Document auditing method and apparatus, device, and storage medium

Abstract

The application belongs to the technical field of data processing, and discloses a document auditing method and device, equipment and a storage medium. The application inputs each block content into a preset auditing model and a preset rule engine for auditing to obtain a first auditing result and a second auditing result, then determines a block auditing result corresponding to each block content according to the first auditing result and / or the second auditing result, and then combines the block auditing results corresponding to each block content to obtain a document auditing result corresponding to the to-be-audited document. The application first performs block processing on the to-be-audited document, then inputs each block content into the preset auditing model and the preset rule engine for parallel auditing, can obtain the first auditing result and the second auditing result corresponding to each block content, combines the block auditing results corresponding to each block content, and obtains the document auditing result corresponding to the entire to-be-audited document, thereby reducing the total time consumption of document auditing and improving the efficiency of document auditing.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a document review method, apparatus, device, and storage medium. Background Technology

[0002] Document review is a core step in the compliance check of credential documents, aiming to automatically identify risks, non-compliance items, or missing information. Currently, document review typically employs a sequential execution method using models and rule engines. The total time consumed is the sum of the processing times of the model and rule engines, and each subsequent step must wait for the previous step to complete completely. This results in lengthy and inefficient document review processes. Summary of the Invention

[0003] The main objective of this application is to provide a document review method, apparatus, device, and storage medium, aiming to solve the technical problem of how to reduce the total time spent on document review and improve the efficiency of document review.

[0004] To achieve the above objectives, this application provides a document review method, which includes the following steps: The document to be reviewed is divided into blocks, resulting in multiple blocks of content. Each block of content is input into the preset review model and the preset rule engine for review, resulting in the first review result and the second review result. Determine the sub-block review result corresponding to each sub-block content based on the first review result and / or the second review result; The review results of each block are merged to obtain the document review result corresponding to the document to be reviewed.

[0005] Optionally, the step of inputting each block of content into a preset review model for review to obtain a first review result includes: Determine the document location of each section's content within the document to be reviewed; Based on the document location, determine the context content corresponding to each block of content, and construct enhanced content based on the block of content and the context content; The enhanced content is input into a preset review model for review, and a first review result is obtained.

[0006] Optionally, the step of inputting the enhanced content into a preset rule engine for review to obtain a second review result includes: Determine the fields to be reviewed corresponding to the content to be reviewed in the enhanced content, wherein the fields to be reviewed include voucher number and attribute identifier; Search the second-level cache of the preset database for the attribute name corresponding to the credential number and the attribute identifier; If the target attribute corresponding to the attribute name does not exist, the attribute list corresponding to the voucher number is searched from the first-level cache of the preset database; Filter the target attributes that match the attribute identifier from the attribute list, and determine the target field corresponding to the target attribute; The target field value is input into the preset rule engine for review, and a second review result is obtained.

[0007] Optionally, after determining the field to be reviewed corresponding to the content to be reviewed in the enhanced content, the method further includes: Determine whether the field to be reviewed is configured with a manual correction value; If the manually corrected value is not empty, then the manually corrected value is used as the target field value; If the manually corrected value is empty, then the attribute list corresponding to the voucher number is retrieved from the first-level cache of the preset database.

[0008] Optionally, merging the block review results corresponding to each block of content to obtain the document review result corresponding to the document to be reviewed includes: Determine the block priority of the block review results corresponding to each block content; Select the target priority from the block priorities; The document review result corresponding to the document to be reviewed is determined based on the target review result corresponding to the target priority.

[0009] Optionally, before dividing the document to be reviewed into multiple blocks, the process further includes: Determine the document length corresponding to the document to be reviewed; If the document length is less than the preset length, the document to be reviewed is input into the preset review model and the preset rule engine for review, and the document review result corresponding to the document to be reviewed is determined according to the obtained third review result and fourth review result. If the document length is greater than or equal to the preset length, the document to be reviewed will be divided into blocks.

[0010] Optionally, after merging the block review results corresponding to each block of content to obtain the document review result corresponding to the document to be reviewed, the method further includes: Determine whether the first review result and the second review result are consistent; If not, the document to be reviewed will be pushed to the manual review queue.

