target vector, and vector 9: currency vector (including transaction currency, transaction price currency). The 9 types of vectors are combined in various ways to form matrices, and after disambiguation, a first layer matrix group is formed. In the first layer matrix group, each two matrices are combined according to matrix operations (operation methods include addition, subtraction, multiplication, transposition, conjugate, conjugate transposition, etc.) to form a new matrix, and after disambiguation, a second layer matrix group is formed. In this way, after multiple layers of matrix group aggregation and disambiguation, the judgment condition matrix aggregation is formed, which is the cross-validation module. In specific implementation, the disambiguation processing in the embodiment of the present application refers to eliminating the text that the machine cannot explain or understand, solving the language ambiguity, i.e., performing multi-level language analysis and processing ambiguity on the matrix, improving the accuracy of the judgment condition matrix aggregation, and further improving the accuracy of the subsequent internal transaction cross-validation.

[0058] As known from the above, in one embodiment, Figure 5 The flowchart of the internal transaction cross-validation processing method in another embodiment of the present application is shown in Figure 5 As shown, a plurality of types of transaction feature vectors are extracted from the internal transaction information that has not completed cross-validation to form a judgment condition matrix aggregation, which can include the following steps:

[0059] Step 201: a plurality of types of transaction feature vectors are extracted from the internal transaction information that has not completed cross-validation;

[0060] Step 202: the plurality of types of transaction feature vectors are combined in different ways to form matrices;

[0061] Step 203: the matrices combined in different ways are disambiguated to form a first layer matrix group;

[0062] Step 204: taking the first layer matrix group as an initial layer matrix group, the following steps for forming a judgment condition matrix aggregation are repeatedly executed, and the following operations are performed in each cycle:

[0063] Step 2041: obtaining a current cycle layer matrix group;

[0064] Step 2042: combining each two matrices in the current cycle layer matrix group according to matrix operations to form a new matrix, and combining a plurality of new matrices after disambiguation to form a next cycle layer matrix group;

[0065] Step 2043: when the next cycle layer matrix group does not satisfy the preset termination condition, performing a next cycle operation on the next cycle layer matrix group as the current cycle layer matrix group until the next cycle layer matrix group satisfies the preset termination condition, and obtaining the judgment condition matrix aggregation.

[0066] In specific implementation, the implementation manner of the judgment condition matrix aggregation can further improve the efficiency and accuracy of cross-validation.

[0067] If the cross-validation passes, it indicates that the preliminary matching of the two or more internal transactions passes, and the data cloud image verification module is continuously run to verify whether the secondary matching passes. If the cross-validation does not pass, it indicates that the preliminary matching of the two or more internal transactions does not pass, and the data cloud image restoration module is run.

[0068] 4) The data cloud image verification module includes the following: the internal transactions passing the cross-validation are respectively traced back to the source system such as the financial accounting system or the front-end business system according to the invoice number or the business voucher number, and the data source image (including the invoice image, the business voucher image, the contract image, etc.) is called. The source image is analyzed by the cloud computing algorithm, the key fields (elements) are extracted, and the cloud image measurement transaction element item set is formed. The cloud image measurement transaction element item set is matched again to verify whether the previous cross-validation is accurate.

[0069] In specific implementation, in order to facilitate understanding of how to implement, the following takes an example of the cloud computing algorithm and the cloud image measurement transaction element item set: the data source image is abstracted into a cloud virtual resource vector, the element set formed by extracting the key elements from the cloud virtual resource vector is the cloud image measurement transaction element item set, the resource utilization rate can be improved, the internal performance of the computer running the internal transaction cross-validation method of the application can be improved, the running efficiency can be improved, and the efficiency and accuracy of the internal transaction cross-validation can be further improved.

[0070] If the internal transaction passing the cross-validation is verified again by the data cloud image verification module, it indicates that the secondary matching of the internal transaction is consistent and correct, and the confidence degree of the matching is high, so that the internal transaction can be marked and included in the marked internal transaction database, and further processed by the subsequent consolidated data output disclosure module. If the internal transaction passing the cross-validation is not verified by the data cloud image verification module, it indicates that the preliminary matching result is incorrect, and the internal transaction is run by the data cloud image restoration module.

[0071] 5) Run the Consolidated Data Output and Disclosure Module on the marked internal transactions included in the marked internal transaction database through steps 2) or 4). This module screens all marked internal transactions for transactions with non-zero debit amounts and generates unilateral consolidated data for subsequent reporting of group consolidation management information to regulatory authorities and external disclosure of group-related information. The consolidated data generated by running the Consolidated Data Output and Disclosure Module eliminates duplicate reporting and disclosure issues, improving the accuracy of consolidated data.

