A global unique identification-based audit full-link data processing and verification method and system

Abstract

This invention belongs to the field of data processing technology, specifically relating to a method and system for end-to-end data processing and verification based on globally unique identifiers. It includes: after a client submits an application, a globally unique work order ID is generated and associated with the records to be reviewed. Upon completion of the review, the information is serialized and stored in a snapshot table with the work order ID as the primary key. Based on a sampling strategy, work orders that have not undergone quality inspection are grouped and sampled to generate quality inspection tasks. During task execution, key fields are read from the snapshot table for data verification, and a quality inspection report is output. This application helps solve problems such as inconsistent identifiers leading to broken traceability, changes in data status leading to inconsistent quality inspection benchmarks, fixed and singular sampling strategies, reliance on manual item-by-item comparison, and a lack of quality analysis dimensions. It significantly improves the traceability efficiency and accuracy, quality inspection consistency and automation level of the entire review process, and achieves flexible sampling that precisely matches business risks, as well as refined management and accurate attribution of review quality.

Description

Technical Field

[0001] This invention belongs to the field of data processing technology, specifically relating to a method and system for end-to-end data processing and verification based on a globally unique identifier. Background Technology

[0002] This invention relates to the field of review and quality inspection technology for ride-hailing platforms, specifically to a method and system for full-link data processing and verification based on a globally unique identifier.

[0003] In the daily operation of ride-hailing platforms, various applications such as driver recruitment, vehicle replacement, and information changes need to be reviewed. To control the quality of the review process, the platform sets up a dedicated quality inspection stage, where quality inspectors conduct random checks on completed review orders. The existing quality inspection methods typically employ the following workflow:

[0004] In the lifecycle of an audit application, an application number is generated during the application submission stage, an audit record ID is used during the audit stage, and a quality inspection task ID is generated during the quality inspection stage. Each system module uses its own independent local identifier. When the quality inspection process finds a problem with an audit work order, if it is necessary to trace back the original application content and the complete situation at the time of the audit, it often requires multiple join queries between multiple business databases. The tracing path is long, the operation is cumbersome, and it is easy to cause data errors due to mismatched join conditions.

[0005] Regarding the acquisition of quality inspection content, the current practice typically involves the quality inspection system querying driver, vehicle, and document information databases in real time after a quality inspection task is generated to obtain and display the latest data status. However, driver and vehicle information is constantly changing. After an audit work order is completed, the driver may have changed their phone number, re-uploaded their ID photo, and the vehicle may have undergone a transfer of ownership or a change in usage. The quality inspector sees the latest data during review, not the data they are facing and relying on at the time of the audit. This time lag in data status leads to a discrepancy between the basis of the quality inspection conclusion and the basis of the audit decision, resulting in misjudgments due to data changes.

[0006] In the sampling inspection stage, existing methods mostly involve random sampling at a fixed ratio or full sampling. This approach results in a relatively simplistic configuration of sampling strategies, making it impossible to differentiate sampling settings based on dimensions such as audit type, city, vehicle type, or specific auditor, or to set different sampling ratios and execution frequencies for different combinations of dimensions. Furthermore, the allocation of sampling tasks is typically done manually by managers or through simple equal distribution, failing to automatically and proportionally split tasks according to the weights set by quality inspectors. This leads to low allocation efficiency and uneven load distribution when the workload is high.

[0007] When quality inspectors actually perform the review, the system typically only lists and displays the application form content and relevant document information. For key inspection items such as the format and uniqueness of the vehicle VIN code, the consistency between the registration date and the vehicle registration certificate, compliance with usage period requirements, the format and uniqueness of the license plate number, whether the document photos are real-life photos without obstructions or glare, and whether they are valid, the existing system lacks the ability to perform real-time automatic verification during the quality inspection process. Quality inspectors need to rely on their own experience to manually compare each item, which is prone to oversights, and the efficiency and accuracy of the review are greatly affected by human factors.

[0008] Regarding statistical analysis after quality inspection, current practices typically involve calculating the overall audit error rate, making it difficult to perform drill-down and cross-analysis based on multiple dimensions such as auditor, audit type, city, vehicle model, and date. Due to the lack of a mechanism to accurately link audit records with quality inspection results, when quality issues are discovered, it is impossible to precisely attribute errors to specific auditors, audit scenarios, or error types, hindering managers from implementing targeted quality improvements.

[0009] Existing technologies have several drawbacks, including inconsistent labeling in the application review and quality inspection stages, making traceability difficult; the ability to obtain only the latest data during quality inspection and not reconstruct the historical state at the time of review; a single sampling strategy with uneven task allocation; a lack of real-time automatic verification during the quality inspection process; and a single dimension of quality analysis that makes it impossible to accurately attribute causes. Summary of the Invention

[0010] To address these issues, this application provides a method and system for end-to-end data processing and verification based on globally unique identifiers. This helps to solve the problems of existing technologies, such as fragmented data identification leading to broken traceability links, inconsistencies between quality inspection data benchmarks and audit decision benchmarks, rigid sampling strategies and unbalanced task allocation, lack of real-time verification mechanisms in the review process, and isolated quality analysis dimensions that make accurate attribution difficult.

