Intelligent document auditing method and device, electronic equipment and storage medium

Abstract

The invention provides an intelligent document auditing method and device, electronic equipment and a storage medium. The method comprises the following steps: firstly, determining an auditing type of a to-be-audited document; and if the to-be-audited document is the document which is audited for the first time, generating a document abstract of the to-be-audited document based on the document type of the to-be-audited document. If the to-be-audited document is not the document which is audited for the first time, obtaining a historical document which is submitted for the last time of the to-be-audited document; and calculating the similarity between the to-be-audited document and the historical document, processing the to-be-audited document according to the similarity, and generating a document abstract of the to-be-audited document or generating a comparison file of the to-be-audited document and the historical document. Therefore, by means of the method, the document abstract of the to-be-audited document or the comparison file of the to-be-audited document and the historical document can be automatically generated according to the type of the to-be-audited document, reference is provided for auditing personnel, the auditing personnel are assisted in conducting document auditing, and the auditing efficiency can be effectively improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, device, electronic device and storage medium for intelligent document review. Background Art

[0002] In the modern financial environment, the banking industry faces increasingly complex economic environments and regulatory requirements. To ensure operational stability, banks have significantly increased their attention to risk management and procedural norms. In actual operations, the bank's approval process requires the collaboration of multiple departments, and each step requires detailed review and confirmation to prevent potential risks.

[0003] However, when the audit document has a large amount of content, the auditor needs to spend a lot of time reading a large amount of document content, or when the audit document is modified, the auditor needs to re-read the document content to find the differences, and the audit efficiency is low. Summary of the invention

[0004] In view of this, the present application provides a method, device, electronic device and storage medium for intelligent document review to solve the problem of low review efficiency in the prior art.

[0005] To achieve the above objectives, this application provides the following technical solutions:

[0006] The first aspect of the present application discloses a method for intelligent document review, comprising:

[0007] Determine the audit type of the document to be audited; wherein the audit type includes initial audit and non-initial audit;

[0008] If the document to be reviewed is a document to be reviewed for the first time, a document summary of the document to be reviewed is generated based on the document type of the document to be reviewed; wherein the document type includes a fixed template class and a supplementary information class;

[0009] If the document to be reviewed is not a document to be reviewed for the first time, then obtaining the most recent historical document submitted for the document to be reviewed;

[0010] The similarity between the document to be reviewed and the historical document is calculated, and the document to be reviewed is processed according to the similarity to generate a document summary of the document to be reviewed or a comparison file of the document to be reviewed and the historical document.

[0011] Optionally, in the above method, generating a document summary of the document to be reviewed based on the document type of the document to be reviewed includes:

[0012] If the document to be reviewed is of the fixed template type, then obtaining the template corresponding to the document to be reviewed;

[0013] Extracting key words and phrases from the document to be reviewed, and filling the key words and phrases into the template to generate a document summary of the document to be reviewed;

[0014] If the document to be reviewed is of the supplementary information type, the document to be reviewed is segmented into sentences using a preset sorting algorithm, and a document summary of the document to be reviewed is generated.

[0015] Optionally, in the above method, the step of processing the document to be reviewed according to the similarity to generate a document summary of the document to be reviewed or generating a comparison file between the document to be reviewed and the historical document includes:

[0016] If the similarity is greater than a preset threshold, generating a document summary of the document to be reviewed based on the document type of the document to be reviewed;

[0017] If the similarity is not greater than the threshold, the longest common subsequence algorithm is used to find out the difference between the document to be reviewed and the historical document, and generate the comparison file.

[0018] Optionally, the above method, before determining the audit type of the document to be audited, further includes:

[0019] Perform text preprocessing on the document to be reviewed.

[0020] The second aspect of the present application discloses a device for intelligent document review, comprising:

[0021] A determination unit, used to determine the audit type of the document to be audited; wherein the audit type includes initial audit and non-initial audit;

[0022] A first generating unit is used to generate a document summary of the document to be reviewed based on the document type of the document to be reviewed if the document to be reviewed is a document to be reviewed for the first time; wherein the document type includes a fixed template class and a supplementary information class;

[0023] An acquisition unit, configured to acquire a historical document of the most recent submission of the document to be reviewed if the document to be reviewed is not a document to be reviewed for the first time;

[0024] The second generating unit is used to calculate the similarity between the document to be reviewed and the historical document, and process the document to be reviewed according to the similarity to generate a document summary of the document to be reviewed or generate a comparison file between the document to be reviewed and the historical document.