[0072] 6) For internal transactions that fail step 3) or 4), run the data cloud map restoration module. The connotation of the data cloud map restoration module is: internal transactions are traced back to the source system such as the financial accounting system or the front-end business system according to the invoice number or business voucher number, and the data source image (including invoice image, business voucher image, contract image, etc.) is retrieved. The source image is analyzed by cloud computing and other algorithms to extract key elements (which may include: counterparty, counterparty group, transaction contract number, transaction contract amount, transaction amount, transaction balance, transaction price-amount, transaction price-non-amount, transaction date, transaction subject, transaction summary, COA account number, transaction currency, transaction price currency, etc.) and then build the cloud image measurement transaction element knowledge graph. By running the regression function, the edge features of the cloud image measurement transaction element knowledge graph can be restored to "regression internal transaction", and three matching collision operations are performed in sequence. The three matching collision operations are detailed in Figure 8 introduce.

[0073] The first matching collision is based on the internal transactions of Institution A (the first transaction party, the transaction initiator) that failed the cross-validation module and / or the data cloud graph verification module. The internal transactions of Subsidiary B (the second transaction party, the transaction counterparty) that failed the cross-validation module or the data cloud graph verification module are then run through the data cloud graph restoration module to create the "regressed internal transactions" and then matched against them using a certain logic (which can be random logic) to test whether the matching collision passes. If the matching collision passes, the human-machine training module is run; if not, the second matching collision is run.

[0074] The second matching collision is based on the internal transactions of Subsidiary B that failed the cross-validation module or the data cloud graph verification module. The matching collision is performed using a certain logic (which can be random logic) with the "regressed internal transactions" of Institution A that failed the cross-validation module or the data cloud graph verification module. If the matching collision passes, the human-machine training module is executed; if not, the third matching collision is executed.

[0075] The third collision match is based on the "returned internal transactions" of Institution A, which were restored after running the data cloud restoration module, based on the internal transactions of Institution A that failed the cross-validation module or the data cloud verification module. A collision match is then performed using a specific logic (possibly random logic) to test whether the collision match succeeds. If the collision match succeeds, the human-machine training module is run; if not, the internal transactions are temporarily stored in the retained internal transaction database.

[0076] From the above, it can be seen that in one embodiment, Figure 6 FIG. 1 is a flow chart of a cross-verification processing method for internal transactions in another embodiment of the present invention. Figure 6 As shown, the cross-validation processing method for internal transactions may further include the following cloud map restoration and matching operation steps for internal transactions that fail preliminary cross-validation (internal transactions that fail the cross-validation module) and for verifying internal transactions that fail preliminary cross-validation (internal transactions that fail the data cloud map verification module):

[0077] Step 301: when it is determined that the cross-validation between the internal transactions recorded by at least two transaction parties fails according to the judgment condition matrix aggregation, obtaining the internal transactions that failed the preliminary cross-validation;

[0078] Step 302: When the cloud image metering transaction element item sets corresponding to the internal transactions recorded by at least two transaction parties fail to match, the internal transactions that fail the preliminary cross-validation are obtained;

[0079] Step 303: For each internal transaction belonging to the transaction initiator that fails the preliminary cross-validation and the internal transaction that fails the preliminary cross-validation, retrieve the data source image of each transaction initiator from the source system database.

[0080] Step 304: For each internal transaction belonging to the counterparty that failed the preliminary cross-validation and the internal transaction that failed the preliminary cross-validation, retrieve the data source image of each internal transaction belonging to the counterparty that failed the preliminary cross-validation and the internal transaction that failed the preliminary cross-validation from the source system database;

[0081] Step 305: extract key elements from the data source image of each transaction initiator and construct a knowledge graph of cloud image measurement transaction elements for each transaction initiator;

[0082] Step 306: Extract key elements from the data source image of each transaction counterparty and construct a knowledge graph of cloud image measurement transaction elements for each transaction counterparty;

[0083] Step 307: Restore each edge feature of the cloud image measurement transaction element knowledge graph of each transaction initiator to the regressed internal transaction of each transaction initiator;

[0084] Step 308: Restore each edge feature of the cloud image measurement transaction element knowledge graph of each transaction counterparty to the regressed internal transaction of each transaction counterparty;

[0085] Step 309: Based on the internal transactions that failed the initial cross-validation and the internal transactions that failed the verification initial cross-validation belonging to the transaction initiator, the internal transactions that failed the initial cross-validation and the internal transactions that failed the verification initial cross-validation belonging to the transaction counterparty, the regression internal transactions of each transaction initiator, and the regression internal transactions of each transaction counterparty, matching processing is performed to obtain the internal transactions that finally passed the cross-validation.

[0086] In specific implementation, the cloud graph restoration and matching operations for internal transactions that fail the initial cross-validation and the verification of internal transactions that fail the initial cross-validation can further improve the accuracy of internal transaction cross-validation.