[0011] To achieve the above objectives, this application adopts the following technical solution: Firstly, this application provides a method for end-to-end data processing and verification of auditing based on a globally unique identifier, the method comprising: Obtain application request information submitted by the client; the application request information includes application type and application form information, the application type includes driver recruitment, vehicle replacement, and information change, and the application form information includes at least one of driver information, vehicle information, and document information; In response to the application request information, a globally unique work order ID is generated, the work order ID is written into the application record table, and an approval record is generated and stored in the approval record table. The audit record has an audit record ID and is associated with the work order ID, and the audit status field in the audit record table is initially set to pending audit status. The review status field in the review record table is detected. When the review status field is updated from pending review to review completion, the driver information database, vehicle information database, document information database, and application form database are queried according to the review record ID to obtain the driver information, vehicle information, document information, and application form content at the current time. The obtained driver information, vehicle information, document information, and application form content are serialized and written into the review snapshot table for storage using the work order ID as the unique primary key. Based on the results of writing the work order ID as a unique primary key into the audit snapshot table, the preset sampling strategy configuration is read. The sampling strategy configuration includes grouping dimensions, sampling ratio, and execution frequency. The grouping dimensions include audit type, city, vehicle type, and auditor. When the triggering condition corresponding to the execution frequency is met, the work orders in the audit record table whose audit status field is an audit completion identifier and whose corresponding work order ID does not exist in the quality inspection task table are queried. The queried work orders are grouped according to the grouping dimensions to obtain multiple independent sampling buckets. For each sampling bucket, the number of samples is calculated according to the sampling ratio corresponding to the sampling bucket and the sampling is performed. A quality inspection task is generated and written into the quality inspection task table. Each quality inspection task record is associated with the work order ID of the sampled work order. When the quality inspection task is executed, the corresponding snapshot data is read from the audit snapshot table using the work order ID, the preset key fields in the snapshot data are verified, the verification result is generated, and a quality inspection report is generated based on the verification result.

[0012] Secondly, this application provides a data processing and verification system for the entire audit chain based on globally unique identifiers, applied to any one of the above-mentioned data processing and verification methods for the entire audit chain based on globally unique identifiers, the system comprising: The acquisition module is used to acquire application request information submitted by the client; the application request information includes application type and application form information, the application type includes driver recruitment, vehicle replacement, and information change, and the application form information includes at least one of driver information, vehicle information, and document information; The first processing module is used to respond to the application request information, generate a globally unique work order ID, write the work order ID into the application record table, and at the same time generate an audit record and store it in the audit record table. The audit record has an audit record ID and is associated with the work order ID, and the audit status field in the audit record table is initially set to pending audit status. The second processing module is used to detect the audit status field in the audit record table. When the audit status field is updated from pending audit status to audit completion identifier, the module queries the driver information database, vehicle information database, document information database and application form database according to the audit record ID to obtain the driver information, vehicle information, document information and application form content at the current time. After serializing the obtained driver information, vehicle information, document information and application form content, the module writes them into the audit snapshot table for storage with the work order ID as the unique primary key. The third processing module is used to read the results stored in the audit snapshot table based on the work order ID as the unique primary key, read the preset sampling strategy configuration, which includes grouping dimensions, sampling ratio, and execution frequency. The grouping dimensions include audit type, city, vehicle type, and auditor. When the triggering condition corresponding to the execution frequency is met, the module queries the audit record table for work orders whose audit status field is an audit completion identifier and whose corresponding work order ID does not exist in the quality inspection task table. The module groups the queried work orders according to the grouping dimensions to obtain multiple independent sampling buckets. For each sampling bucket, the module calculates the sampling quantity according to the sampling ratio corresponding to the sampling bucket and performs the sampling, generating a quality inspection task and writing it to the quality inspection task table. Each quality inspection task record is associated with the work order ID of the sampled work order. The fourth processing module is used to read the corresponding snapshot data from the audit snapshot table through the work order ID when the quality inspection task is executed, verify the preset key fields in the snapshot data, generate verification results, and generate quality inspection conclusions based on the verification results.

[0013] The application employs the above technical solution and has at least the following beneficial effects: 1. By using a globally unique work order ID to connect the entire process of application, review, snapshot, sampling, and quality inspection, the problem of broken traceability links caused by inconsistent identification at each stage in existing technologies is solved. The work order ID is generated and forcibly written into the application record table and review record table when the application is submitted. Subsequent snapshot storage, sampling task generation, and quality inspection execution all use this work order ID as the association key, which allows direct traceability from the quality inspection conclusion to the original application content and review process without the need for multiple cross-database join queries, significantly improving traceability efficiency and accuracy.

[0014] 2. By monitoring the audit status field and instantly capturing all business data to generate a snapshot upon audit completion, the problem of inconsistencies between quality inspection benchmarks and audit decision benchmarks caused by continuous data changes after auditing in existing technologies is resolved. The snapshot is generated instantly upon audit completion and stored independently in the audit snapshot table. Subsequent quality inspection reads the historical status at the time of auditing, rather than the latest data, fundamentally eliminating quality inspection misjudgments caused by time differences in data status, and ensuring that quality inspection conclusions are based on the exact same data foundation used by the auditor at the time of auditing.