[0025] Optionally, in the above device, the first generating unit includes:

[0026] An acquisition subunit, used for acquiring the template corresponding to the document to be reviewed if the document to be reviewed is of the fixed template type;

[0027] A first generating subunit is used to extract key words and phrases from the document to be reviewed, and fill the key words and phrases into the template to generate a document summary of the document to be reviewed;

[0028] The second generating subunit is used for, if the document to be reviewed is of the supplementary information type, performing sentence segmentation on the document to be reviewed by using a preset sorting algorithm, and generating a document summary of the document to be reviewed.

[0029] Optionally, in the above device, the second generating unit includes:

[0030] A third generating subunit, configured to generate a document summary of the document to be reviewed based on the document type of the document to be reviewed if the similarity is greater than a preset threshold;

[0031] The fourth generating subunit is used for finding out the difference between the document to be reviewed and the historical document by using the longest common subsequence algorithm if the similarity is not greater than the threshold value, and generating the comparison file.

[0032] Optionally, the above device further includes:

[0033] The preprocessing unit is used to perform text preprocessing on the document to be reviewed.

[0034] The third aspect of the present application discloses an electronic device, comprising:

[0035] one or more processors;

[0036] a storage device having one or more programs stored thereon;

[0037] When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of the first aspects of the present invention.

[0038] A fourth aspect of the present application discloses a computer storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the method described in any one of the first aspects of the present invention is implemented.

[0039] It can be seen from the above technical scheme that in a method for intelligent document review provided by the present application, the review type of the document to be reviewed is first determined. If the document to be reviewed is a document that is being reviewed for the first time, a document summary of the document to be reviewed is generated based on the document type of the document to be reviewed. If the document to be reviewed is not a document that is being reviewed for the first time, the historical document that was most recently submitted for the document to be reviewed is obtained; the similarity between the document to be reviewed and the historical document is calculated, and the document to be reviewed is processed according to the similarity to generate a document summary of the document to be reviewed or a comparison file of the document to be reviewed and the historical document. It can be seen that by using the method of the present application, a document summary of the document to be reviewed or a comparison file of the document to be reviewed and the historical document can be automatically generated according to the type of the document to be reviewed, providing a reference for the reviewer, assisting the reviewer in document review, and effectively improving the review efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0041] Figure 1 A flowchart of a method for intelligent document review disclosed in an embodiment of the present application;

[0042] Figure 2 A schematic diagram of a process audit system disclosed in an embodiment of the present application;

[0043] Figure 3 A schematic diagram of a device for intelligent document review disclosed in an embodiment of the present application;

[0044] Figure 4 A schematic diagram of an electronic device disclosed in an embodiment of the present application. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0046] In this application, the terms "comprises", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element.

[0047] Furthermore, in this document, relational terms such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0048] The present application embodiment provides a method for intelligent document review, such as Figure 1 As shown, specifically including:

[0049] S101. Determine the audit type of the document to be audited; wherein the audit type includes initial audit and non-initial audit.

[0050] It should be noted that when conducting a document review, the review type of the document to be reviewed must be determined first; the review types include initial review and non-initial review. Initial review refers to the first time the document is submitted and has not been modified. Non-initial review refers to documents that have been submitted, but have been sent back for modification after the work order has been circulated, and need to be reviewed again.

[0051] Optionally, in another embodiment of the present application, before executing step S101, the following may also be included:

[0052] Perform text preprocessing on the documents to be reviewed.

[0053] It should be noted that the document to be reviewed is divided into sentence sets according to complete sentences. A custom user dictionary is established, and uncommon professional words are added to the existing common dictionary. For each sentence, the word is segmented according to the dictionary, and the part of speech, such as adjective, noun, verb, etc., is marked, and unnecessary punctuation marks and function words are removed to obtain the sentence and word segmentation results of the document.

[0054] S102: If the document to be reviewed is a document to be reviewed for the first time, a document summary of the document to be reviewed is generated based on the document type of the document to be reviewed; wherein the document type includes a fixed template class and a supplementary information class.