[0087] As can be seen from the above, in one embodiment, matching is performed based on the internal transactions belonging to the transaction initiator that failed the initial cross-validation and the internal transactions that failed the verification initial cross-validation, the internal transactions belonging to the transaction counterparty that failed the initial cross-validation and the internal transactions that failed the verification initial cross-validation, the regression internal transactions of each transaction initiator, and the regression internal transactions of each transaction counterparty to obtain the internal transactions that finally passed the cross-validation, which may include:

[0088] Based on the internal transactions of the transaction initiator that failed the initial cross-validation and the internal transactions that failed the initial cross-validation, a first match is performed with the regressed internal transactions of each transaction counterparty. When the first match is successful, the internal transactions that finally passed the cross-validation are obtained.

[0089] In specific implementation, the above-mentioned first matching operation for internal transactions that fail the initial cross-validation and the verification of internal transactions that fail the initial cross-validation can further improve the accuracy of internal transaction cross-validation.

[0090] As can be seen from the above, in one embodiment, the cross-verification processing method for internal transactions may further include:

[0091] When the first matching fails, the internal transactions that failed the initial cross-validation and the internal transactions that failed the initial cross-validation belonging to the counterparty are used as the basis for a second matching with the regressed internal transactions of each transaction initiator. When the second matching passes, the internal transactions that finally passed the cross-validation are obtained.

[0092] In specific implementation, the above-mentioned second matching operation for internal transactions that failed the initial cross-validation and the verification of internal transactions that failed the initial cross-validation can further improve the accuracy of internal transaction cross-validation.

[0093] From the above, it can be seen that in one embodiment, the cross-verification processing method of the internal transaction can also include:

[0094] If the second matching fails, the third matching is performed based on the regressed internal transaction of each transaction initiator and the regressed internal transaction of each transaction counterparty. If the third matching passes, the final cross-validated internal transaction is obtained.

[0095] In specific implementation, the third matching operation for internal transactions that fail the initial cross-validation and the verification of internal transactions that fail the initial cross-validation can further improve the accuracy of internal transaction cross-validation.

[0096] In specific implementation, steps 301 to 309 can be performed as follows Figure 2 The data cloud map restoration module shown is used to achieve this, that is, by tracing back the image information of the source system, extracting key elements and building a cloud image measurement transaction element knowledge graph, and running the preset regression function to match and collide internal transactions that have not passed cross-validation or cloud map verification to increase the matching success rate.

[0097] The detained internal transaction database is judged once a day, and the judgment condition is "Is the data loading date of the detained internal transaction database close to the reporting and regulatory deadline?" If it is judged to be close, a risk reminder will be pushed to the two trading entities of the detained internal transaction, and the reminder content will be "The internal transaction with contract number XXX between your organization and XXX counterparty is suspected of misreporting / omission. The automatic cross-validation has not been completed within the XXX time limit. Please handle it in time." If it is not close, the detained transaction will be merged with the internal transactions that failed the cross-validation module or data cloud map verification module on the next day, and the data cloud map restoration module will be repeatedly run until the matching collision is passed. The human-computer training module will be run, or a risk reminder will be pushed when the data loading date of the detained internal transaction database is close to the reporting and regulatory deadline.

[0098] As can be seen from the above, in one embodiment, the cross-verification processing method for internal transactions may further include: when the third matching fails, temporarily storing the internal transaction in a retained internal transaction database;

[0099] A test is run once a day on the stranded internal transaction database. When it is detected that the data loading date of the stranded internal transaction database is approaching the reporting and supervision deadline, risk warning information is pushed to both trading entities of the stranded internal transaction.

[0100] During specific implementation, the above-mentioned specific implementation method for handling retained internal transactions can further improve the accuracy of internal transaction cross-verification.

[0101] From the above, it can be seen that in one embodiment, the cross-validation processing method of the above internal transactions may also include: when it is detected that the data loading date of the retained internal transaction database is not close to the reporting and supervision deadline date, performing a cloud map restoration and matching operation of the internal transaction on the next day for the retained internal transaction.

[0102] For specific implementation, please refer to the steps for detecting and storing stranded transactions. Figure 2 Implementation plan for the stranded internal transaction database in.

[0103] 7) Internal transactions that pass the matching collision will be run through the human-machine training module, which manually verifies the accuracy of the matching collision. If the human-machine training passes, it indicates that the matching collision in the data cloud map restoration module is accurate and the confidence level of the internal transaction matching is high. The internal transaction will then be marked and included in the marked internal transaction database, awaiting further processing by the subsequent consolidated data output and disclosure module. If the human-machine training fails, it indicates that the matching collision in the data cloud map restoration module is inaccurate. The test results will be fed back to the system, which will automatically initiate optimization. At the same time, the internal transactions that failed the human-machine training will be temporarily stored in the retained internal transaction database, pending the next day's re-run of the data cloud map restoration module.