[0011] Furthermore, to achieve the above objectives, this application also provides a document review device, the document review device comprising: The document segmentation module is used to segment documents to be reviewed into multiple segments. The content review module is used to input each block of content into a preset review model and a preset rule engine for review, and obtain the first review result and the second review result. The result determination module is used to determine the sub-block review result corresponding to each sub-block content based on the first review result and / or the second review result. The result merging module is used to merge the block review results corresponding to each block of content to obtain the document review result corresponding to the document to be reviewed.

[0012] In addition, to achieve the above objectives, this application also proposes a document review device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the document review method described above.

[0013] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and which, when executed by a processor, implements the steps of the document review method described above.

[0014] This application divides the document to be reviewed into multiple blocks, then inputs each block into a preset review model and a preset rule engine for parallel review, yielding a first review result and a second review result. Based on the first and / or second review results, a block review result is determined for each block. Finally, these block review results are merged to obtain the final document review result for the entire document. This approach reduces the overall review time and improves review efficiency. Attached Figure Description

[0015] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0016] 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, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the first embodiment of the document review method for this application. Figure 2 This is a flowchart illustrating the second embodiment of the document review method for this application. Figure 3 This is a flowchart illustrating the third embodiment of the document review method for this application. Figure 4 This is a schematic diagram of the overall process of one embodiment of the document review method of this application; Figure 5 This is a structural block diagram of the first embodiment of the document review device for this application; Figure 6 This is a schematic diagram of the structure of a document review device in the hardware operating environment involved in the embodiments of this application.

[0018] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0020] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0021] The main solution of this application embodiment is: to divide the document to be reviewed into blocks to obtain multiple blocks; to input each block into a preset review model and a preset rule engine for review to obtain a first review result and a second review result; to determine the block review result corresponding to each block based on the first review result and / or the second review result; and to merge the block review results corresponding to each block to obtain the document review result corresponding to the document to be reviewed.

[0022] Document review is a core step in the compliance check of credential documents, aiming to automatically identify risks, non-compliance items, or missing information. Currently, document review typically employs a sequential execution method using models and rule engines. The total time consumed is the sum of the processing times of the model and rule engines, and each subsequent step must wait for the previous step to complete completely. This results in lengthy and inefficient document review processes.

[0023] This application divides the document to be reviewed into multiple blocks, then inputs each block into a preset review model and a preset rule engine for parallel review, yielding a first review result and a second review result. Based on the first and / or second review results, a block review result is determined for each block. Finally, these block review results are merged to obtain the final document review result for the entire document. This approach reduces the overall review time and improves review efficiency.

[0024] It should be noted that the executing entity of this application can be a computing service device with data processing, network communication, and program execution functions, such as a computer, or an electronic device or document review device capable of performing the above functions. The following description uses a document review device as an example to illustrate this embodiment and the subsequent embodiments.

[0025] Based on this, this application provides a document review method, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the document review method for this application.

[0026] In this embodiment, the document review method includes the following steps: Step S10: Divide the document to be reviewed into blocks to obtain multiple blocks of content.

[0027] Understandably, documents pending review refer to voucher documents that need to be reviewed. Voucher documents are written documents that record the occurrence or completion of economic transactions. They are the basis for financial accounting, audit tracking and business review. They usually refer to business documents with evidentiary effect, such as invoices, purchase orders, warehouse receipts, acceptance slips and contracts.

[0028] It should be understood that this embodiment can segment the document to be reviewed. In one feasible embodiment, semantic segmentation can be performed based on the title level, table boundaries, etc. of the document's structure tree, rather than fixed page number segmentation. Segmentation overlap areas can also be set, such as the end of the previous segment overlapping the beginning of the next segment by N lines to ensure semantic continuity across segments. The optimal segment size can also be dynamically calculated based on the preset review model context window length, and the size of each segment can be the context window length, resulting in multiple segment contents.

[0029] Furthermore, in order to effectively determine whether document segmentation is required, in this embodiment, before step S10, the method further includes: determining the document length corresponding to the document to be reviewed; if the document length is less than a preset length, the document to be reviewed is input into a preset review model and a preset rule engine for review, and the document review result corresponding to the document to be reviewed is determined based on the obtained third review result and fourth review result; if the document length is greater than or equal to the preset length, the document to be reviewed is segmented.

[0030] Understandably, the document length of the document to be reviewed can be determined, which could be the number of pages. If the document length is less than the preset length, it indicates that the document content is relatively short, and the document to be reviewed can be reviewed directly. Specifically, the document to be reviewed can be input into the preset review model and the preset rule engine for review, resulting in a third review result and a fourth review result. Then, based on the third and fourth review results, the document review result corresponding to the document to be reviewed can be determined, allowing for direct review of the document to be reviewed, thereby improving document review efficiency.

