Reverse checking method and system for SAAS platform

Abstract

The invention discloses a reverse checking method and system for an SAAS platform. The invention relates to the field of computers. Loading a preset matching rule and data check range configuration, and initializing a rule engine; according to a predefined checking range, extracting original data and calculation result data which need to be checked from a database in batches; calling a rule engine, and matching and checking the original data and the calculation result data; the rule engine judges whether the data are consistent or not according to a preset matching rule, and records a preliminary difference data set D; through an automatic and intelligent data checking mechanism, the data checking period can be remarkably shortened, and the time and energy required by manual spot check are reduced. An enterprise can find and solve data problems more quickly, and the overall operation efficiency is improved. According to the method, the matching rule engine and the collaborative filtering algorithm are adopted, data differences can be accurately recognized, and omissions and errors in manual spot check are effectively avoided.

Description

Technical Field

[0001] The present invention relates to the field of computers, in particular to the field of verification of fund calculation accuracy, and in particular to a reverse verification method and system for a SAAS platform. Background Art

[0002] In traditional financial or fiscal SAAS processing platforms, the conventional method of ensuring the accuracy of amount calculations relies on the platform's built-in calculation logic, supplemented by manual spot checks and result data verification.

[0003] CN201811462669.7 discloses a two-way transaction clearing method and system based on bill data (publication date: 2019-03-29); CN202010793537.3 discloses a clearing and settlement test method, device, terminal and storage medium (publication date: 2020-12-22);

[0004] This verification method has several inherent limitations and efficiency issues. For example, the current means based on manual spot checks or monitoring of data trends can only cover a portion of the data set, and cannot achieve comprehensive verification of all transactions and calculation results. This limitation leads to potential calculation errors or outliers not being discovered in time, thereby increasing financial risks and uncertainty in data processing.

[0005] Further analysis shows that manual spot checks are not only time-consuming and laborious, requiring a large amount of human resources, but are also susceptible to human factors during execution, such as fatigue, negligence, or misjudgment, which weaken the effectiveness of verification. Especially after the system logic changes, manual spot checks and calculations need to be re-performed each time to ensure the accuracy of calculations under the new logic. This process is not only cumbersome and inefficient, but also very easy to miss certain key verification scenarios. These missed scenarios involve complex transactions, edge cases, or data processing under specific conditions.

[0006] To this end, the present invention proposes a reverse verification method and system for a SAAS platform. Summary of the invention

[0007] In view of this, the present invention hopes to provide a reverse verification method and system for SAAS platform to solve or alleviate the technical problems existing in the prior art, that is, how to achieve no manual spot check and calculation every time the system logic changes, thereby ensuring the accuracy of calculation; the technical solution of the present invention is implemented as follows:

[0008] First, the reverse verification method for SAAS platform:

[0009] 1. Overview:

[0010] The present invention aims to build an efficient and automated data verification and monitoring mechanism to replace the traditional manual spot check and accounting method. By loading the preset matching rules and data verification range configuration, the system can intelligently identify and extract the original data and calculation result data to be verified. Using rule engine technology, the system can accurately match and verify these data, quickly identify data differences, and generate detailed difference data sets. Furthermore, algorithms such as collaborative filtering are used to conduct in-depth analysis of the difference data to more accurately troubleshoot and solve data problems. In the end, not only will the difference data set and its related information be stored in the database for subsequent query and analysis, but it will also trigger the monitoring and early warning mechanism to send the difference data to the business party for processing in a timely manner.

[0011] (II) Technical solution:

[0012] To achieve the above technical goals, when receiving the activation instruction input by the user (such as triggering the program startup through API call, system interface button click, etc.), perform the following steps.

[0013] 2.1 Step S1, load configuration:

[0014] Load the preset matching rules and data verification range configuration, and initialize the rule engine.

[0015] 2.1.1 Step S100, initializing the matching rules and data verification range configuration, including:

[0016] S1000, Classification and verification type: According to data type and business requirements, the verification type is divided into three categories: upstream and downstream voucher data verification, and voucher data and calculation result data verification.