[0104] As can be seen from the above, in one embodiment, the cross-verification processing method of the internal transaction may further include: if the matching is successful, sending the matched internal transaction to the terminal of the business personnel to verify the matching result.

[0105] In specific implementation, this step can be performed as follows Figure 2 The human-machine training module is implemented as shown. The human-machine training module manually checks whether the matching collision is accurate and feeds the test results back to the system to improve the rationality and accuracy of the system's automatic matching parameter settings.

[0106] As can be seen from the above, in one embodiment, the cross-verification processing method for internal transactions may further include:

[0107] The marked internal transactions included in the marked internal transaction database will be consolidated and pushed to the regulatory agency.

[0108] In implementation, this step can be realized by a "combined statement data output disclosure module" as shown in the accompanying drawings. Figure 2 The combined statement data output disclosure module, i.e., all matching marked internal transaction screening extraction single transactions form combined statement data, which is used for subsequent reporting of group combined statement management information to regulatory agencies and disclosure of group related information. The combined statement data formed after running the combined statement data output disclosure module can exclude data duplication reporting and duplication disclosure, improve the accuracy of combined statement data, and reduce the difficulty of manually matching transactions and preparing combined statement data.

[0109] In implementation, the specific implementation schemes of the cross-validation module, the data cloud map verification module and the man-machine training module are shown in the accompanying drawings. Figure 7 Figure 7 FIG. 1 is a flowchart of a preliminary cross-validation and cloud map verification processing method in an embodiment of the present application. The specific implementation scheme of the data cloud map restoration module is shown in the accompanying drawings. Figure 8 Figure 8 FIG. 2 is a flowchart of a cloud map restoration processing method for internal transactions in an embodiment of the present application.

[0110] 8) Parameterized dynamic adjustment:

[0111] According to the maximum transaction occurrence date difference of random two transactions in the internal transaction database, the extreme value of the vector 4 (transaction occurrence date difference vector) in the judgment condition matrix aggregation of the cross-validation module is dynamically adjusted. In an embodiment, the cross-validation processing method for internal transactions can further include: according to the maximum transaction occurrence date difference of random two transactions in the internal transaction database, the extreme value of the transaction occurrence date difference vector in the judgment condition matrix aggregation of the cross-validation module is dynamically adjusted, which can further improve the accuracy of internal transaction cross-validation.

[0112] According to the transaction occurrence date difference of the matched and marked transactions, a normal distribution graph is dynamically generated, and the priority selection of the vector 4 (transaction occurrence date difference vector) in the judgment condition matrix aggregation of the cross-validation module is sequentially fed back from large to small according to the date difference distribution probability. In an embodiment, the cross-validation processing method for internal transactions can further include: according to the transaction occurrence date difference of the internal transactions that have completed cross-validation, a normal distribution graph is dynamically generated, and the priority selection of the transaction occurrence date difference vector in the judgment condition matrix aggregation is sequentially fed back from large to small according to the date difference distribution probability in the normal distribution graph, which can further improve the accuracy of internal transaction cross-validation.

[0113] An internal transaction cross-validation processing device is also provided in an embodiment of the present application, as described in the following embodiments. Since the principle of solving the problem of the device is similar to that of the cross-validation processing method for internal transactions, the implementation of the device can be referred to the implementation of the cross-validation processing method for internal transactions, and the repeated parts will not be described again. ​​

[0114] Figure 9 FIG. 1 is a schematic diagram of the structure of a cross-verification processing device for internal transactions according to an embodiment of the present invention. Figure 9 As shown, the device includes:

[0115] Acquisition unit 01 is used to acquire internal transaction information that has not completed cross-validation from an internal transaction database; the internal transaction database is an internal transaction database pre-established based on internal transaction information acquired from multiple source system databases;

[0116] Matrix aggregation construction unit 02 is used to extract multiple types of transaction feature vectors from the internal transaction information that has not completed cross-validation to form a judgment condition matrix aggregation;

[0117] A preliminary cross-validation unit 03 is configured to obtain an internal transaction that has passed the preliminary cross-validation when it is determined that the cross-validation between the internal transactions recorded by at least two transaction parties has passed according to the judgment condition matrix aggregation;

[0118] The backtracking image unit 04 is used to retrieve the data source image of the internal transaction that has passed the preliminary cross-verification from the source system database according to the identifier of the internal transaction that has passed the preliminary cross-verification;

[0119] The cloud image metrology processing unit 05 is configured to extract key elements from the data source image to form a cloud image metrology transaction element item set, and match the cloud image metrology transaction element item sets corresponding to the internal transactions recorded by at least two transaction parties. If a match is successful, the internal transaction that has passed the preliminary cross-validation of the cloud image verification is obtained;

[0120] The optimized cross-validation unit 06 is used to mark the internal transactions that have passed the preliminary cross-validation as internal transactions that have completed the cross-validation and then include them into the marked internal transaction database.