[0031] It should be understood that if the document length is greater than or equal to the preset length, it means that the document content is too large. In this case, the document to be reviewed can be divided into blocks.

[0032] Step S20: Input each block of content into the preset review model and the preset rule engine for review, and obtain the first review result and the second review result.

[0033] Understandably, each segment of content can be input into a preset review model and a preset rule engine for parallel review, yielding a first review result and a second review result. The preset review model can be a pre-trained AI review model, such as BERT or ResNet. The preset rule engine can be a review system based on predefined rules, for example, "IF Amount > 10000 THEN Trigger Large Transaction Review".

[0034] In the specific implementation, the audit code is as follows: ScaVouAuditServiceImpl.audit() uses CompletableFuture to achieve concurrent execution: / / Preset review model review tasks (asynchronously submitted to a dedicated thread pool) CompletableFuture <reviewresponse>modelFuture = CompletableFuture.supplyAsync(() ->{ return scaVouAuditManager.audit(classifiedItems, checkPoints, txnNo); }, auditAgentExecutor); / / Preset rule engine review task (asynchronously submitted to a dedicated thread pool) CompletableFuture <List <audititem>>ruleFuture = CompletableFuture.supplyAsync(() ->{ return scaVouAuditManager.ruleAudit(txnNo, sceneCode); }, auditAgentExecutor); / / Waiting for both tasks to complete CompletableFuture.allOf(modelFuture, ruleFuture).join(); / / Get results (Exception isolation: an exception in one task does not affect another) ReviewResponse modelResult = modelFuture.exceptionally(e ->fallbackResponse()).get(); List <audititem>ruleResult = ruleFuture.exceptionally(e ->Collections.emptyList()).get(); Thread pool isolation: auditAgentExecutor uses a dedicated thread pool for auditing, isolated from other business thread pools in the system, to prevent auditing tasks from blocking other business requests. Configuring an independent thread pool for auditing tasks, isolated from other business thread pools, prevents auditing tasks (which are typically time-consuming) from blocking other business requests and ensures overall system responsiveness.

[0035] Step S30: Determine the sub-block review result corresponding to each sub-block content based on the first review result and / or the second review result.

[0036] It should be understood that the block review result corresponding to each block of content can be determined based on the first review result and / or the second review result. If the first review result and the second review result are consistent, the first review result or the second review result can be used as the block review result corresponding to each block of content. If the first review result and the second review result are inconsistent, the more serious result will be used as the block review result. For example, if the first review result is "pass" and the second review result is "disagree", then the block review result will be "disagree". If the first review result is not identified due to an anomaly in the preset review model, then the second review result will be used as the block review result. If the second review result is not identified due to an anomaly in the preset rule engine, then the first review result will be used as the block review result.

[0037] Step S40: Merge the block review results corresponding to each block content to obtain the document review result corresponding to the document to be reviewed.

[0038] Understandably, the review results of each block can be merged. For example, if the review result of block 1 is "passed" and the review result of block 2 is "disagree", then the document review agency corresponding to the document to be reviewed will also be "disagree".

[0039] Furthermore, in this embodiment, after step S40, the method further includes: determining whether the first review result and the second review result are consistent; if not, pushing the document to be reviewed to the manual review queue.

[0040] It should be understood that it is possible to determine whether the first review result and the second review result are consistent. For example, if the first review result is "pass" and the second review result is "disagree", it indicates a high-risk scenario. If the first review result and the second review result are inconsistent, it can be marked as a conflict of conclusions and pushed to the manual review queue. The result of manual arbitration is taken as the final conclusion and fed back to the model for continuous learning.

[0041] This embodiment divides the document to be reviewed into multiple blocks, then inputs each block into a preset review model and a preset rule engine for review, obtaining a first review result and a second review result. Based on the first and / or second review results, the block review result corresponding to each block is determined. Finally, the block review results corresponding to each block are merged to obtain the final document review result for the entire document to be reviewed. This embodiment first divides the document to be reviewed into blocks, then inputs each block into a preset review model and a preset rule engine for parallel review, obtaining a first and second review result for each block. Finally, the block review results are merged to obtain the final document review result for the entire document to be reviewed, thereby reducing the total review time and improving the efficiency of document review.