[0017] S1001, upstream and downstream voucher data verification: transaction voucher data between suppliers or customers, including purchase orders, sales orders and logistics orders.

[0018] S1001, Verification of voucher data and calculation result data: Verification of the internal financial voucher data of the enterprise with its corresponding calculation results, including verification of accounting entries with general ledger balances and detailed ledger balances.

[0019] S1002, define the scope of verification:

[0020] The verification scope for each verification type includes which data tables and fields need to be involved in the verification, as well as the time range for the verification (such as monthly verification or quarterly verification, etc.).

[0021] 2.1.2 Step S101, initializing the rule engine, including:

[0022] S1010, upstream and downstream voucher data verification: If the upstream supplier invoice amount is not equal to the downstream customer payment amount, the difference is recorded and marked as the upstream and downstream voucher amount does not match;

[0023] S1011, check the voucher data with the calculation result data: if the final amount of the voucher is not equal to the sum of the account transaction amounts, record the difference and mark it as the voucher and account transaction amounts do not match;

[0024] Note: The "final amount of the voucher" here involves adjustments such as reverse refunds and changes during the intermediate process;

[0025] S1012, Inventory and Sales Reconciliation: If the inventory reduction amount is not equal to the sales outbound amount, record the difference and mark it as "Inventory does not match sales outbound amount;

[0026] S1013, configure the conditional statements of S1010 to S1012 into the rule engine, and set the priority of the rule and the parameters of the trigger condition.

[0027] 2.2 Step S2, data collection:

[0028] According to the predefined verification scope, the original data and calculation result data that need to be verified are extracted in batches from the database.

[0029] 2.3 Step S3, initial matching and verification:

[0030] The rule engine is called to match and check the original data and the calculated result data; the rule engine determines whether the data is consistent based on the preset matching rules and records the preliminary difference data set D.

[0031] 2.3.1 Step S300, rule engine preparation:

[0032] Start and load the configured rule engine to ensure that it is in operation and can accept input data for processing; check whether the scope, matching conditions and priority are consistent with the configuration loaded in step S1; if so, proceed to S301; if not, issue an error warning.

[0033] 2.3.2 Step S301, data input and preprocessing:

[0034] The raw data and calculation result data collected in S2 are converted and adapted, deduplication, sorting, and format standardization are performed, and then input into the rule engine.

[0035] 2.3.3 Step S302, matching and checking execution:

[0036] The rule engine compares the original data and the calculated result data item by item according to the preset matching rules, including equal amounts, timestamp matching and business logic consistency.

[0037] For example, if the final amount of the original data (after reverse refunds and changes) is not equal to the sum of the account flow, it is considered a discrepancy. The rule engine will automatically calculate and compare these data items to determine whether they meet the rule conditions.

[0038] During the matching process, the rule engine will automatically identify data records that do not meet the matching rules and mark them as preliminary difference data.

[0039] 2.3.4 Step S303, preliminary difference data set recording

[0040] All identified preliminary difference data records are organized into a preliminary difference data set D, which includes the data ID, difference type and difference description of the preliminary difference data.

[0041] 2.4 Step S4, preliminary difference data set processing:

[0042] For the preliminary difference data set D, a collaborative filtering algorithm is executed, and the above-mentioned "raw data and calculation result data dependent on the calculation" are regarded as the "data matching and identification with user-item structure" problem. The collaborative filtering algorithm finds other rows similar to the target row (user or item) in the preliminary difference data set D, sorts out the difference data and stores its information, timestamp and warning conditions in three dimensions to form the difference data set D'.