[0121] In one embodiment, the matrix aggregation construction unit is specifically used to:

[0122] Extracting various types of transaction feature vectors from internal transaction information that has not completed cross-validation;

[0123] Construct a matrix by combining various types of transaction feature vectors in different ways;

[0124] The matrices formed by different combinations are processed by disambiguation to form the first-layer matrix group;

[0125] With the first layer matrix group as the initial layer matrix group, the following steps of forming the judgment condition matrix aggregation are cyclically executed. The following operations are performed in each cycle:

[0126] Get the current periodic layer matrix group;

[0127] In the current periodic layer matrix group, each two matrices are used to form a new matrix according to matrix operations, and multiple new matrices are processed through disambiguation to form the next periodic layer matrix group;

[0128] When the next periodic layer matrix group does not meet the preset termination condition, the next periodic layer matrix group is used as the current periodic layer matrix group to perform the next periodic operation until the next periodic layer matrix group meets the preset termination condition, thereby obtaining the judgment condition matrix aggregation.

[0129] In one embodiment, the cross-validation processing apparatus for internal transactions may further include a cloud graph restoration unit configured to perform the following cloud graph restoration and matching operations on internal transactions that fail the preliminary cross-validation and verify internal transactions that fail the preliminary cross-validation:

[0130] When it is determined that the cross-validation between the internal transactions recorded by at least two transaction parties fails according to the judgment condition matrix aggregation, the internal transaction that fails the preliminary cross-validation is obtained;

[0131] When the cloud image measurement transaction element item set corresponding to the internal transactions recorded by at least two transaction parties fails to match, the internal transaction that fails the preliminary cross-validation is obtained;

[0132] For internal transactions belonging to the transaction initiator that have failed preliminary cross-validation and internal transactions that have failed preliminary cross-validation, retrieve the data source image of each transaction initiator from the source system database.

[0133] For internal transactions belonging to counterparties that have failed preliminary cross-validation and internal transactions that have failed preliminary cross-validation, retrieve the data source image of each counterparty from the source system database.

[0134] After extracting key elements from the data source image of each transaction initiator, a knowledge graph of cloud image measurement transaction elements of each transaction initiator is constructed;

[0135] After extracting key elements from the data source image of each transaction counterparty, a knowledge graph of cloud image measurement transaction elements for each transaction counterparty is constructed;

[0136] Restore each edge feature of the cloud image measurement transaction element knowledge graph of each transaction initiator to the regressed internal transaction of each transaction initiator;

[0137] Restore each edge feature of the cloud image measurement transaction element knowledge graph of each transaction counterparty to the regressed internal transaction of each transaction counterparty;

[0138] Based on the internal transactions that failed the initial cross-validation and verification of the transaction initiator, the internal transactions that failed the initial cross-validation and verification of the transaction counterparty, the regression internal transactions of each transaction initiator, and the regression internal transactions of each transaction counterparty, matching processing is performed to obtain the internal transactions that finally passed the cross-validation.

[0139] In one embodiment, matching is performed based on the internal transactions belonging to the transaction initiator that failed the initial cross-validation and the internal transactions that failed the verification initial cross-validation, the internal transactions belonging to the transaction counterparty that failed the initial cross-validation and the internal transactions that failed the verification initial cross-validation, the regression internal transactions of each transaction initiator, and the regression internal transactions of each transaction counterparty to obtain the internal transactions that finally passed the cross-validation, which may include:

[0140] Based on the internal transactions of the transaction initiator that failed the initial cross-validation and the internal transactions that failed the initial cross-validation, a first match is performed with the regressed internal transactions of each transaction counterparty. When the first match is successful, the internal transactions that finally passed the cross-validation are obtained.

[0141] In one embodiment, matching is performed based on the internal transactions belonging to the transaction initiator that failed the initial cross-validation and the internal transactions that failed the verification initial cross-validation, the internal transactions belonging to the transaction counterparty that failed the initial cross-validation and the internal transactions that failed the verification initial cross-validation, the regression internal transactions of each transaction initiator, and the regression internal transactions of each transaction counterparty to obtain the internal transactions that finally passed the cross-validation, which may also include:

[0142] When the first matching fails, the internal transactions that failed the initial cross-validation and the internal transactions that failed the initial cross-validation belonging to the counterparty are used as the basis for a second matching with the regressed internal transactions of each transaction initiator. When the second matching passes, the internal transactions that finally passed the cross-validation are obtained.