[0055] It should be noted that if the document to be reviewed is a document for the first review, the document type of the document to be reviewed is identified, wherein the document type includes a fixed template class and a supplementary information class, the fixed template class is a document generated according to a fixed template, and the supplementary information class is a document not generated according to a fixed template. Then, based on the document type of the document to be reviewed, different processing methods are used to process the document to be reviewed, and a document summary of the document to be reviewed is generated.

[0056] Optionally, in another embodiment of the present application, an implementation of step S102 may include:

[0057] If the document to be reviewed is a fixed template class, obtain the template corresponding to the document to be reviewed.

[0058] Extract key words and phrases from the document to be reviewed, and fill the key words and phrases into the template to generate a document summary of the document to be reviewed.

[0059] If the document to be reviewed is of the supplementary information type, the document to be reviewed is segmented into sentences using a preset sorting algorithm, and a document summary of the document to be reviewed is generated.

[0060] It should be noted that if the document to be reviewed is a fixed template type, the matching template specified in the document title is used to find the existing template in the system, and the corresponding fields in the document are automatically identified according to the preset rules and the key words and phrases in the document are extracted. This information is filled into the template summary to generate a structured summary, reducing the manual workload of the reviewer. For example, in the online system, the work order submitted by the bill lading person must include a business test report document to prove that the content to be put online has been verified. The business test report has a fixed template, including fields such as project name, development department, test department, test content, test date, and auditor's signature. After the bill lading person uploads the document to the system, the system first identifies the document name, matches it to the business test report template in the system, and pre-processes the submitted document, that is, removes interference elements such as font color, type, paragraph format, extra blank lines, indentation, etc., to ensure that the text content in the document can be correctly parsed. Then, according to the pre-defined template structure, the document is scanned field by field to match the key fields to be filled in the fixed template summary. The system extracts the matched information and organizes it according to the structure in the template. Based on the extracted information, the system generates a concise template summary for the reviewer: This test was conducted on multiple contents such as n1 and n2 of Project X, which was led by Department A and developed by Department B, between Date C and Date D. The tester was E and the reviewer was F. The test was completed without any exceptions and all were signed.

[0061] If the document to be reviewed is of the supplementary information type, the TextRank algorithm is used to segment the document to be reviewed and generate a document summary of the document to be reviewed. TextRank is a graph-based sorting algorithm, mainly used for text processing tasks. It divides the text into several components (such as words and sentences), builds edges based on the similarity between units, and establishes a graph model. It uses a voting mechanism to sort the important components in the text, so as to achieve keyword extraction and summary.

[0062] For example, in the stamping process, the bill of lading holder submits materials that need to be stamped, and the materials do not belong to a fixed template. In this case, the method extracts the material summary according to the following algorithm process to help reviewers quickly understand the main content of the materials, thereby improving review efficiency.

[0063] Take each sentence as a node and calculate the similarity between the two sentences by the number of words that appear in both sentences. The calculation formula is:

[0064]

[0065] Where s represents a sentence and w represents a word, and a similarity matrix between sentences is constructed.

[0066] Based on the similarity matrix, the sentence graph model is obtained, and the weight of each node is updated iteratively. The weight is calculated based on the similarity. The weight calculation formula for each node is:

[0067]

[0068] w represents weight, v represents node, and d is coefficient. That is, the weight of each node is calculated by the weight of its incoming edge nodes. In each round of iteration, the weight of each node is redistributed according to the weight of its neighboring nodes until convergence. The weight of each node reflects its importance in the text. After convergence, the high-ranking sentences are extracted according to the node weight value and combined into a text summary.

[0069] S103: If the document to be reviewed is not a document to be reviewed for the first time, obtain the most recent historical document submitted for the document to be reviewed.

[0070] It should be noted that if the document to be reviewed is not a document for the first review, then the most recently submitted historical document of the document to be reviewed is obtained, that is, the historical document is the original document, and the document to be reviewed is a document modified based on the historical document.

[0071] S104: Calculate the similarity between the document to be reviewed and the historical documents, and process the document to be reviewed according to the similarity to generate a document summary of the document to be reviewed or produce a comparison file between the document to be reviewed and the historical documents.