[0015] 3. By enabling multi-dimensional binning sampling through configurable grouping dimensions, sampling ratios, and execution frequencies, the system solves the problem of fixed and singular sampling strategies in existing technologies. Sampling strategies can be freely combined and configured according to dimensions such as audit type, city, vehicle type, and auditor. Different combinations of dimensions can set different sampling ratios and execution frequencies. The system automatically bins according to the strategy configuration and independently calculates the sampling quantity for each bin, ensuring that sampling behavior is accurately matched with the distribution of business risks.

[0016] 4. By automating real-time verification of snapshot data during quality inspection, the problem of existing technologies lacking systematic verification methods and relying on manual item-by-item comparison is solved. The verification covers multiple key fields such as VIN code format and uniqueness, registration date consistency and usage period, license plate number format and uniqueness, and document photo compliance. The verification results automatically return abnormal field identifiers and reason descriptions, helping quality inspectors quickly locate problems and reduce human error.

[0017] 5. By linking quality inspection conclusions with audit records and auditor information through work order IDs and performing multi-dimensional aggregation statistics, the problem of single-dimensional quality analysis and inability to accurately attribute causes in existing technologies is solved. Audit error rates and error type distributions can be calculated by auditor, audit type, city, vehicle type, date, etc., enabling refined management and accurate attribution of audit quality, and providing data support for process improvement and personnel training.

[0018] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram illustrating a full-link data processing and verification process for auditing based on a globally unique identifier, according to an exemplary embodiment. Figure 2 This is a schematic diagram illustrating the composition of a full-link data processing and verification system for auditing based on a globally unique identifier, according to an exemplary embodiment. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be described in detail below. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other implementation methods obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] Example 1 In one embodiment, see Figure 1 , Figure 1 This is a schematic diagram illustrating a full-link data processing and verification process for auditing based on a globally unique identifier, according to an exemplary embodiment. The method includes: S1. This embodiment uses the quality inspection of three types of review work orders—driver recruitment, vehicle replacement, and information change—on the application scenario of a ride-hailing platform. The system first obtains the application request information submitted by the client. This application request information includes the application type and application form information. The application type is one of the aforementioned driver recruitment, vehicle replacement, and information change. The application form information includes at least one of driver information, vehicle information, and document information. The driver information specifically includes the driver's name, mobile phone number, ID card number, driver's license number, driving experience, professional qualification certificate number, and account status; the vehicle information includes the license plate number, brand, model, vehicle class, VIN code, engine number, registration date, and usage nature; the document information includes photos and corresponding text recognition content of the ID card, driver's license, vehicle registration certificate, and operating permit.

[0023] S2. In response to the received application request, the system generates a globally unique work order ID. This work order ID is generated using the Snowflake algorithm by a distributed ID service when the application request reaches the gateway. The generated work order ID is a 64-bit long integer value, which is converted to a 16-bit fixed-length hexadecimal string and returned to all subsequent stages. The system writes this work order ID to the application record table and simultaneously creates an audit record in the audit record table. This audit record has an audit record ID and is associated with the aforementioned work order ID. The audit status field in the audit record table is initially set to "pending audit".

[0024] S3. The system continuously monitors the review status field in the review record table. The monitoring method is periodic polling, with a configurable polling interval, such as once per second. When the review status field is updated from pending review to review completed, the system uses the review record ID to query the driver information database, vehicle information database, document information database, and application form database to obtain the driver information, vehicle information, document information, and application form content at the time of review completion. After serializing the above information, the system persistently stores it in the review snapshot table using the work order ID as the unique primary key.

[0025] S4. After the snapshot data of completed work orders is available in the review snapshot table, the system reads the administrator's preset sampling strategy configuration. This sampling strategy configuration includes grouping dimensions, sampling ratio, and execution frequency. The grouping dimensions include review type, city, vehicle type, and reviewer. The execution frequency is divided into two types: cyclic execution and single execution. Cyclic execution means triggering at preset intervals, such as sampling the previous day's (T+1) data at midnight every day; single execution means triggering once at a specified time. When the triggering condition corresponding to the execution frequency is met, the system queries the work order records in the review record table where the review status field is a review completion identifier and the corresponding work order ID has not yet appeared in the quality inspection task table. The query results are grouped according to the grouping dimensions in the strategy configuration to form multiple independent sampling buckets. The number of samples to be extracted is calculated according to the corresponding sampling ratio for each sampling bucket and the extraction is performed. Based on the extraction results, a quality inspection task is generated and written to the quality inspection task table. Each quality inspection task record is associated with the work order ID of the extracted work order.

[0026] S5. When a quality inspector claims or opens a quality inspection task on the quality inspection workbench, the system reads the corresponding snapshot data from the audit snapshot table using the work order ID associated with the task, verifies the preset key fields in the snapshot data, generates a verification result, and generates a quality inspection conclusion based on the verification result. The quality inspection conclusion includes two states: quality inspection passed and quality inspection failed. When the quality inspection fails, the error type and error description are further marked.