[0042] refer to Figure 2 , Figure 2 This is a flowchart illustrating the second embodiment of the document review method for this application.

[0043] Based on the first embodiment described above, in this embodiment, step S20 includes: Step S201: Determine the document location of each block of content in the document to be reviewed.

[0044] Understandably, the document location of each content block within the document to be reviewed can be determined, for example, the page and line number of each content block within the document to be reviewed.

[0045] Step S202: Determine the context content corresponding to each block of content based on the document location, and construct enhanced content based on the block content and the context content.

[0046] It should be understood that the context content corresponding to each block can be determined based on the document location, and enhanced content can be constructed based on the block content and the context content. For example, the current review scope is the content of each block, and the context content includes the preceding context and the following context. The preceding context is for understanding only and is not within the scope of review. It can be the last 50 lines of the previous block. The following context is for understanding only and is not within the scope of review. It can be the first 50 lines of the next block.

[0047] In practice, through clear scope marking, the pre-defined review model can: understand references such as "the above clauses" and "the foregoing agreement" using the preceding context; understand forward references such as "details to follow" using the following context; and only output review opinions for the content within the "current review scope" without reviewing the context portion repeatedly.

[0048] Step S203: Input the enhanced content into the preset review model for review and obtain the first review result.

[0049] Understandably, the enhanced content can be input into a preset review model for review, resulting in the first review result, such as pass or fail.

[0050] This embodiment determines the document position of each content block within the document to be reviewed, then determines the corresponding contextual content based on the document position, constructs enhanced content based on the content blocks and contextual content, and finally inputs the enhanced content into a preset review model for review, obtaining a first review result. This embodiment constructs enhanced content based on the content blocks and contextual content, attaches preceding and following context during block review, and clearly marks the review scope, thus resolving the semantic understanding problem of cross-block clause references (such as "the aforementioned clauses" or "the foregoing agreement").

[0051] Step S204: Determine the fields to be reviewed corresponding to the content to be reviewed in the enhanced content. The fields to be reviewed include the voucher number and attribute identifier.

[0052] It should be understood that the fields to be reviewed in the enhanced content can be determined. The content to be reviewed can be segmented content. The fields to be reviewed include voucher number and attribute identifier. The voucher number, for example, VOUCHER_001, is used to identify which voucher it is; the attribute identifier, for example, amount, is used to identify which specific attribute needs to be reviewed.

[0053] Furthermore, in this embodiment, after step S204, the method further includes: determining whether the field to be reviewed is configured with a manual correction value; if the manual correction value is not empty, then using the manual correction value as the target field value; if the manual correction value is empty, then searching for the attribute list corresponding to the voucher number from the first-level cache of the preset database.

[0054] Step S205: Search for the attribute name corresponding to the credential number and the attribute identifier from the second-level cache of the preset database.

[0055] Understandably, the voucher number and attribute identifier can be concatenated into a key, which can then be used to look up the attribute name, such as the amount, in the second-level cache of the pre-defined database.

[0056] Step S206: If the target attribute corresponding to the attribute name does not exist, then search for the attribute list corresponding to the credential number from the first-level cache of the preset database.

[0057] It should be understood that if the target attribute corresponding to the attribute name does not exist, it means that there is no record for this in the second-level cache. The reasons may be: the cache is not loaded, the key is misspelled, or the voucher does not actually have this attribute. In this case, the attribute list corresponding to the voucher number can be searched from the first-level cache in the preset database. For example, voucherNo = "VOUCHER_001" can be used to search for the corresponding attribute list in the first-level cache.

[0058] Step S207: Filter the target attributes that match the attribute identifier from the attribute list, and determine the target field corresponding to the target attribute.

[0059] Understandably, you can filter the attribute list to match the attribute identifier, for example, {propertyId: "amount", propertyName: "amount", fieldValue: "1000"}. Then, you can determine the target field corresponding to the target attribute; the target field can be the entire attribute object or a specific field value.

[0060] In practical implementation, if the target attribute corresponding to the attribute name exists, the target field corresponding to that target attribute can be directly determined.

[0061] Step S208: Input the target field value in the target field into the preset rule engine for review, and obtain the second review result.

[0062] It should be understood that the target field value can be extracted from the target field, for example, fieldValue = "1000". The target field value is then input into the preset rule engine for review. For example, input: fieldValue = "1000", rule example: IF Amount > 500 THEN Risk Level = High, judgment result: 1000 > 500 is true, output: Second review result = {Risk Level: High, Review Comment: "Amount Exceeds Limit"}.