[0072] It should be noted that the feature vector of the document to be reviewed and the feature vector of the historical document are extracted, and the similarity between the document to be reviewed and the historical document is calculated using cosine similarity. The document to be reviewed is processed according to the similarity to determine whether to generate a document summary of the document to be reviewed or to generate a comparison file of the document to be reviewed and the historical document.

[0073] Optionally, in another embodiment of the present application, an implementation of the above step S104 may include:

[0074] If the similarity is greater than a preset threshold, a document summary of the document to be reviewed is generated based on the document type of the document to be reviewed;

[0075] If the similarity is not greater than the threshold, the longest common subsequence algorithm is used to find the differences between the document to be reviewed and the historical document, and a comparison file is generated.

[0076] It should be noted that if the similarity is greater than a preset threshold, a document summary of the document to be reviewed is generated based on the document type of the document to be reviewed, wherein the threshold can be set according to actual conditions, for example, 70%. The implementation process of generating a document summary of the document to be reviewed based on the document type of the document to be reviewed can be referred to the content of the above embodiment, and will not be repeated here.

[0077] If the similarity is not greater than the threshold, the longest common subsequence algorithm (LCS algorithm) is used to compare the content between the document to be reviewed and the historical document. First, each line of the new and old files is compared, and the LCS algorithm identifies and marks the differences. These differences are then integrated and output as a comparison file, listing the new, modified, and deleted content in detail. This method not only simplifies the reviewer's search process for modified content, but also helps them compare differences more efficiently and improve review efficiency.

[0078] In a method for intelligent document review provided by an embodiment of the present application, the review type of the document to be reviewed is first determined. If the document to be reviewed is a document that is being reviewed for the first time, a document summary of the document to be reviewed is generated based on the document type of the document to be reviewed. If the document to be reviewed is a document that is not being reviewed for the first time, the historical document that was most recently submitted for the document to be reviewed is obtained. The similarity between the document to be reviewed and the historical document is calculated, and the document to be reviewed is processed based on the similarity to generate a document summary of the document to be reviewed or a comparison file of the document to be reviewed and the historical document. It can be seen that by using the method of the present application, a document summary of the document to be reviewed or a comparison file of the document to be reviewed and the historical document can be automatically generated based on the type of the document to be reviewed, providing a reference for the reviewer, assisting the reviewer in document review, and effectively improving the review efficiency.

[0079] In another embodiment of the present application, a process audit system based on the document intelligent audit method is provided, such as Figure 2 As shown, it mainly includes three links: bill of lading by the initiator, system circulation and review by the reviewer.

[0080] 1. Originator's bill of lading

[0081] This link is the starting point of the entire system. The initiator initiates the business process, enters the necessary business information on the bill of lading interface, uploads relevant documents, and fills out a detailed application form.

[0082] The form interface includes four parts: filling instructions module, filling module, attachment upload module and submission module. Among them, the filling instructions module guides the bill of lading person on how to fill in the form; the filling module can be divided into multiple areas, and the attachments can be checked after uploading. According to the different review objects of multiple reviewers in the next link, in the submission module, the bill of lading person submits to different reviewers through the submit button. When submitting a bill of lading, the system performs form verification, file upload and data storage. The system performs preliminary verification of the data entered by the user, automatically detects and marks common errors and incomplete information, provides modification suggestions, ensures the integrity and accuracy of the data, and thus reduces the probability of rejection. Once submitted successfully, the bill of lading information will enter the system and generate a unique bill of lading number for subsequent tracking and management.

[0083] When submitting a bill of lading, the bill of lading holder can choose to submit it to multiple reviewers for parallel processing. At this time, each reviewer will make a circulation opinion on the bill of lading. If there is no objection and agreement, it will be submitted to the reviewer of the next link according to the work order circulation rules. If a reviewer rejects the content of this bill of lading, the bill of lading holder can choose to retrieve other parts of the form module to be reviewed and modify them together, or use the replacement function to modify the content in the review part reviewed by the reviewer, replace the attachment, and resubmit it without affecting the review of other reviewers. If the bill of lading holder needs additional explanations after the reviewer has completed the review, the work order can be transferred to the bill of lading holder. On the basis of not modifying the submitted part, the bill of lading can be directly transferred to the reviewer after additional explanations are required by the current reviewer. The previous reviewer will receive a reminder to supplement the work order, but is not forced to review it again.