[0027] Example 2 This embodiment further defines the method for generating work order IDs. When the system responds to the received application request information, it executes the following steps to generate a globally unique work order ID. First, it obtains the current millisecond-level timestamp of the server when the application request information is received. Preferably, this timestamp uses 00:00:00 on January 1, 2025 as the zero point reference and records the millisecond offset from the reference time to the current moment. Second, it obtains the machine identifier of the service node that processes the application request. This machine identifier is specified through a configuration file or environment variable when the service node is deployed and is an integer value between 0 and 1023. The machine identifiers of each service node are different. Finally, it obtains the number of work orders generated by the service node in the current millisecond. This work order count increments automatically within the same millisecond, with a count value ranging from 0 to 4095. The count is incremented by 1 for each work order ID generated within the same millisecond, and is reset to zero when the millisecond time changes.

[0028] The system combines the timestamp, machine identifier, and work order count according to the snowflake algorithm structure to generate a work order ID. The final generated work order ID is a 64-bit long integer value, whose structure from left to right is: 1 sign bit (fixed to 0, indicating a positive number), 41-bit timestamp, 10-bit machine ID, and 12-bit sequence number. After generation, the 64-bit value is converted into a 16-bit fixed-length hexadecimal string, which serves as a globally unique work order ID throughout the entire process of subsequent review, snapshot generation, spot checks, and quality inspection. Among them, the 41-bit timestamp ensures that the work order ID increases in a macro time dimension, the 10-bit machine ID isolates different service nodes, and the 12-bit sequence number ensures the concurrent uniqueness of a single node within the same millisecond.

[0029] Example 3 This embodiment further defines the serialization and storage method of snapshot data based on Embodiment 1. When the system needs to serialize the acquired driver information, vehicle information, document information, and application form content and store them in the audit snapshot table, the following operations are performed. The system first encapsulates the above four types of information into a data object, then performs JSON serialization processing on the data object, calling the JSON serialization function to convert the data object into a JSON format string. Finally, the JSON string is written into the snapshot data field of the audit snapshot table, and the work order ID is set as the unique index field of the audit snapshot table. The constraint of the unique index field ensures that there can only be one record with the same work order ID in the audit snapshot table. If the work order ID is duplicated during insertion, the insertion operation is rejected by the database, thereby ensuring the uniqueness and consistency of the snapshot data.

[0030] Example 4 The audit snapshot table is the core data table that stores a snapshot of all business data at the moment the audit is completed. It contains at least the following fields: Primary Key ID, which is an auto-incrementing bigint type and serves as the physical primary key in the table; Work Order ID, which is a varchar(64) type and stores a globally unique work order ID; Audit Record ID, which is a bigint type and is associated with the audit record table; Audit Type, which is a tinyint type and has a value of 1 for driver recruitment, 2 for vehicle replacement, and 3 for information change; Audit Result, which is a tinyint type and has a value of 1 for audit approval and 2 for audit rejection; Audit Completion Time, which is a datetime type and records the exact moment when the audit status is updated to the completion mark; Auditor ID, which is a varchar(32) type and records the auditor ID that performed the audit operation; Driver ID, which is a varchar(32) type and has a value of 32 for driver ID. The `har(32)` type records the identifier of the driver being reviewed; the driver information snapshot is of type JSON and stores the full information of the driver at the time of review completion; the vehicle ID is of type varchar(32) and records the identifier of the vehicle being reviewed. It can be empty when the review type is driver recruitment and there is no vehicle information; the vehicle information snapshot is of type JSON and stores the full information of the vehicle at the time of review completion; the document information snapshot is of type JSON and stores the photos and text information of various documents at the time of review completion; the application form snapshot is of type JSON and stores the content of the application form at the time of review completion; the snapshot version number is of type varchar(16) and the default value is 1.0, which is used to identify the version of the snapshot data structure; the creation time is of type datetime and the default value is the current timestamp, which records the insertion time of the snapshot record.

[0031] Example 5 This embodiment further defines the proportion calculation and extraction method within the sampling bucket based on Embodiment 1.

[0032] When the system calculates and executes the sampling for each sampling bin according to the corresponding sampling ratio, a segmented sampling strategy is adopted. First, the work orders in each sampling bin are divided into several segments according to a preset fixed number. This preset fixed number can be set by the administrator in the strategy configuration, for example, the default is 100. The work orders in the bin are arranged in ascending order according to the approval completion time, and every 100 work orders starting from the first one are divided into segments. For each segment, the corresponding number of work orders are extracted according to the sampling ratio corresponding to that bin. For example, when the sampling ratio is 10%, the first 10 work orders are extracted from the 100 work orders in each segment. For the remaining work orders that are less than the preset fixed number after division, the remaining number of work orders is multiplied by the sampling ratio and rounded down, and the corresponding number of work orders are extracted. For example, when there are 50 remaining work orders and the sampling ratio is 10%, 50 × 10% = 5 is calculated, and the first 5 work orders are extracted. The work orders extracted from each segment and the remaining part are uniformly generated into quality inspection tasks and written to the quality inspection task table in batches.

[0033] Example 6 After grouping the queried work orders into multiple independent sampling buckets according to the grouping dimension, the system performs double deduplication filtering before extracting from each sampling bucket and writing to the quality inspection task table. The first deduplication is performed in memory: the system maintains a Set collection in memory to store the work order IDs of the extracted work orders. When extracting a new work order, it first checks whether the work order ID already exists in the Set collection. If it exists, it means that the work order has been extracted in the current sampling process, so the work order is skipped and the next one is processed. If it does not exist, the work order ID is added to the Set collection and the next step is taken. The Set collection is implemented using a hash structure, and the time complexity of a single deduplication operation is O(1), which can efficiently handle the deduplication judgment of a large number of work orders.

