Loan business document auditing method and device, equipment, storage medium and product

Abstract

The invention provides a loan business document auditing method and device, equipment, a storage medium and a product, and relates to the technical field of finance. The method comprises the steps of obtaining text information corresponding to a to-be-audited loan business document, and performing vectorization processing on the text information to obtain text vector information; field vector information corresponding to at least one auditing index is extracted from the text vector information, a preset prompt word template is called according to the field vector information corresponding to each auditing index, and auditing prompt words corresponding to the field vector information are generated; and inputting the field vector information corresponding to the at least one auditing index and the auditing prompt word corresponding to each piece of field vector information into an artificial intelligence auditing model, and generating auditing result information corresponding to the loan business document. The method can improve the auditing efficiency of the loan business document.

Description

Technical Field

[0001] This application relates to the field of financial technology, and in particular to a method, apparatus, device, storage medium and product for reviewing loan business documents. Background Technology

[0002] Currently, with the rapid development of the financial industry, the complexity and compliance requirements of banking business audits are increasing. Therefore, it is necessary to audit loan business documents to ensure their compliance and validity.

[0003] In this technology, staff manually review loan documents based on relevant business documents. For example, they review loan contracts and transaction records to ensure the documents are compliant and valid.

[0004] However, the inventors have discovered at least the following technical problems in the related technology: because for some loan business documents, when manually reviewing loan business documents, a large number of business-related documents need to be referenced, resulting in a long time spent on manual review of loan business documents, and therefore the efficiency of manual review of loan business documents is low. Summary of the Invention

[0005] This application provides a method, apparatus, device, storage medium, and product for reviewing loan business documents, in order to solve the technical problem of low efficiency in manual review of loan business documents in the prior art.

[0006] Firstly, this application provides a method for reviewing loan business documents, including:

[0007] Obtain the text information corresponding to the loan business document to be reviewed, and perform vectorization processing on the text information to obtain text vector information;

[0008] Extract at least one field vector information corresponding to the audit indicator from the text vector information, and based on the field vector information corresponding to each audit indicator, call the preset prompt word template to generate the audit prompt word corresponding to the field vector information;

[0009] Input the field vector information corresponding to at least one audit indicator and the audit prompt words corresponding to each field vector information into the artificial intelligence audit model to generate the audit result information corresponding to the loan business document.

[0010] Secondly, this application provides a loan business document review device, comprising:

[0011] The acquisition module is used to acquire the text information corresponding to the loan business document to be reviewed, and to perform vectorization processing on the text information to obtain text vector information.

[0012] The generation module is used to extract field vector information corresponding to at least one audit indicator from the text vector information, and to call a preset prompt word template to generate audit prompt words corresponding to the field vector information based on the field vector information corresponding to each audit indicator.

[0013] The review module is used to input the field vector information corresponding to at least one review indicator and the review prompt words corresponding to each field vector information into the artificial intelligence review model to generate the review result information corresponding to the loan business document.

[0014] Thirdly, this application provides an electronic device, including: a memory and a processor;

[0015] The memory stores the instructions that the computer executes;

[0016] The processor executes computer execution instructions stored in memory, causing the processor to perform the implementation method described in the first aspect above.

[0017] Fourthly, this application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the embodiments described in the first aspect above.

[0018] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, is used to implement the embodiments described in the first aspect above.

[0019] This application provides a method, apparatus, device, storage medium, and product for reviewing loan business documents. The method includes: acquiring text information corresponding to the loan business document to be reviewed; vectorizing the text information to obtain text vector information; extracting field vector information corresponding to at least one review indicator from the text vector information; calling a preset prompt word template based on the field vector information corresponding to each review indicator to generate review prompt words corresponding to the field vector information; and inputting the field vector information corresponding to at least one review indicator and the review prompt words corresponding to each field vector information into an artificial intelligence review model to generate review result information corresponding to the loan business document. In this embodiment, because the preset prompt word template is called based on the field vector information corresponding to each review indicator to generate review prompt words corresponding to the field vector information, and different review prompt words are generated for different field vector information based on the prompt word engineering, accurate and effective prompt words are generated to guide the artificial intelligence review model to quickly and accurately generate review result information corresponding to the loan business document, thus improving the review efficiency of loan business documents. Attached Figure Description

[0020] 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.

[0021] Figure 1 A schematic diagram illustrating an application scenario for a loan business document review method provided in this application embodiment;

[0022] Figure 2 The flowchart of a loan business document review method provided in this application embodiment Figure 1 ;

[0023] Figure 3 The flowchart of a loan business document review method provided in this application embodiment Figure 2 ;

[0024] Figure 4 The flowchart of a loan business document review method provided in this application embodiment Figure 3 ;

[0025] Figure 5 A schematic diagram of the structure of a loan business document review device provided in this application embodiment;

[0026] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0027] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0028] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0029] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of the relevant data all comply with the relevant laws, regulations, and standards of the relevant countries and regions, have taken necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation portals for users to choose to authorize or refuse.