[0063] Understandably, the preset database in this embodiment includes a two-layer cache: the first layer cache is a voucher attribute list cache, key: voucherNo (voucher number), value: List <voucherproperty>(A list of all attributes for this voucher type); Second-level cache: Attribute name cache, key: voucherNo + "_" + propertyId (voucher number + attribute ID), value: propertyName (attribute name). Population process: 1. Collect the set of all voucherNos involved in all rules and query the database in batches (one query replaces N queries); 2. Populate the query results into the first-level cache; 3. Traverse the rule list, retrieve field values ​​from the cache, prioritizing manualValue (manually corrected value), followed by fieldValue (system-extracted value); 4. Use ConcurrentHashMap to implement caching, ensuring thread safety during concurrent population operations.

[0064] This embodiment determines the fields to be reviewed in the enhanced content, including the credential number and attribute identifier. It then searches the second-level cache of the preset database for the attribute name corresponding to the credential number and attribute identifier. If the target attribute does not exist, it searches the first-level cache of the preset database for the attribute list corresponding to the credential number. It then filters the attribute list for target attributes matching the attribute identifier and determines the target field corresponding to the target attribute. Finally, it inputs the target field value into the preset rule engine for review, obtaining a second review result. This embodiment reduces the number of database queries for filling rule field values ​​from N to 1-3 using a two-level cache structure: credential number → attribute list and credential number + attribute ID → attribute name, thus eliminating the N+1 query problem.

[0065] refer to Figure 3 , Figure 3 This is a flowchart illustrating the third embodiment of the document review method for this application.

[0066] Based on the above embodiments, in this embodiment, step S40 includes: Step S401: Determine the block priority of the block review results corresponding to each block content.

[0067] Understandably, the priority of the review results for each block of content can be determined, with the priority order being: first priority risk (RISK) > second priority non-conformance (NON_CONFORMANCE) > third priority missing (MISSING) > fourth priority pass (PASS).

[0068] Step S402: Select the target priority from the block priorities.

[0069] It should be understood that a target priority can be selected from the block priorities, and the target priority is the highest priority. For example, the block priorities include the first priority and the second priority, and the target priority is the first priority.

[0070] Step S403: Determine the document review result corresponding to the document to be reviewed based on the target review result corresponding to the target priority.

[0071] Understandably, the target review result corresponding to the target priority can be used as the document review result corresponding to the document to be reviewed. The final conclusion = MAX(Block 1 conclusion, Block 2 conclusion, ..., Block N conclusion). For example, Block 1 conclusion: Pass; Block 2 conclusion: Incompatible; Block 3 conclusion: Pass; Final conclusion: Incompatible (take the strictest one).

[0072] It should be understood that the analysis text of each block is concatenated and labeled with the source block number: [Block 1 Analysis] Article 3 Payment Terms Exist...; [Block 2 Analysis] Article 7 Breach of Contract Liability Agreement.... Clause deduplication: Deduplication is performed based on the combined hash value of "clause location + first 100 characters of the original text" to avoid the same clause being reported repeatedly in multiple blocks, ensuring zero risk omissions and non-redundant results.

[0073] In the specific implementation, refer to Figure 4 , Figure 4 This is a schematic diagram of the overall process of one embodiment of the document review method of this application, as shown below. Figure 4 As shown, the document review service entry point allows for concurrent task submission, parallel AI model review (dedicated thread pool) and rule engine review (dedicated thread pool), block review executor, enhanced context window, strict and limited principle for merging block results, rule field value binder, and double-layer caching + concurrency security.

[0074] This embodiment determines the block priority of the block review results corresponding to each block of content, then selects a target priority from the block priorities, and finally determines the document review result corresponding to the document to be reviewed based on the target review result corresponding to the target priority. This embodiment determines the document review result corresponding to the document to be reviewed based on the target review result corresponding to the target priority, which can take the most stringent value for the block review results of all block content, thereby improving the accuracy of the document review results.

[0075] Reference Figure 5 , Figure 5 This is a structural block diagram of the first embodiment of the document review device of this application.