[0034] The second layer of deduplication is performed at the database level: when the extracted work order IDs are written to the quality inspection task table, this table has a unique index constraint based on the work order ID field. If the attempted write work order ID conflicts with the primary key of an existing record in the table, the database will reject the write operation and return an error. The system catches this error and skips the work order, ensuring that the same work order is not written to the quality inspection task table repeatedly. These two deduplication mechanisms complement each other: the in-memory Set ensures deduplication efficiency during a single sampling inspection, while the database unique index acts as a fallback mechanism to guarantee data consistency at the persistence level.

[0035] Example 7 Based on Example 1, the verification content of preset key fields in the snapshot data is further defined.

[0036] When the system validates the preset key fields in the snapshot data, it performs at least one of the following four types of validation operations: The first category involves format verification and global uniqueness verification of the vehicle VIN code. Format verification checks whether the VIN code consists of a preset number of letters and numbers. Specifically, it verifies whether the VIN code string is 17 characters long and whether each character belongs to a valid character set of letters (AZ, excluding I, O, Q) or numbers (0-9). Global uniqueness verification queries the vehicle information table in the database to confirm whether the VIN code has already been used by another vehicle. If the query result is empty, the VIN code passes the uniqueness verification; if another vehicle is found to have used the same VIN code, the verification fails.

[0037] The second category involves consistency verification of the registration date and usage period verification. Consistency verification compares the registration date with the initial registration date recorded on the vehicle registration certificate to ensure complete consistency, and calculates the difference between the registration date and the vehicle's manufacturing date. It then determines whether this difference falls within a preset date difference range. This preset range is typically set to be within a certain number of months from the manufacturing date; exceeding this range indicates a lack of reasonable logical connection between the registration date and the manufacturing date. Usage period verification calculates the number of years since the registration date and compares this value with a preset usage period threshold. This preset usage period threshold is set according to the platform's entry requirements for different vehicle models. For example, the entry threshold for comfort vehicles is a certain number of years; vehicles exceeding this threshold are not permitted entry.

[0038] The third category involves format validation and global uniqueness validation of license plate numbers. Format validation verifies whether the license plate number string conforms to preset format rules. These rules include: a total length of 7 to 8 characters, the first character being the province abbreviation in Chinese, the second character being a letter, and subsequent characters being a combination of letters and numbers. Global uniqueness validation queries the vehicle information table in the database to confirm whether the license plate number has already been used by another vehicle.

[0039] The fourth category involves compliance verification of ID photos. This compliance verification includes the following checks: 1) Checking whether the ID photo was taken in a real-world setting, i.e., using image analysis algorithms to determine whether the photo is a direct photograph of the physical document or a secondary photograph of an electronic screen, paper copy, or reproduction; 2) Checking for any signs of modification, i.e., using image tampering detection algorithms to determine whether the photo has been cropped, spliced, or had text modified by image editing software; 3) Checking for any obstructions or reflective areas, i.e., using image quality analysis algorithms to determine whether the document information area in the photo is obstructed by objects or has strong light reflection that makes the information unreadable; 4) Checking whether the ID photo is within its validity period, i.e., extracting the expiration date field of the document in the photo and comparing it with the current date to determine if it has expired.

[0040] Specifically, the system detects whether the ID photo was taken in a real-world setting. The implementation involves performing a Fourier transform on the ID photo to obtain a spectrum; calculating the ratio of high-frequency energy to total energy in the spectrum; and determining if the ratio is lower than a preset threshold for determining if the photo is a copy. The principle is that when a secondary photograph is taken of an electronic screen or a paper copy, the presence of pixel grids on the screen or surface texture of the paper causes a regular attenuation of high-frequency components in the spectrum of the captured image, resulting in a significantly lower proportion of high-frequency energy compared to directly photographing a physical ID. In this embodiment, the preferred threshold for determining if the photo is a copy is 0.15, meaning that a high-frequency energy proportion below 15% is considered a copy.

[0041] This method detects whether a document photo has been modified. Specifically, the document photo is divided into multiple image blocks of the same size; the standard deviation of the pixel values ​​for each image block is calculated; if the difference between the standard deviation of a certain image block and the standard deviations of its adjacent image blocks exceeds a preset multiple of the average standard deviation of the adjacent blocks, then the area containing that image block is determined to have been modified. The principle is that the texture changes in an unmodified natural image within a local area are smooth transitions, while operations such as splicing and text modification disrupt this smoothness, creating abrupt changes in standard deviation at the boundaries of the modified area. In this embodiment, the image block size is preferably 32×32 pixels, and the preset multiple is preferably 2 times, meaning that when the difference between the standard deviation of a certain image block and the average standard deviation of its adjacent blocks exceeds twice the average standard deviation of the adjacent blocks, it is determined to be abnormal.