[0084] 2. System Flow

[0085] This link is responsible for circulating and distributing bill of lading information in the system. It consists of four functional modules: task assignment, status recording, reminder notification, and visual tracking. According to the preset business processes and rules, the system automatically distributes bill of lading information to the corresponding processing departments and personnel. To ensure the transparency of the processing progress and status of each link, the reviewer will be intelligently reminded to review during the circulation, the current work order status will be recorded in the log, and a visual chart of the work order circulation will be provided to ensure that the processing process of each bill of lading is traceable and support subsequent audits and analysis.

[0086] Task allocation: The task allocation module is responsible for automatically or manually allocating newly submitted work orders to appropriate processing personnel or departments according to preset rules and conditions. For different types of work orders, the system pre-sets the next review group that can be selected in this link, and circulates the work order according to the selected submission object.

[0087] Status Record: The status record module tracks and records the status changes of work orders in each processing link in real time, including pending, in-process, completed, etc. Each status change will be recorded in detail, including the change time, operator and specific operation content, forming a complete log record to ensure the transparency and traceability of the work order processing process.

[0088] Smart reminders: The smart reminder module sends reminders and notifications to relevant personnel through multiple channels to ensure that work orders can be processed in a timely manner without frequent interruptions due to excessive reminders. The module designs a multi-level reminder strategy, including light reminders (system pop-ups, in-app notifications), medium reminders (email notifications) and heavy reminders (SMS, phone notifications). According to the business type and urgency of each work order, different priority levels are assigned to the work order, and the reviewer of the current link is automatically reminded to different degrees. At the same time, it supports the processing personnel to manually adjust the reminder settings to enhance the flexibility and adaptability of the module.

[0089] Visual tracking: used to monitor and manage the work order flow process in real time. This function uses charts and graphical interfaces to display the status and flow of work orders at various process nodes, helping managers and lading personnel to better understand and track the progress of work order processing. In the visual chart, color coding (such as green for completed, yellow for processing, and red for pending) can be used to intuitively display the current status of the work order in the process in real time. The flow chart vividly displays the flow path of the work order. Each node represents a processing link. The arrows between the nodes indicate the flow direction of the work order. The gray area shows the pending flow link of the process, allowing the lading personnel to better grasp the progress of the bill of lading.

[0090] (III) Review by Reviewer

[0091] This link is the key link of the audit system, responsible for reviewing and approving the work orders in circulation. This link usually includes functions such as approval decision, audit opinion record, and audit task allocation. The reviewer views the detailed information and circulation process records of the bill of lading in the system, fills in the review opinion and makes a review decision, and then submits the work order to the reviewer or bill of lading person in the next link. This link ensures that each bill of lading is fully reviewed and feedback is received, and all review processes and results are recorded for subsequent query and tracking. In order to help reviewers quickly and accurately understand the content of the work order and complete the review, the review process is based on the document intelligent review method proposed in this patent.

[0092] During the initial review of the work order, all submitted documents will be processed comprehensively. Submitted documents are divided into two categories: fixed template category and supplementary information category. If the submitted document is generated by an existing template, the preset rules will be used to pre-review the filled content and generate a fixed format summary. When submitting, the review results will be placed next to the document for reviewers to view. For supplementary information documents, a summary description of the document is output based on the TextRank algorithm.

[0093] After a work order is submitted repeatedly, a comprehensive comparison will be made between the submitted document and the previously submitted version. For a rejected work order, if the submitter needs to modify the attached document, click "Replace" to replace the old document submitted last time with the modified new document. The reviewer can click "Compare" to obtain a document showing the differences between the two documents, which not only allows the submitter to more quickly understand the modification location and content of the modification, but also avoids ignoring small modification points.