[0043] 2.4.1 Step S400, similarity (i.e., "similar other rows") calculation:

[0044] The collaborative filtering algorithm used in this solution belongs to the user-based collaborative filtering algorithm; considering that the preliminary difference data set D contains outliers, the traditional similarity calculation method (such as cosine similarity and Pearson correlation coefficient) is not enough to accurately reflect the differences between users. Therefore, this step adopts another similarity calculation method:

[0045]

[0046] Where: u and v represent user-item (original data-difference data) respectively; y(u,v) represents the similarity between user-item (original data-difference data); Iuv represents the set of items that users u and v have evaluated together; r ui Represents the cosine similarity between user u and item i; and Respectively represent the average cosine similarity of users u and v; σ u and σ v Represent the standard deviation of users u and v respectively;

[0047] wi represents the weight of item i, which is set according to the importance of the item. In this scheme, importance can be regarded as the discreteness of the data (entropy value); then, the relative importance of each indicator (item) is reflected based on the discreteness of the data itself (i.e., entropy value). The basic idea is that the greater the discreteness of the indicator, the greater the influence of the indicator on the comprehensive evaluation (i.e., weight); the weight w i The assignment method is as follows:

[0048] S4000, standardize the item scores (or difference indexes) in the difference dataset D to eliminate the influence of different index dimensions and magnitudes:

[0049] Among them, r ui is the original rating (or difference index) of user u for item i, r ui ′ is the standardized score, min(r i ) and max(r i ) are the minimum and maximum ratings of item i among all users, respectively.

[0050] S4001, calculate entropy value: calculate the entropy value e of each item i i :

[0052] Where m is the number of users, p ui It is the ratio of user u's rating (or difference index) for item i to the total rating of the item, that is:

[0053] Where k is a constant To ensure the entropy value e i The value range is between 0 and 1.

[0054] S4002, calculate the difference d of each item i , which is 1 minus the entropy value e i :d i =1-e i ;

[0055] Difference d i It reflects the degree of dispersion of item i in all user ratings, that is, the contribution of the item to user differences.

[0056] S4003, get weight

[0057] Where n is the number of items.

[0058] 2.4.2 Step S401, execute the difference analysis algorithm:

[0059] After finding similar users, a specific difference analysis algorithm is needed to further check and confirm the authenticity and importance of the difference data. This solution uses a difference analysis algorithm based on rules and statistical analysis:

[0060] S4010, difference data aggregation: calculate the frequency of occurrence and degree of difference of each similarity y(u,v);

[0061] S4011, difference rule definition: When a certain similarity y(u,v) exceeds the threshold T, it is regarded as an important difference item;

[0062] S4012, Difference Cause Analysis: For the screened difference items, analyze them in combination with business logic and context information to determine the causes of the differences, including data entry errors, calculation logic problems, and business process changes.

[0063] 2.4.3 Step S402, output:

[0064] The difference items are combined into a difference data set D', which contains the information of confirmed important and significant difference items, timestamps, warning conditions, and difference cause analysis.

[0065] 2.5 Step S5, result storage and feedback:

[0066] Generate a report or log file of the difference data set D', recording the information and execution status of this verification.

[0067] At this point, the solution is executed or enters the loop execution mode (depending on the system configuration). In this state, the system has completed the data verification and monitoring task and is ready to process the next task.

[0068] (III) Mechanism for solving technical problems:

[0069] The matching rule engine is used to intelligently match and check the collected raw data and the calculated result data. These rules are preset based on business logic and data processing requirements, and can accurately determine the consistency of data. Through automated matching and verification, the system can quickly identify data differences, avoiding the tediousness and omissions of manual spot checks.

[0070] For discrepancy data found during the matching process, algorithms such as collaborative filtering are used to identify and resolve data issues, as well as to record detailed information, timestamps, and warning conditions for the discrepancy data.

[0071] Second, the reverse verification system for SAAS platform:

[0072] like Figure 2As shown, the system includes a memory storing program instructions and a processor connected to the memory. When the processor executes the above program instructions, the reverse verification method for the SAAS platform described above can be implemented, and also includes:

[0073] (1) Load the configuration loading module for the preset matching rules and data verification range configuration:

[0074] Output: The initialized rule engine is ready to perform data matching tasks.

[0075] (2) A data collection module that extracts the raw data and calculation result data that need to be checked in batches from the database according to the predefined verification scope:

[0076] Input: Configuration load module provides the check scope configuration.

[0077] Output: extracted raw data and calculation result data for use by subsequent modules.

[0078] (3) Call the rule engine to match and check the original data and the calculated result data, determine whether the data is consistent, and record the initial matching and verification module of the initial difference data set D:

[0079] Input: original data and calculation result data provided by the data collection module, and matching rules provided by the configuration loading module.