[0042] The system detects whether a document photo contains reflective areas. Specifically, it converts the document photo to a grayscale image, marks pixels with grayscale values ​​higher than a preset highlight threshold as highlight pixels, aggregates interconnected highlight pixels into connected regions, and determines a connected region as a reflective area if its area exceeds a preset area threshold. In this embodiment, the grayscale value range of the document photo is 0 to 255, the preset highlight threshold is preferably 220 (i.e., pixels with grayscale values ​​exceeding 220 are determined to be highlight pixels), and the preset area threshold is preferably 5% of the total pixel area of ​​the document photo (i.e., a connected region is determined to be a reflective area when its area exceeds 5% of the total photo area).

[0043] The system detects whether an ID photo is obstructed. Specifically, it performs edge detection on the ID photo and extracts the document's border outline. If the outline has an unclosed gap, and the pixel brightness variance at the gap is lower than a preset texture judgment value, then the gap is determined to be obstructed. In this embodiment, the preset texture judgment value is preferably 50, meaning that when the pixel brightness variance in the gap area is lower than 50, it indicates a lack of texture variation in that area, and is determined to be caused by an obstruction.

[0044] The system checks whether a photo of an ID document is valid. Specifically, it performs optical character recognition (OCR) on the photo to obtain text lines and their corresponding text content. It then searches the recognition results for any keyword among "validity period," "expiration date," and "expiration date." The system extracts the date text from the line containing that keyword, parses it into year, month, and day values, and compares these values ​​with the current system date. If the current system date is later than the extracted date, the document is considered expired.

[0045] Example 8 The execution method for each verification operation is further defined. When the system performs verification operations on preset key fields in the snapshot data, an asynchronous call and concurrent processing architecture is adopted. First, after loading the snapshot data and completing page rendering, the quality inspection front-end sends a verification request to the server through an asynchronous interface. This asynchronous interface is implemented based on the asynchronous request mechanism of the HTTP protocol. After the front-end sends the request, it does not block the main thread of the page, and the quality inspector can browse the snapshot data synchronously. The verification request carries the work order ID of the current work order and the snapshot data to be verified, including the VIN code, registration date, license plate number, and the storage address or content of the document photo.

[0046] Upon receiving a verification request, the server processes the multiple verification operations involved in the request concurrently. Specifically, the four tasks—VIN code verification, registration date verification, license plate number verification, and ID photo verification—are submitted to different threads in a thread pool for parallel execution. Each thread independently completes its own verification logic and returns a result. After all threads have finished executing, the server aggregates the return results from each thread and encapsulates them into a single verification result object. This verification result includes error field identifiers and error reason descriptions. The error field identifiers indicate which field is abnormal, and the error reason description provides a detailed explanation of the exception.

[0047] After the server returns the validation results to the front end, the front end, based on the error field identifiers in the validation results, marks the abnormal fields on the quality inspection interface with a preset highlighting style. The preset highlighting style includes adding an orange or red background color to the text area or input box of the abnormal field, and displaying a floating tooltip next to the abnormal field containing a description of the error reason. The highlighting operation is implemented by modifying the CSS class name of the DOM element, without blocking page interaction.

[0048] Example 9 Based on Example 1, the statistical analysis steps after the quality inspection conclusion is generated are further defined. After the system generates the quality inspection conclusion, the following statistical analysis processing is also performed. The system uses the work order ID to perform a correlation query with the corresponding audit record and auditor information. The specific method of correlation query is as follows: using the work order ID as the connection key, the quality inspection task table, audit record table, and auditor information table are joined to obtain the complete business context corresponding to each quality inspection conclusion. Then, the correlation results are grouped and statistically analyzed according to the preset aggregation dimensions. The aggregation dimensions include at least one of auditor, audit type, city, vehicle type, and date. The core indicator of statistical calculation is the audit error rate, and the calculation formula is: Audit error rate = Number of work orders with a quality inspection conclusion of failure under this dimension ÷ Total number of work orders sampled under this dimension 100%; Percentage of a certain type of error = Number of errors of that type ÷ Total number of errors × 100%; Quality inspector completion rate = (Number of quality inspections completed ÷ Total number of assigned items) × 100%; City audit quality = Number of incorrect orders in the city ÷ Total number of random inspections in the city × 100%.

[0049] In addition, it includes error type distribution statistics, which shows the percentage of each error type in the total number of errors in that dimension, and quality inspector completion rate statistics, which shows the percentage of quality inspectors who have completed their inspection tasks out of the total number of assigned tasks. The statistical results are finally output in the form of data reports, supporting visualization charts such as line charts and bar charts, and can generate periodic reports by time granularity such as day, week, and month.

[0050] Example 10 Please see Figure 2 , Figure 2 A data processing and verification system for the entire audit chain based on a globally unique identifier is provided. This system is used to execute the method described in any of the foregoing embodiments. The system includes an acquisition module, a first processing module, a second processing module, a third processing module, and a fourth processing module. Each module can be physically deployed on the same server or distributed across different server nodes. Data is transferred between modules through internal interfaces or message queues.

[0051] The acquisition module is used to retrieve application request information submitted by the client. This application request information includes the application type and application form information. The application type includes driver recruitment, vehicle replacement, and information modification. The application form information includes at least one of driver information, vehicle information, and document information. The acquisition module exposes a unified HTTP interface to receive application requests from the driver's app or the management backend, and performs format validation and legality checks on the request parameters.