[0094] When reviewers face submitted work orders, they will have three review results: rejection, approval, and pending. If the reviewer believes that the work order is incomplete, the information is wrong, it does not meet the regulations or standards, or there are obvious risks or problems, the work order will be rejected and returned to the submitter for modification or supplementation. The previous processes involved in the modification by the submitter need to be reviewed again. If the reviewer believes that the work order is complete, the information is accurate, it meets the regulations and standards, and no risks or problems are found, the work order will enter the next processing link after being reviewed and the entire process is finally completed. The pending result is usually caused by the following situations: there is unclear or vague information in the work order, and it is impossible to make a decision immediately, and further verification or additional information is required. If a pending result is given, the reviewer writes the comments and matters to be discussed, and signs the work order to other co-reviewers. After other reviewers give their opinions, the work order returns to the review link of the reviewer to give the final result of approval or rejection.

[0095] Another embodiment of the present application also discloses a device for intelligent document review, such as Figure 3 As shown, specifically including:

[0096] The determination unit 301 is used to determine the audit type of the document to be audited; wherein the audit type includes initial audit and non-initial audit.

[0097] The first generating unit 302 is used to generate a document summary of the document to be reviewed based on the document type of the document to be reviewed if the document to be reviewed is a document to be reviewed for the first time; wherein the document type includes a fixed template class and a supplementary information class.

[0098] The acquisition unit 303 is used to acquire the most recently submitted historical document of the document to be reviewed if the document to be reviewed is not a document to be reviewed for the first time.

[0099] The second generating unit 304 is used to calculate the similarity between the document to be reviewed and the historical document, and process the document to be reviewed according to the similarity to generate a document summary of the document to be reviewed or generate a comparison file between the document to be reviewed and the historical document.

[0100] In this embodiment, the specific execution process of the determination unit 301, the first generation unit 302, the acquisition unit 303 and the second generation unit 304 can be referred to in the corresponding Figure 1 The content of the method embodiment will not be repeated here.

[0101] In a device for intelligent document review provided by an embodiment of the present application, first, a determination unit 301 determines the review type of the document to be reviewed. If the document to be reviewed is a document that is being reviewed for the first time, a first generation unit 302 generates a document summary of the document to be reviewed based on the document type of the document to be reviewed. If the document to be reviewed is a document that is not being reviewed for the first time, an acquisition unit 303 acquires the historical document that was most recently submitted by the document to be reviewed. A second generation unit 304 calculates the similarity between the document to be reviewed and the historical document, and processes the document to be reviewed based on the similarity, generating a document summary of the document to be reviewed or generating a comparison file of the document to be reviewed and the historical document. It can be seen from this that, using the method of the present application, a document summary of the document to be reviewed or a comparison file of the document to be reviewed and the historical document can be automatically generated based on the type of the document to be reviewed, providing a reference for the reviewer, assisting the reviewer in conducting document review, and effectively improving the review efficiency.

[0102] Optionally, in another embodiment of the present application, an implementation of the first generating unit 302 includes:

[0103] The acquisition subunit is used to obtain the template corresponding to the document to be reviewed if the document to be reviewed is a fixed template class.

[0104] The first generating subunit is used to extract key words and sentences in the document to be reviewed, and fill the key words and sentences into the template to generate a document summary of the document to be reviewed.

[0105] The second generating subunit is used for, if the document to be reviewed is of the supplementary information type, performing sentence segmentation on the document to be reviewed by using a preset sorting algorithm, and generating a document summary of the document to be reviewed.

[0106] In this embodiment, the specific execution process of the acquisition subunit, the first generation subunit and the second generation subunit can be found in the corresponding method embodiment content mentioned above, which will not be repeated here.

[0107] Optionally, in another embodiment of the present application, an implementation of the second generating unit 304 includes:

[0108] The third generating subunit is configured to generate a document summary of the document to be reviewed based on the document type of the document to be reviewed if the similarity is greater than a preset threshold.

[0109] The fourth generating subunit is used for finding the difference between the document to be reviewed and the historical document by using the longest common subsequence algorithm if the similarity is not greater than the threshold value, and generating a comparison file.

[0110] In this embodiment, the specific execution process of the third generating subunit and the fourth generating subunit can be found in the corresponding method embodiment described above, and will not be repeated here.

[0111] Optionally, in another embodiment of the present application, the above-mentioned document intelligent review device may further include:

[0112] The preprocessing unit is used to perform text preprocessing on the document to be reviewed.

[0113] In this embodiment, the specific execution process of the preprocessing unit can refer to the corresponding method embodiment content mentioned above, which will not be repeated here.