[0030] Furthermore, the technical solution involved in this application, which involves big data analysis of user information (including but not limited to personal biometrics, identity data, consumption data, asset data, electronic terminal operation data, etc.) and the use of artificial intelligence technology for automated decision-making, and makes decisions that have a significant impact on personal rights based on the results of automated decision-making, provides users with corresponding operation entry points for users to choose to agree to or reject the results of automated decision-making; if the user chooses to reject, the process will proceed to the expert decision-making process.

[0031] It should be noted that the loan business document review methods, devices, equipment, storage media and products provided in this application can be used in the field of financial technology, or in any field other than financial technology. The application fields of the loan business document review methods, devices, equipment, storage media and products in this application are not limited.

[0032] Currently, with the rapid development of the financial industry, the complexity and compliance requirements of banking business audits are increasing. Therefore, it is necessary to audit loan business documents to ensure their compliance and validity.

[0033] In related technologies, loan business documents are manually reviewed by staff based on relevant business documents. For example, loan contracts and transaction records are reviewed using documents related to bank loan business to ensure compliance and validity. However, for some loan business documents, the number of relevant business documents that need to be consulted during manual review is large, resulting in a long time consumption and low efficiency.

[0034] Therefore, improving the efficiency of reviewing loan application documents is a pressing technical problem that needs to be solved.

[0035] To solve the above-mentioned technical problems, the inventors' technical concept is as follows: This invention provides an artificial intelligence model based on Prompt Engineering, which, by designing a reasonable prompt word model, constructs an efficient prompt word generation and model interaction system to achieve intelligent compliance review of loan business documents.

[0036] Accordingly, the specific steps may include: First, obtaining the text information corresponding to the loan business document to be reviewed, and vectorizing the text information to obtain text vector information. Then, extracting field vector information corresponding to at least one review indicator from the text vector information, and based on the field vector information corresponding to each review indicator, calling a preset prompt word template to generate review prompt words corresponding to the field vector information. Finally, inputting the field vector information corresponding to at least one review indicator and the review prompt words corresponding to each field vector information into the artificial intelligence review model to generate review result information corresponding to the loan business document.

[0037] In this technical solution, by calling a preset prompt word template based on the field vector information corresponding to each audit indicator, audit prompt words corresponding to the field vector information are generated. Based on the prompt word engineering, different audit prompt words are generated for different field vector information, so as to generate accurate and effective prompt words. This guides the artificial intelligence audit model to quickly and accurately generate the audit result information corresponding to the loan business document, thereby improving the audit efficiency of the loan business document.

[0038] Figure 1 This is a schematic diagram illustrating an application scenario for a loan business document review method provided in an embodiment of this application. For example... Figure 1 As shown, terminal 101 and server 102 are connected via a network. Users can send review requests and loan business documents to be reviewed to server 102 through terminal 101. Server 102 receives the review request, reviews the loan business documents using the review method provided in this application, and generates corresponding review result information. Server 102 returns the generated review result information to terminal 101 for display.

[0039] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0040] Figure 2 The flowchart of a loan business document review method provided in this application embodiment Figure 1 ,like Figure 2 As shown, the method includes:

[0041] S201. Obtain the text information corresponding to the loan business document to be audited, perform vectorization processing on the text information, and obtain text vector information.

[0042] In some embodiments, the number of loan business documents to be audited can be one or more. Optionally, preprocessing can be performed on each loan business document to obtain structured data that can be processed by the prompt word model. Optionally, as Figure 3 shown, this step may include the following steps S11 to S13:

[0043] S11. Obtain the initial text information corresponding to the loan business document to be audited, delete the invalid characters and preset format characters in the initial text information, and obtain the text information corresponding to the loan business document to be audited.

[0044] Optionally, for editable loan business documents, key information can be directly parsed and extracted to obtain the initial text information corresponding to the loan business document. For non-editable documents, such as PDF format documents, OCR (Optical Character Recognition) technology can be used for text recognition to obtain the initial text information corresponding to the loan business document.

[0045] It should be noted that by cleaning the initial text information, invalid characters, noise information, etc. are removed to obtain the text information corresponding to the loan business document to be audited. Among them, invalid characters may include punctuation marks such as colons, periods, and double quotes. Preset format characters (i.e., noise information) may include extra spaces, line breaks, format symbols, etc. Preset format characters may also include repeated words, for example, "of", "is", "and", etc.