[0052] The first processing module responds to the application request information received by the acquisition module by calling the distributed ID service to generate a globally unique work order ID, writing the work order ID into the application record table, and simultaneously creating an approval record in the approval record table. This approval record has an approval record ID associated with the work order ID, and the approval status field in the approval record table is initially set to pending approval.

[0053] The second processing module is used to detect the audit status field in the audit record table. When the audit status field is detected to have been updated from pending audit to audit completed, the module queries the driver information database, vehicle information database, document information database, and application form database based on the audit record ID to obtain the driver information, vehicle information, document information, and application form content at the current moment. The obtained information is then serialized and written into the audit snapshot table for storage using the work order ID as the unique primary key.

[0054] The third processing module reads the preset sampling strategy configuration. This configuration includes grouping dimensions, sampling ratios, and execution frequencies. Grouping dimensions include audit type, city, vehicle type, and auditor. When the triggering condition corresponding to the execution frequency is met, the third processing module queries the audit record table for work orders whose audit status field is an audit completion identifier and whose corresponding work order ID does not exist in the quality inspection task table. It groups the queried work orders by grouping dimensions to obtain multiple independent sampling buckets. For each sampling bucket, it calculates the sampling quantity according to the corresponding sampling ratio and executes the sampling, generating a quality inspection task and writing it to the quality inspection task table. Each quality inspection task record is associated with the work order ID of the sampled work order. The third processing module internally includes a timed scheduling submodule, responsible for triggering the sampling process according to the execution frequencies configured for each strategy.

[0055] The fourth processing module is used to read the corresponding snapshot data from the audit snapshot table using the work order ID when a quality inspection task is executed. It then validates preset key fields in the snapshot data, generates validation results, and produces a quality inspection conclusion based on these results. This fourth processing module is also responsible for writing the quality inspection conclusion into the quality inspection result table and linking the quality inspection conclusion with the audit record and auditor information using the work order ID, providing a data foundation for subsequent performance statistical analysis.

[0056] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0057] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for end-to-end data processing and verification in auditing based on globally unique identifiers, characterized in that, The method includes: Obtain application request information submitted by the client; the application request information includes application type and application form information, the application type includes driver recruitment, vehicle replacement, and information change, and the application form information includes at least one of driver information, vehicle information, and document information; In response to the application request information, a globally unique work order ID is generated, the work order ID is written into the application record table, and an approval record is generated and stored in the approval record table. The audit record has an audit record ID and is associated with the work order ID, and the audit status field in the audit record table is initially set to pending audit status. The review status field in the review record table is detected. When the review status field is updated from pending review to review completion, the driver information database, vehicle information database, document information database, and application form database are queried according to the review record ID to obtain the driver information, vehicle information, document information, and application form content at the current time. The obtained driver information, vehicle information, document information, and application form content are serialized and written into the review snapshot table for storage using the work order ID as the unique primary key. Based on the results of writing the work order ID as a unique primary key into the audit snapshot table, the preset sampling strategy configuration is read. The sampling strategy configuration includes grouping dimensions, sampling ratio, and execution frequency. When the triggering condition corresponding to the execution frequency is met, the work orders in the audit record table whose audit status field is an audit completion identifier and whose corresponding work order ID does not exist in the quality inspection task table are queried. The queried work orders are grouped according to the grouping dimensions to obtain multiple independent sampling buckets. For each sampling bucket, the number of samples is calculated according to the sampling ratio corresponding to the sampling bucket and the sampling is performed. A quality inspection task is generated and written to the quality inspection task table. Each quality inspection task record is associated with the work order ID of the sampled work order. The grouping dimensions include audit type, city, vehicle type, and auditor. When the quality inspection task is executed, the corresponding snapshot data is read from the audit snapshot table using the work order ID, the preset key fields in the snapshot data are verified, the verification result is generated, and a quality inspection report is generated based on the verification result.

2. The method according to claim 1, characterized in that, In response to the application request information, a globally unique work order ID is generated, including: In response to the obtained application request information, the server obtains the current millisecond-level timestamp of the server when the application request information is received, the machine identifier of the service node that processes the application request, and the number of work orders generated by the service node in the current millisecond; The server's current millisecond-level timestamp at the time of the application request, the machine identifier of the service node making the application request, and the number of work orders generated by the service node in the current millisecond are combined according to the snowflake algorithm to generate a work order ID; The machine identifier is a fixed value configured during the deployment of the service node, and the work order count increments automatically within the same millisecond.

3. The method according to claim 1, characterized in that, The step of serializing the obtained driver information, vehicle information, document information, and application form content, and then storing them in the audit snapshot table using the work order ID as the unique primary key, includes: The driver information, vehicle information, document information, and application form content are serialized into a JSON string. Write the JSON string into the snapshot data field of the audit snapshot table, and set the work order ID as the unique index field of the audit snapshot table.

4. The method according to claim 3, characterized in that, The audit snapshot table must contain at least the following fields: primary key ID, work order ID, audit record ID, audit type, audit result, audit completion time, auditor ID, driver ID, driver information snapshot, vehicle ID, vehicle information snapshot, document information snapshot, application form snapshot, snapshot version number, and creation time.