[0143] In one embodiment, matching is performed based on the internal transactions belonging to the transaction initiator that failed the initial cross-validation and the internal transactions that failed the verification initial cross-validation, the internal transactions belonging to the transaction counterparty that failed the initial cross-validation and the internal transactions that failed the verification initial cross-validation, the regression internal transactions of each transaction initiator, and the regression internal transactions of each transaction counterparty to obtain the internal transactions that finally passed the cross-validation, which may also include:

[0144] If the second matching fails, the third matching is performed based on the regressed internal transaction of each transaction initiator and the regressed internal transaction of each transaction counterparty. If the third matching passes, the final cross-validated internal transaction is obtained.

[0145] In one embodiment, the cross-verification processing apparatus for internal transactions may further include a retention processing unit configured to:

[0146] If the third matching fails, the internal transaction will be temporarily stored in the retained internal transaction database;

[0147] A test is run once a day on the stranded internal transaction database. When it is detected that the data loading date of the stranded internal transaction database is approaching the reporting and supervision deadline, risk warning information is pushed to both trading entities of the stranded internal transaction.

[0148] In one embodiment, the above-mentioned retention processing unit can also be used to: when it is detected that the data loading date of the retained internal transaction database is not close to the reporting and supervision deadline date, perform the cloud graph restoration and matching operation of the internal transaction on the next day for the retained internal transaction.

[0149] In one embodiment, the cross-validation processing method for internal transactions may further include a dynamic adjustment unit configured to:

[0150] According to the maximum transaction date difference between two random transactions retained in the internal transaction database, the extreme value of the transaction date difference vector in the judgment condition matrix aggregation is dynamically adjusted.

[0151] In one embodiment, the dynamic adjustment unit is further configured to:

[0152] Based on the transaction date differences of the internal transactions that have completed cross-validation, a normal distribution graph is dynamically generated. Then, according to the distribution probability of the date differences in the normal distribution graph, they are sequentially fed back from large to small as the priority selection of the transaction date difference vector in the judgment condition matrix aggregation.

[0153] In one embodiment, the cross-validation processing method for internal transactions may further include a push unit configured to push the marked internal transactions included in the marked internal transaction database into consolidated data to a regulatory agency.

[0154] In one embodiment, the cross-verification processing method for internal transactions may further include a sending unit configured to send the matched internal transactions to a terminal of a business person if the matching is successful, so as to verify the matching result.

[0155] An embodiment of the present invention further provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above method when executing the computer program.

[0156] An embodiment of the present invention further provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned cross-verification processing method for internal transactions when executing the computer program.

[0157] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the computer program implements the cross-verification processing method for internal transactions.

[0158] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the cross-verification processing method for internal transactions.

[0159] Compared with the technical solutions in the prior art that combine automatic data acquisition and manual identification to perform internal transaction cross-validation, which have low efficiency and accuracy, the cross-validation processing solution for internal transactions provided by the embodiments of the present invention has the following beneficial technical effects: an internal transaction database is pre-established from internal transaction information obtained from multiple source system databases, and internal transaction information that has not completed cross-validation is obtained from the internal transaction database; multiple types of transaction feature vectors are extracted from the internal transaction information that has not completed cross-validation to form a judgment condition matrix aggregation; based on the judgment condition matrix aggregation and the extraction of key elements from the data source image, a cloud image metering transaction element item set is formed to perform cross-validation of internal transactions, which can achieve efficient and accurate cross-validation of internal transactions and improve the accuracy and efficiency of cross-validation processing of internal transactions.