[0076] like Figure 5 As shown, the document review device proposed in this application includes: Document segmentation module 10 is used to segment the document to be reviewed into multiple segments. The content review module 20 is used to input each block of content into the preset review model and the preset rule engine for review, and obtain the first review result and the second review result. The result determination module 30 is used to determine the sub-block review result corresponding to each sub-block content based on the first review result and / or the second review result. The result merging module 40 is used to merge the block review results corresponding to each block content to obtain the document review result corresponding to the document to be reviewed.

[0077] This embodiment divides the document to be reviewed into multiple blocks, then inputs each block into a preset review model and a preset rule engine for review, obtaining a first review result and a second review result. Based on the first and / or second review results, the block review result corresponding to each block is determined. Finally, the block review results corresponding to each block are merged to obtain the final document review result for the entire document to be reviewed. This embodiment first divides the document to be reviewed into blocks, then inputs each block into a preset review model and a preset rule engine for parallel review, obtaining a first and second review result for each block. Finally, the block review results are merged to obtain the final document review result for the entire document to be reviewed, thereby reducing the total review time and improving the efficiency of document review.

[0078] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of this application. In practical applications, those skilled in the art can select some or all of it to achieve the purpose of this embodiment according to actual needs, and no restrictions are imposed here.

[0079] In addition, for technical details not described in detail in this embodiment, please refer to the document review method provided in any embodiment of this application, which will not be repeated here.

[0080] Based on the first embodiment of the document review device described in this application, a second embodiment of the document review device of this application is proposed.

[0081] In this embodiment, the content review module 20 is further configured to determine the document position of each block of content in the document to be reviewed; determine the context content corresponding to each block of content based on the document position, and construct enhanced content according to the block of content and the context content; input the enhanced content into a preset review model for review, and obtain a first review result.

[0082] Furthermore, the content review module 20 is also used to determine the fields to be reviewed corresponding to the content to be reviewed in the enhanced content, the fields to be reviewed including credential number and attribute identifier; search for the attribute name corresponding to the credential number and the attribute identifier from the second-level cache of the preset database; if the target attribute corresponding to the attribute name does not exist, search for the attribute list corresponding to the credential number from the first-level cache of the preset database; filter the target attributes that match the attribute identifier from the attribute list, and determine the target field corresponding to the target attribute; input the target field value in the target field into the preset rule engine for review, and obtain the second review result.

[0083] Furthermore, the content review module 20 is also used to determine whether the field to be reviewed is configured with a manual correction value; if the manual correction value is not empty, then the manual correction value is used as the target field value; if the manual correction value is empty, then the attribute list corresponding to the voucher number is searched from the first-level cache of the preset database.

[0084] Furthermore, the result merging module 40 is also used to determine the block priority of the block review result corresponding to each block content; select a target priority from the block priorities; and determine the document review result corresponding to the document to be reviewed based on the target review result corresponding to the target priority.

[0085] Furthermore, the document segmentation module 10 is also used to determine the document length corresponding to the document to be reviewed; if the document length is less than the preset length, the document to be reviewed is input into the preset review model and the preset rule engine for review, and the document review result corresponding to the document to be reviewed is determined according to the obtained third review result and fourth review result; if the document length is greater than or equal to the preset length, the document to be reviewed is segmented.

[0086] Furthermore, the result merging module 40 is also used to determine whether the first review result and the second review result are consistent; if not, the document to be reviewed is pushed to the manual review queue.

[0087] Other embodiments or specific implementations of the document review device in this application can be found in the above-described method embodiments, and will not be repeated here.

[0088] This application provides a document review device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform the document review method in Embodiment 1 above.

[0089] The following is for reference. Figure 6 The diagram illustrates a structural schematic of a document review device suitable for implementing embodiments of this application. The document review device in these embodiments may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 6 The document review device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0090] like Figure 6 As shown, the document review device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the document review device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. Communication device 1009 allows the document review device to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows document review devices with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0091] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0092] The document review device provided in this application, employing the document review method described in the above embodiments, can solve the technical problem of how to reduce the total time spent on document review and improve the efficiency of document review. Compared with the prior art, the beneficial effects of the document review device provided in this application are the same as those of the document review method provided in the above embodiments, and other technical features of this document review device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0093] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0094] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0095] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the document review method in the above embodiments.

[0096] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0097] The aforementioned computer-readable storage medium may be included in the document review device; or it may exist independently and not be assembled into the document review device.