5. The method according to claim 1, characterized in that, For each sampling bin, the sampling quantity is calculated according to the sampling ratio corresponding to that bin, and the sampling is performed to generate a quality inspection task, which is then written into the quality inspection task table. This includes: The work orders in each sampling bin are divided into several pieces according to a preset fixed number. For each piece, the corresponding number of work orders are extracted according to the sampling ratio. The work orders are used to generate quality inspection tasks and written into the quality inspection task table. For any remaining work orders that are less than the preset fixed number, the remaining number of work orders is multiplied by the sampling ratio and rounded down. Then, a corresponding number of work orders are selected, and the work orders are used to generate quality inspection tasks, which are then written into the quality inspection task table.

6. The method according to claim 5, characterized in that, After grouping the queried work orders according to the grouping dimension to obtain multiple independent sampling buckets, for each sampling bucket, before calculating the sampling quantity according to the sampling ratio corresponding to that bucket and performing the sampling, and before generating a quality inspection task and writing it into the quality inspection task table, the process further includes: The work order IDs of the extracted work orders are stored in a Set collection in memory. When extracting a new work order, it is first determined whether the work order ID already exists in the Set collection. If it does, the work order is skipped. When writing the extracted work order ID into the quality inspection task table, if the work order ID violates the unique index constraint established based on the work order ID field in the quality inspection task table, the write operation is rejected.

7. The method according to claim 1, characterized in that, The validation of preset key fields in the snapshot data includes: The vehicle VIN code is subjected to format verification and global uniqueness verification; the format verification is to verify whether the VIN code is composed of a preset number of alphanumeric characters, and the global uniqueness verification is to query the database to confirm that the VIN code has not been used by other vehicles. The registration date is checked for consistency and the service life is checked. The consistency check is to compare the logical relationship between the registration date, the vehicle registration date and the vehicle manufacturing date. The service life check is to calculate the number of years from the registration date to the current time and compare it with a preset service life threshold. The license plate number is subjected to format verification and global uniqueness verification; the format verification is to verify whether the license plate number conforms to the preset license plate number format rules, and the global uniqueness verification is to query the database to confirm that the license plate number is not occupied by other vehicles; The compliance verification of the ID photo includes: checking whether the ID photo was taken in a real scene, checking whether the ID photo has been modified, checking whether the ID photo has any obstructed or reflective areas, and checking whether the ID photo has expired.

8. The method according to claim 7, characterized in that, The verification operation is performed in the following manner: A verification request is sent to the server via an asynchronous interface; the verification request carries the work order ID and the snapshot data to be verified. The server performs concurrent processing on the multiple verification operations involved in the verification request to generate a verification result. The verification result includes an error field identifier and a description of the error reason; After receiving the verification result, based on the error field identifier, the fields with abnormalities are marked with a preset highlighting style on the preset quality inspection interface to obtain the quality inspection result.

9. The method according to claim 1, characterized in that, Also includes: The quality inspection results are associated with the corresponding audit records and auditor information through the work order ID, and statistical analysis is performed in groups according to preset aggregation dimensions to calculate the audit error rate under each aggregation dimension; the aggregation dimensions include at least one of auditor, audit type, city, vehicle type, and date.

10. A data processing and verification system for the entire audit chain based on a globally unique identifier, applied to the data processing and verification method for the entire audit chain based on a globally unique identifier as described in any one of claims 1-9, characterized in that, The system includes: The acquisition module is used to acquire application request information submitted by the client; the application request information includes application type and application form information, the application type includes driver recruitment, vehicle replacement, and information change, and the application form information includes at least one of driver information, vehicle information, and document information; The first processing module is used to respond to the application request information, generate a globally unique work order ID, write the work order ID into the application record table, and at the same time generate an audit record and store it in the audit record table. The audit record has an audit record ID and is associated with the work order ID, and the audit status field in the audit record table is initially set to pending audit status. The second processing module is used to detect the audit status field in the audit record table. When the audit status field is updated from pending audit status to audit completion identifier, the module queries the driver information database, vehicle information database, document information database and application form database according to the audit record ID to obtain the driver information, vehicle information, document information and application form content at the current time. After serializing the obtained driver information, vehicle information, document information and application form content, the module writes them into the audit snapshot table for storage with the work order ID as the unique primary key. The third processing module is used to read the results stored in the audit snapshot table based on the work order ID as the unique primary key, read the preset sampling strategy configuration, which includes grouping dimensions, sampling ratio, and execution frequency. The grouping dimensions include audit type, city, vehicle type, and auditor. When the triggering condition corresponding to the execution frequency is met, the module queries the audit record table for work orders whose audit status field is an audit completion identifier and whose corresponding work order ID does not exist in the quality inspection task table. The module groups the queried work orders according to the grouping dimensions to obtain multiple independent sampling buckets. For each sampling bucket, the module calculates the sampling quantity according to the sampling ratio corresponding to the sampling bucket and performs the sampling, generating a quality inspection task and writing it to the quality inspection task table. Each quality inspection task record is associated with the work order ID of the sampled work order. The fourth processing module is used to read the corresponding snapshot data from the audit snapshot table through the work order ID when the quality inspection task is executed, verify the preset key fields in the snapshot data, generate verification results, and generate quality inspection conclusions based on the verification results.

Images