[0160] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0161] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0162] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0163] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0164] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A cross-validation processing method for internal transactions, characterized in that: include: obtaining internal transaction information that has not completed cross-validation from an internal transaction database; the internal transaction database is an internal transaction database pre-established based on internal transaction information obtained from multiple source system databases; Extracting various types of transaction feature vectors from the internal transaction information that has not completed cross-validation to form a judgment condition matrix aggregation; When it is determined that the cross-validation between the internal transactions recorded by at least two transaction parties passes according to the judgment condition matrix aggregation, the internal transactions that have passed the preliminary cross-validation are obtained; According to the identifier of the internal transaction that has passed the preliminary cross-verification, the data source image of the internal transaction that has passed the preliminary cross-verification is retrieved from the source system database; After extracting key elements from the data source image, a cloud image measurement transaction element item set is formed. The cloud image measurement transaction element item sets corresponding to the internal transactions recorded by at least two transaction parties are matched. If the match is successful, the internal transaction that has passed the preliminary cross-validation of the cloud image verification is obtained; Internal transactions that have passed the preliminary cross-validation will be marked as internal transactions that have completed cross-validation and included in the marked internal transaction database; The cross-validation processing method for internal transactions also includes the following cloud image restoration and matching operations for internal transactions that fail preliminary cross-validation and internal transactions that fail verification of preliminary cross-validation: when it is determined according to the judgment condition matrix aggregation that the cross-validation between the internal transactions recorded by at least two transaction parties fails, the internal transactions that fail preliminary cross-validation are obtained; when the cloud image measurement transaction element item set corresponding to the internal transactions recorded by at least two transaction parties fails to match, the internal transactions that fail verification of preliminary cross-validation are obtained; for internal transactions that fail preliminary cross-validation and internal transactions that fail verification of preliminary cross-validation belonging to the transaction initiator, the data source image of each transaction initiator of the internal transactions that fail preliminary cross-validation and internal transactions that fail verification of preliminary cross-validation belonging to the transaction initiator are retrieved from the source system database; for internal transactions that fail preliminary cross-validation and internal transactions that fail verification of preliminary cross-validation belonging to the transaction counterparty, the data source image of each transaction initiator of the internal transactions that fail preliminary cross-validation and internal transactions that fail verification of preliminary cross-validation belonging to the transaction counterparty are retrieved from the source system database. The data source image of each transaction counterparty of the internal transactions that failed the cross-validation; after extracting the key elements from the data source image of each transaction initiator, the cloud image measurement transaction element knowledge graph of each transaction initiator is constructed; after extracting the key elements from the data source image of each transaction counterparty, the cloud image measurement transaction element knowledge graph of each transaction counterparty is constructed; each edge feature of the cloud image measurement transaction element knowledge graph of each transaction initiator is restored to the regressed internal transaction of each transaction initiator; each edge feature of the cloud image measurement transaction element knowledge graph of each transaction counterparty is restored to the regressed internal transaction of each transaction counterparty; according to the internal transactions that failed the preliminary cross-validation of the transaction initiator and the internal transactions that failed the preliminary cross-validation of the verification, the internal transactions that failed the preliminary cross-validation of the transaction counterparty and the internal transactions that failed the preliminary cross-validation of the verification, the regressed internal transactions of each transaction initiator, and the regressed internal transactions of each transaction counterparty, matching processing is performed to obtain the internal transactions that finally passed the cross-validation.

2. The method according to claim 1, wherein Extract various types of transaction feature vectors from the internal transaction information of uncompleted cross-validation to form a judgment condition matrix aggregation, including: Extracting various types of transaction feature vectors from internal transaction information that has not completed cross-validation; Construct a matrix by combining various types of transaction feature vectors in different ways; The matrices formed by different combinations are processed by disambiguation to form the first-layer matrix group; With the first layer matrix group as the initial layer matrix group, the following steps of forming the judgment condition matrix aggregation are cyclically executed. The following operations are performed in each cycle: Get the current periodic layer matrix group; In the current periodic layer matrix group, each two matrices are used to form a new matrix according to matrix operations, and multiple new matrices are processed through disambiguation to form the next periodic layer matrix group; When the next periodic layer matrix group does not meet the preset termination condition, the next periodic layer matrix group is used as the current periodic layer matrix group to perform the next periodic operation until the next periodic layer matrix group meets the preset termination condition, thereby obtaining the judgment condition matrix aggregation.

3. The method according to claim 1, wherein Based on the internal transactions that failed the initial cross-validation and verification of the transaction initiator, the internal transactions that failed the initial cross-validation and verification of the transaction counterparty, the regression internal transactions of each transaction initiator, and the regression internal transactions of each transaction counterparty, a matching process is performed to obtain the internal transactions that finally passed the cross-validation, including: Based on the internal transactions of the transaction initiator that failed the initial cross-validation and the internal transactions that failed the initial cross-validation, a first match is performed with the regressed internal transactions of each transaction counterparty. When the first match is successful, the internal transactions that finally passed the cross-validation are obtained.

4. The method according to claim 1, wherein Also includes: When the first matching fails, the internal transactions that failed the initial cross-validation and the internal transactions that failed the initial cross-validation belonging to the counterparty are used as the basis for a second matching with the regressed internal transactions of each transaction initiator. When the second matching passes, the internal transactions that finally passed the cross-validation are obtained.

5. The method according to claim 4, wherein Also includes: If the second matching fails, the third matching is performed based on the regressed internal transaction of each transaction initiator and the regressed internal transaction of each transaction counterparty. If the third matching passes, the final cross-validated internal transaction is obtained.

6. The method according to claim 5, wherein Also includes: If the third matching fails, the internal transaction will be temporarily stored in the retained internal transaction database; A test is run once a day on the stranded internal transaction database. When it is detected that the data loading date of the stranded internal transaction database is approaching the reporting and supervision deadline, risk warning information is pushed to both trading entities of the stranded internal transaction.