[0080] Output: preliminary difference dataset D for subsequent modules to process.

[0081] (4) A collaborative filtering algorithm is executed on the preliminary difference data set D to identify the difference data, and its information, timestamp and warning conditions are stored in three dimensions to form a preliminary difference data set processing module for the difference data set D':

[0082] Input: The preliminary difference dataset D provided by the initial matching and verification module.

[0083] Output: difference dataset D', for storage and feedback of subsequent modules.

[0084] (5) Generate a report or log file of the difference data set D', record the information and execution status of this verification, and provide the business party with a result storage and feedback module for the verification results:

[0085] Input: The difference dataset D' provided by the preliminary difference dataset processing module.

[0086] Output: Report or log file as a record of system execution.

[0087] Compared with the prior art, the present invention has the following beneficial effects:

[0088] 1. Improve data verification efficiency: Through the automated and intelligent data verification mechanism, the present invention can significantly shorten the data verification cycle and reduce the time and effort required for manual spot checks. It helps enterprises to discover and solve data problems more quickly and improve overall operational efficiency.

[0089] 2. Ensure data accuracy: The present invention uses a matching rule engine and collaborative filtering algorithm to accurately identify data differences and effectively avoid omissions and errors in manual spot checks. It helps to ensure the accuracy and reliability of enterprise data and provide strong support for business decision-making.

[0090] 3. Reduce labor costs: Through automated data verification, the present invention can significantly reduce the reliance on manual spot checks, thereby reducing the labor costs of enterprises. It helps enterprises invest more resources in core businesses and improve overall competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0092] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0093] Figure 2 It is a schematic diagram of the system composition of the present invention;

[0094] Figure 3 It is a schematic diagram of the composition of the rule engine of the present invention;

[0095] Figure 4 is a schematic flow chart of step S4 of the present invention;

[0096] Figure 5 A schematic diagram of the business logic on which the "credentials" of step S1 of the present invention depend;

[0097] Figure 6 It is a schematic diagram of “checking” in step S3 of the present invention. DETAILED DESCRIPTION

[0098] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below;

[0099] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the various embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0100] Explanation of relevant terms:

[0101] (1) Matching rules: refers to the preset standards or logic used by the system to determine whether the original data and the calculated result data are consistent;

[0102] (2) Data verification scope: defines the type and scope of data that needs to be verified, that is, which raw data and calculation result data need to be extracted by the system and matched and verified;

[0103] (3) Original data: refers to the basic data that business logic calculations rely on and is an input item in the data verification process;

[0104] (4) Calculation result data: refers to the result obtained through business logic calculation based on the original data. It is another input item in the data verification process and is used to match with the original data;

[0105] (5) Collaborative filtering algorithm: It is a data analysis method used in this scheme to process preliminary difference data sets and help identify and resolve inconsistencies in the data;

[0106] (6) Target row: In collaborative filtering algorithms, it refers to the benchmark data row for which similar rows need to be found, usually used to identify discrepant data;

[0107] (7) Information, timestamp, and warning conditions: These are important components of the difference data set. Information refers to the specific content of the difference data, the timestamp records the time when the difference occurs, and the warning conditions are used to trigger the monitoring and warning mechanism.

[0108] Embodiment 1: Figure 1 , 3As shown in Figures 1 to 6, this embodiment discloses the business logic of the reverse verification method for the SAAS platform in the financial verification of the online car-hailing platform; in the online car-hailing SAAS platform, financial verification is a key link to ensure the accuracy of platform revenue, driver settlement and passenger payment. This solution aims to reversely verify the financial data of the online car-hailing platform through automated and intelligent means, timely discover and resolve differences, and ensure financial accuracy. The method includes the following steps S1 to S5.

[0109] In this embodiment, regarding step S1: loading configuration:

[0110] S100: Initialize matching rules and data verification range configuration.

[0111] S1000: Classification and verification type. In the online car-hailing platform, the verification types can be divided into: verification of passenger payment data and driver settlement data, verification of passenger payment data and platform income data, and verification of driver settlement data and platform expenditure data.