[0098] The aforementioned computer-readable storage medium carries one or more programs. When the document review device executes the aforementioned one or more programs, the document review device causes the document review device to: divide the document to be reviewed into blocks to obtain multiple blocks of content; input each block of content into a preset review model and a preset rule engine for review to obtain a first review result and a second review result; determine the block review result corresponding to each block of content based on the first review result and / or the second review result; and merge the block review results corresponding to each block of content to obtain the document review result corresponding to the document to be reviewed.

[0099] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Python, Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0100] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0101] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0102] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., computer programs) for executing the above-described document review method. This addresses the technical problem of reducing the total time required for document review and improving its efficiency. Compared to the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the document review method provided in the above embodiments, and will not be elaborated upon here.

[0103] The above description is only a part of the embodiments of this application and does not limit the scope of protection of this application. All equivalent structural transformations made under the technical concept of this application and using the content of this application specification and drawings, or direct / indirect applications in other related technical fields, are included in the scope of protection of this application.< / voucherproperty> < / audititem> < / audititem> < / reviewresponse>

Claims

1. A document review method, characterized in that, The method includes the following steps: The document to be reviewed is divided into blocks, resulting in multiple blocks of content. Each block of content is input into the preset review model and the preset rule engine for review, resulting in the first review result and the second review result. Determine the sub-block review result corresponding to each sub-block content based on the first review result and / or the second review result; The review results of each block are merged to obtain the document review result corresponding to the document to be reviewed.

2. The document review method as described in claim 1, characterized in that, The step of inputting each segment of content into a preset review model for review to obtain a first review result includes: Determine the document location of each section's content within the document to be reviewed; Based on the document location, determine the context content corresponding to each block of content, and construct enhanced content based on the block of content and the context content; The enhanced content is input into a preset review model for review, and a first review result is obtained.

3. The document review method as described in claim 2, characterized in that, The step of inputting the enhanced content into a preset rule engine for review to obtain a second review result includes: Determine the fields to be reviewed corresponding to the content to be reviewed in the enhanced content, wherein the fields to be reviewed include voucher number and attribute identifier; Search the second-level cache of the preset database for the attribute name corresponding to the credential number and the attribute identifier; If the target attribute corresponding to the attribute name does not exist, the attribute list corresponding to the voucher number is searched from the first-level cache of the preset database; Filter the target attributes that match the attribute identifier from the attribute list, and determine the target field corresponding to the target attribute; The target field value is input into the preset rule engine for review, and a second review result is obtained.

4. The document review method as described in claim 3, characterized in that, After determining the field to be reviewed corresponding to the content to be reviewed in the enhanced content, the method further includes: Determine whether the field to be reviewed is configured with a manual correction value; If the manually corrected value is not empty, then the manually corrected value is used as the target field value; If the manually corrected value is empty, then the attribute list corresponding to the voucher number is retrieved from the first-level cache of the preset database.

5. The document review method as described in claim 1, characterized in that, The step of merging the block review results corresponding to each block of content to obtain the document review result corresponding to the document to be reviewed includes: Determine the block priority of the block review results corresponding to each block content; Select the target priority from the block priorities; The document review result corresponding to the document to be reviewed is determined based on the target review result corresponding to the target priority.

6. The document review method as described in any one of claims 1 to 5, characterized in that, Before dividing the document to be reviewed into multiple blocks, the process also includes: Determine the document length corresponding to the document to be reviewed; If the document length is less than the preset length, the document to be reviewed is input into the preset review model and the preset rule engine for review, and the document review result corresponding to the document to be reviewed is determined according to the obtained third review result and fourth review result. If the document length is greater than or equal to the preset length, the document to be reviewed will be divided into blocks.

7. The document review method as described in any one of claims 1 to 5, characterized in that, After merging the block review results corresponding to each block of content to obtain the document review result corresponding to the document to be reviewed, the process further includes: Determine whether the first review result and the second review result are consistent; If not, the document to be reviewed will be pushed to the manual review queue.

8. A document review device, characterized in that, The document review device includes: The document segmentation module is used to segment documents to be reviewed into multiple segments. The content review module is used to input each block of content into a preset review model and a preset rule engine for review, and obtain the first review result and the second review result. The result determination module is used to determine the sub-block review result corresponding to each sub-block content based on the first review result and / or the second review result. The result merging module is used to merge the block review results corresponding to each block of content to obtain the document review result corresponding to the document to be reviewed.

9. A document review device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the document review method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the document review method as described in any one of claims 1 to 7.

Images