7. The method according to claim 6, wherein Also includes: When it is detected that the data loading date of the retained internal transaction database is not close to the reporting and supervision deadline date, the cloud graph restoration and matching operation of the retained internal transaction is performed on the next day.

8. The method according to claim 6, wherein Also includes: According to the maximum transaction date difference between two random transactions retained in the internal transaction database, the extreme value of the transaction date difference vector in the judgment condition matrix aggregation is dynamically adjusted.

9. The method according to claim 1, wherein Also includes: Based on the transaction date differences of the internal transactions that have completed cross-validation, a normal distribution graph is dynamically generated. Then, according to the distribution probability of the date differences in the normal distribution graph, they are sequentially fed back from large to small as the priority selection of the transaction date difference vector in the judgment condition matrix aggregation.

10. The method according to claim 1, wherein Also includes: The marked internal transactions included in the marked internal transaction database will be consolidated and pushed to the regulatory agency.

11. The method according to any one of claims 1 to 10, wherein: Also includes: If the matching is successful, the internal transaction that has passed the matching will be sent to the business personnel's terminal to verify the matching result.

12. A cross-verification processing device for internal transactions, characterized in that: include: an acquiring unit, configured to acquire internal transaction information that has not completed cross-validation from an internal transaction database; the internal transaction database is a pre-established internal transaction database based on internal transaction information acquired from multiple source system databases; A matrix aggregation construction unit is used to extract various types of transaction feature vectors from internal transaction information that has not completed cross-validation to form a judgment condition matrix aggregation; A preliminary cross-validation unit, configured to obtain an internal transaction that has passed the preliminary cross-validation when it is determined that cross-validation between internal transactions recorded by at least two transaction parties has passed according to the judgment condition matrix aggregation; The backtracking image unit is used to retrieve the data source image of the internal transaction that has passed the preliminary cross-verification from the source system database according to the identifier of the internal transaction that has passed the preliminary cross-verification; A cloud image metrology processing unit is configured to extract key elements from the data source image to form a cloud image metrology transaction element item set, match the cloud image metrology transaction element item sets corresponding to internal transactions recorded by at least two transaction parties, and obtain an internal transaction that has passed preliminary cross-validation of the cloud image verification if a match is successful; Optimize the cross-validation unit to mark internal transactions that have passed preliminary cross-validation as completed cross-validation internal transactions and include them in the marked internal transaction database; The cross-validation processing device for internal transactions further includes a cloud image restoration unit, which is used to perform the following cloud image restoration and matching operations for internal transactions that have failed preliminary cross-validation and internal transactions that have failed verification of preliminary cross-validation: when it is determined according to the judgment condition matrix aggregation that cross-validation between internal transactions recorded by at least two transaction parties has failed, the internal transactions that have failed preliminary cross-validation are obtained; when the cloud image measurement transaction element item set corresponding to the internal transactions recorded by at least two transaction parties fails to match, the internal transactions that have failed verification of preliminary cross-validation are obtained; for internal transactions that have failed preliminary cross-validation and internal transactions that have failed verification of preliminary cross-validation belonging to the transaction initiator, the data source image of each transaction initiator of the internal transactions that have failed preliminary cross-validation and internal transactions that have failed verification of preliminary cross-validation belonging to the transaction initiator is retrieved from the source system database; for internal transactions that have failed preliminary cross-validation and internal transactions that have failed verification of preliminary cross-validation belonging to the transaction counterparty, the internal transactions that have failed preliminary cross-validation belonging to the transaction counterparty are retrieved from the source system database. It is easy to verify the data source image of each transaction counterparty of the internal transactions that failed the preliminary cross-validation; after extracting the key elements from the data source image of each transaction initiator, a knowledge graph of the cloud image measurement transaction elements of each transaction initiator is constructed; after extracting the key elements from the data source image of each transaction counterparty, a knowledge graph of the cloud image measurement transaction elements of each transaction counterparty is constructed; each edge feature of the cloud image measurement transaction element knowledge graph of each transaction initiator is restored to the regressed internal transaction of each transaction initiator; each edge feature of the cloud image measurement transaction element knowledge graph of each transaction counterparty is restored to the regressed internal transaction of each transaction counterparty; according to the internal transactions that failed the preliminary cross-validation of the transaction initiator and the internal transactions that failed the preliminary cross-validation of the verification, the internal transactions that failed the preliminary cross-validation of the transaction counterparty and the internal transactions that failed the preliminary cross-validation of the verification, the regressed internal transactions of each transaction initiator, and the regressed internal transactions of each transaction counterparty, matching processing is performed to obtain the internal transactions that finally passed the cross-validation.

13. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 11 is implemented.

14. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 11 is implemented.

15. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 11 is implemented.

Images