[0112] S1001: Define the upstream and downstream voucher data verification, including passenger payment vouchers, driver settlement vouchers and platform bank statements.

[0113] S1002: Check the passenger payment data, driver settlement data and platform bank transaction data for the past month.

[0114] S101: Initialize the rule engine:

[0115] S1010: Set a rule such that if the amount paid by the passenger is not equal to the amount settled by the driver, the difference is recorded.

[0116] S1011: Set rules such that if the total passenger payment is not equal to the total platform revenue, the difference is recorded.

[0117] S1012: Set rules. If the driver's settlement amount is not equal to the platform's total expenditure (taking into account platform service fees, etc.), record the difference.

[0118] S1013: Configure the above rules into the rule engine, and set the priority and triggering conditions of the rules.

[0119] In this embodiment, regarding step S2: data collection: extract the passenger payment data, driver settlement data and platform bank transaction data of the most recent month from the database.

[0120] In this embodiment, regarding step S3: initial matching and verification:

[0121] S300: Rule engine preparation. Start and load the configured rule engine to ensure it is in operation.

[0122] S301: Data input and preprocessing: The collected data is cleaned, formatted and standardized, and then input into the rule engine.

[0123] S302: Matching and verification execution. The rule engine compares the data item by item according to the preset rules, such as whether the passenger payment amount is consistent with the driver settlement amount, whether the passenger payment amount is consistent with the platform revenue amount, etc.

[0124] S303: Preliminary difference data set recording: All data records that do not conform to the rules are sorted into a preliminary difference data set D.

[0125] In this embodiment, regarding step S4: preliminary difference data set processing:

[0126] S400: Similarity calculation. A user-based collaborative filtering algorithm is used to calculate the similarity between the data items in the preliminary difference data set. Considering the particularity of financial data, a weighted cosine similarity calculation method can be used, where the weight is set according to the discrete degree (entropy value) of the data:

[0127] Step S400, similarity (i.e., "similar other rows") calculation:

[0128] The collaborative filtering algorithm used in this solution is a user-based collaborative filtering algorithm; considering that the preliminary difference data set D contains outliers, the traditional similarity calculation method (such as cosine similarity and Pearson correlation coefficient) is not enough to accurately reflect the differences between users. Therefore, this step uses another similarity calculation method:

[0129]

[0130] Where: u and v represent user-item (original data-difference data) respectively; y(u,v) represents the similarity between user-item (original data-difference data); Iuv represents the set of items that users u and v have evaluated together; r ui Represents the cosine similarity between user u and item i; and Respectively represent the average cosine similarity of users u and v; σ u and σ v Represent the standard deviation of users u and v respectively;

[0131] w i represents the weight of item i, which is set according to the importance of the item. In this scheme, importance can be regarded as the discreteness of the data (entropy value); then, the relative importance of each indicator (item) is reflected based on the discreteness of the data itself (i.e., entropy value). The basic idea is that the greater the discreteness of the indicator, the greater the influence of the indicator on the comprehensive evaluation (i.e., weight); the weight w iThe assignment method is as follows:

[0132] S4000, standardize the item scores (or difference indexes) in the difference dataset D to eliminate the influence of different index dimensions and magnitudes:

[0133] Among them, r ui is the original rating (or difference index) of user u for item i, r ui ′ is the standardized score, min(r i ) and max(r i ) are the minimum and maximum ratings of item i among all users, respectively.

[0134] S4001, calculate entropy value: calculate the entropy value e of each item i i :

[0136] Where m is the number of users, p ui It is the ratio of user u's rating (or difference index) for item i to the total rating of the item, that is:

[0137] Where k is a constant To ensure the entropy value e i The value range is between 0 and 1.

[0138] S4002, calculate the difference d of each item i , which is 1 minus the entropy value e i :d i =1-e i ;

[0139] Difference d i It reflects the degree of dispersion of item i in all user ratings, that is, the contribution of the item to user differences.

[0140] S4003, get weight

[0141] Where n is the number of items.