[0114] Another embodiment of the present application further provides an electronic device, such as Figure 4 As shown, specifically including:

[0115] One or more processors 401 .

[0116] The storage device 402 stores one or more programs.

[0117] When one or more programs are executed by one or more processors 401 , the one or more processors 401 implement any one of the methods in the above embodiments.

[0118] Another embodiment of the present application further provides a computer storage medium on which a computer program is stored, wherein when the computer program is executed by a processor, any one of the methods in the above embodiments is implemented.

[0119] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can refer to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiment. The system and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without creative work.

[0120] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0121] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for intelligent document review, characterized in that: include: Determine the audit type of the document to be audited; wherein the audit type includes initial audit and non-initial audit; If the document to be reviewed is a document to be reviewed for the first time, a document summary of the document to be reviewed is generated based on the document type of the document to be reviewed; wherein the document type includes a fixed template class and a supplementary information class; If the document to be reviewed is not a document to be reviewed for the first time, then obtaining the most recent historical document submitted for the document to be reviewed; The similarity between the document to be reviewed and the historical document is calculated, and the document to be reviewed is processed according to the similarity to generate a document summary of the document to be reviewed or a comparison file of the document to be reviewed and the historical document.

2. The method according to claim 1, characterized in that: The step of generating a document summary of the document to be reviewed based on the document type of the document to be reviewed includes: If the document to be reviewed is of the fixed template type, then obtaining the template corresponding to the document to be reviewed; Extracting key words and phrases from the document to be reviewed, and filling the key words and phrases into the template to generate a document summary of the document to be reviewed; If the document to be reviewed is of the supplementary information type, the document to be reviewed is segmented into sentences using a preset sorting algorithm, and a document summary of the document to be reviewed is generated.

3. The method according to claim 1, characterized in that The step of processing the document to be reviewed according to the similarity to generate a document summary of the document to be reviewed or generating a comparison file between the document to be reviewed and the historical document includes: If the similarity is greater than a preset threshold, generating a document summary of the document to be reviewed based on the document type of the document to be reviewed; If the similarity is not greater than the threshold, the longest common subsequence algorithm is used to find out the difference between the document to be reviewed and the historical document, and generate the comparison file.

4. The method according to claim 1, characterized in that: Before determining the audit type of the document to be audited, the process further includes: Perform text preprocessing on the document to be reviewed.

5. A device for intelligent document review, characterized in that: include: A determination unit, used to determine the audit type of the document to be audited; wherein the audit type includes initial audit and non-initial audit; A first generating unit is used to generate a document summary of the document to be reviewed based on the document type of the document to be reviewed if the document to be reviewed is a document to be reviewed for the first time; wherein the document type includes a fixed template class and a supplementary information class; An acquisition unit, configured to acquire a historical document of the most recent submission of the document to be reviewed if the document to be reviewed is not a document to be reviewed for the first time; The second generating unit is used to calculate the similarity between the document to be reviewed and the historical document, and process the document to be reviewed according to the similarity to generate a document summary of the document to be reviewed or generate a comparison file between the document to be reviewed and the historical document.

6. The device according to claim 5, characterized in that The first generating unit comprises: An acquisition subunit, used for acquiring the template corresponding to the document to be reviewed if the document to be reviewed is of the fixed template class; A first generating subunit is used to extract key words and phrases from the document to be reviewed, and fill the key words and phrases into the template to generate a document summary of the document to be reviewed; The second generating subunit is used for, if the document to be reviewed is of the supplementary information type, performing sentence segmentation on the document to be reviewed by using a preset sorting algorithm, and generating a document summary of the document to be reviewed.

7. The device according to claim 5, characterized in that The second generating unit comprises: A third generating subunit, configured to generate a document summary of the document to be reviewed based on the document type of the document to be reviewed if the similarity is greater than a preset threshold; The fourth generating subunit is used for finding out the difference between the document to be reviewed and the historical document by using the longest common subsequence algorithm if the similarity is not greater than the threshold value, and generating the comparison file.

8. The device according to claim 5, characterized in that Also includes: The preprocessing unit is used to perform text preprocessing on the document to be reviewed.

9. An electronic device, characterized in that: include: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the method according to any one of claims 1 to 4.

10. A computer storage medium, characterized in that: A computer program is stored thereon, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images