[0142] Step S401, executing the difference analysis algorithm:

[0143] After finding similar users, a specific difference analysis algorithm is needed to further check and confirm the authenticity and importance of the difference data. This solution uses a difference analysis algorithm based on rules and statistical analysis:

[0144] S4010, difference data aggregation: calculate the frequency of occurrence and degree of difference of each similarity y(u,v);

[0145] S4011, difference rule definition: When a certain similarity y(u,v) exceeds the threshold T, it is regarded as an important difference item;

[0146] S4012, Difference Cause Analysis: For the screened difference items, analyze them in combination with business logic and context information to determine the causes of the differences, including data entry errors, calculation logic problems, and business process changes.

[0147] Specifically, step S402, output: the difference items are formed into a difference data set D', which includes information of confirmed important and significant difference items, timestamps, warning conditions and difference cause analysis.

[0148] In this embodiment, regarding step S5: result storage and feedback:

[0149] Generate a report or log file of the difference data set D', recording the information and execution status of this verification.

[0150] Store the report or log file in a designated location for financial personnel to review and process.

[0151] Whether to enter the cyclic execution mode is determined based on the system configuration to continuously monitor changes in financial data.

[0152] In summary, in this embodiment, through the automated processing of the rule engine and collaborative filtering algorithm, the system can automatically adapt to changes in business logic and complete data verification without manual intervention. This greatly reduces the time and cost of manual spot checks and improves verification efficiency and accuracy. The collaborative filtering algorithm can identify and exclude abnormal data, further improving the accuracy of verification. The automated and intelligent financial data verification process can significantly improve the efficiency and accuracy of financial work and reduce the occurrence of human errors and fraud.

[0153] Example 2: This example further provides a python execution program for the reverse verification method for SAAS platform as described in Example 1 in the financial verification of online car-hailing platform:

[0154]

[0155]

[0156]

[0157]

[0158]

[0159] Note: The data passed in here should contain the necessary fields.

[0160] store_and_feedback_results(important_differences)

[0161] if __name__ == '__main__':

[0162] main()

[0163] In the above program, the load_config function performs the loading of configuration. The collect_data function extracts passenger payment data, driver settlement data, and platform revenue data from the database.

[0164] The initial_match_and_check function performs preliminary matching and verification of the data according to the loaded rules and records the differences. The process_initial_differences function processes the preliminary difference data set. The similarity calculation in the collaborative filtering algorithm is performed here, and the cosine similarity is used.

[0165] The store_and_feedback_results function generates a report or log file and records the information and execution status of this verification. The main function is the entry point of the program, which calls the functions of each module in sequence to execute the entire financial verification process.

[0166] All the above embodiments only express the implementation methods of the relevant practical applications of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be based on the attached claims.

[0167] For those skilled in the art, it can be further appreciated that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0168] At the same time, those skilled in the art can understand that all or part of the processes in all the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided in this application and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

Claims

1. A reverse verification method for a SAAS platform, characterized in that: After receiving the activation command input by the user, perform the following steps: S1, load the preset matching rules and data verification range configuration, and initialize the rule engine; S2, extracting the original data and calculation result data that need to be checked according to the predefined checking range; S3, calling the rule engine to match and check the original data and the calculated result data; the rule engine determines whether the data is consistent according to the preset matching rules, and records the preliminary difference data set D; S4, the collaborative filtering algorithm finds other rows similar to the target row in the preliminary difference data set D, sorts out the difference data and stores its information, timestamp and warning conditions in three dimensions to form the difference data set D'.

2. The reverse verification method according to claim 1, characterized in that: In S1, the matching rules and data verification range configuration are initialized, including: S1000, Classification and verification type: According to data type and business requirements, the verification type is divided into three categories: upstream and downstream voucher data verification, voucher data and calculation result data verification; S1001, upstream and downstream voucher data verification: transaction voucher data between suppliers or customers, including purchase orders, sales orders and logistics orders; S1001, Verification of voucher data and calculation result data: Verification of the internal financial voucher data of the enterprise with its corresponding calculation results, including verification of accounting entries with general ledger balances and detailed ledger balances.

3. The reverse verification method according to claim 2, characterized in that: In S1, the rules for initializing the rule engine include: S1010, upstream and downstream voucher data verification: If the upstream supplier invoice amount is not equal to the downstream customer payment amount, the difference is recorded and marked as the upstream and downstream voucher amount does not match; S1011, check the voucher data with the calculation result data: if the final amount of the voucher is not equal to the sum of the account transaction amounts, record the difference and mark it as the voucher and account transaction amounts do not match; S1012, Inventory and Sales Reconciliation: If the inventory reduction is not equal to the sales outbound quantity, the difference is recorded and marked as "Inventory does not match sales outbound quantity." 4. The reverse verification method according to claim 1, characterized in that: The execution method of S3 includes: S300, start and load the configured rule engine to ensure that it is in operation and can accept input data for processing; check whether the scope, matching conditions and priority are consistent with the configuration loaded in step S1; if so, enter S301; S301, converting and adapting the original data and calculation result data collected in S2, performing deduplication, sorting, and format standardization, and then inputting them into the rule engine; S302, the rule engine compares the original data and the calculated result data one by one according to the preset matching rules, including the equality of amount, timestamp matching and business logic consistency; S303, organizing all identified preliminary difference data records into a preliminary difference data set D, which includes the data ID, difference type and difference description of the preliminary difference data.

5. The reverse verification method according to claim 1, characterized in that: In S4, the collaborative filtering algorithm is a user-based collaborative filtering method, and the execution method is as follows: Where: u and v represent users and items, i.e., original data and difference data, respectively; y(u,v) represents the similarity between users and items; Iuv represents the set of items that users u and v have evaluated together; r ui Represents the cosine similarity between user u and item i; and Respectively represent the average cosine similarity of users u and v; σ u and σ v Represent the standard deviation of users u and v respectively; w i represents the weight of item i; S401, executing the difference analysis algorithm: S4010, difference data aggregation: calculate the frequency of occurrence and degree of difference of each similarity y(u,v); S4011, difference rule definition: When a certain similarity y(u,v) exceeds the threshold T, it is regarded as an important difference item; S4012, Difference Cause Analysis: For the screened difference items, analyze them in combination with business logic and context information to determine the causes of the differences, including data entry errors, calculation logic problems, and business process changes.

6. The reverse verification method according to claim 5, characterized in that: The weight w i The assignment strategy is: assign weight w i The importance represented is regarded as the degree of discreteness of the data. The relative importance of each indicator is reflected based on the degree of discreteness of the data itself. The greater the degree of discreteness of the indicator, the greater the impact of the indicator on the comprehensive evaluation.

7. The reverse verification method according to claim 6, characterized in that: The weight w i The assignment method is as follows: S4000, standardize the item scores in the difference dataset D: Among them, r ui is the original rating of user u for item i, r ui ′ is the standardized score, min(r i ) and max(r i ) are the minimum and maximum ratings of item i among all users respectively; S4001, calculate the entropy value e of each item i i : Where m is the number of users, p ui is the ratio of user u's rating of item i to the total rating of the item; S4002, calculate the difference d of each item i :d i =1-e i ; S4003, get weight Where n is the number of items.

8. The reverse verification method according to claim 1, characterized in that: Also includes: S5, generating a report or log file of the difference data set D', recording the information and execution status of this verification.

9. A system for implementing the reverse verification method according to any one of claims 1 to 8, characterized in that: The system comprises: A configuration loading module for loading preset matching rules and data verification range configuration; A data collection module that batch extracts the raw data and calculation result data that need to be checked from the database according to the predefined checking scope; Call the rule engine to match and check the original data and the calculated result data, determine whether the data is consistent, and record the initial matching and verification module of the initial difference data set D; A collaborative filtering algorithm is executed on the preliminary difference data set D to identify the difference data, and its information, timestamp and warning conditions are stored in three dimensions to form a preliminary difference data set processing module of the difference data set D'.

10. The system according to claim 9, characterized in that: It also includes a report or log file for generating a difference data set D', recording the information and execution status of this verification, and providing a result storage and feedback module for the business party to obtain the verification results.

Images