[0046] S12. For each sentence text information in the text information, determine the sentence vector information corresponding to the sentence text information.

[0047] Optionally, in the process of converting text information into structured data that can be understood by the model, in order to retain the context information, logical association, and semantic features in the text information, feature extraction and vectorization are required. Correspondingly, this step may include:

[0048] (21) For each sentence text information in the text information, perform vectorization processing on each word information in the sentence text information to obtain the word vector corresponding to the word information.

[0049] (22) For each word information in the sentence text information, obtain the word frequency parameter and inverse document frequency parameter corresponding to the word information, and combine the word frequency parameter and inverse document frequency parameter to determine the weight value corresponding to the word information.

[0050] Among them, the vocabulary in the document (i.e., the text information mentioned above) can be statistically analyzed. The frequency of each word in the document can be calculated using the following formula to obtain the word frequency parameter corresponding to the word information.

[0051] Formula 1:

[0052]

[0053] Where t represents word information, d represents document, TF(t,d) represents the word frequency parameter corresponding to the word information, and f(t,d) represents the number of times the word information appears in the document, ∑ ω∈d f(ω,d) represents the total number of times all word information appears in the document.

[0054] Optionally, the inverse document frequency (IVF) parameter is used to characterize the importance of word information among all documents (i.e., the text information corresponding to the multiple loan business documents to be reviewed mentioned above). Accordingly, the IVF parameter corresponding to each word information can be calculated using the following Formula 2;

[0055] Formula 2:

[0056]

[0057] Where t represents word information, IDF(t, D) represents the inverse document frequency parameter corresponding to word information t, |D| represents the total number of documents in the corpus, and |{d∈D:t∈d}| is the number of documents containing word information t (i.e., the number of documents in which the word appears at least once).

[0058] Optionally, the weight value corresponding to the word information is determined by combining the term frequency parameter and the inverse document frequency parameter, including: determining the weight value corresponding to the word information based on the product of the term frequency parameter and the inverse document frequency parameter. This can be calculated using the following formula:

[0059] Formula 3: TF-IDF(t,d,D)=TF(t,d)×IDF(t,D);

[0060] Where TF-IDF(t,d,D) represents the weight value corresponding to the word information, TF(t,d) represents the word frequency parameter corresponding to the word information, and IDF(t,D) represents the inverse document frequency parameter corresponding to the word information t.

[0061] (23) Based on the weight value corresponding to each word information, the word vector corresponding to each word information is weighted and summed to obtain the sentence vector information corresponding to the sentence text information.

[0062] Optionally, each word is mapped to a fixed-dimensional vector, word embeddings are generated using a word vector model, and then the word vectors of the entire sentence are weighted and summed to obtain the sentence vector information. This can be calculated using the following formula:

[0063] Formula 4:

[0064] Where S represents sentence vector information, v(t) i ) represents the i-th weighted word vector, and N represents the total number of word vectors in the sentence vector information.

[0065] S13. Combine the sentence vector information corresponding to the text information of each sentence to obtain the text vector information.

[0066] S202. Extract at least one field vector information corresponding to the audit indicator from the text vector information. Based on the field vector information corresponding to each audit indicator, call the preset prompt word template to generate the audit prompt word corresponding to the field vector information. The audit indicators include one or more of the following: interest rate terms, repayment period, risk warning, guarantee terms, and default terms.

[0067] In some embodiments, the loan document is segmented into logical paragraphs or clauses through document preprocessing and natural language processing technologies, key fields are extracted, and finally, approval prompts are generated by combining the key fields with prompt templates.

[0068] Optionally, the audit indicators include one or more of the following: interest rate terms, repayment period, risk warnings, guarantee terms, and default terms; correspondingly, at least one field vector information corresponding to the audit indicator is extracted from the text vector information, including: extracting one or more of the following: field vector information corresponding to the interest rate terms, field vector information corresponding to the repayment period, field vector information corresponding to the risk warnings, field vector information corresponding to the guarantee terms, and field vector information corresponding to the default terms.

[0069] For example, the interest rate field includes an interest rate figure and description, such as "annual interest rate of 5%". The term field includes a time description, such as "repayment period of 36 months". The default clause field includes a penalty percentage, such as "penalty of 3% of the outstanding amount".

[0070] In some embodiments, the goal of prompt word design is to generate questions or task instructions that can guide deep learning models to answer effectively.

[0071] Optionally, the prompt design includes static and dynamic prompts. Static prompts are suitable for general compliance metrics, generating fixed inspection tasks; dynamic prompts are generated dynamically based on the context of the document content, adapting to specific scenarios and content. Accordingly, such as Figure 4 As shown, based on the field vector information corresponding to each audit indicator, a preset prompt word template is called to generate audit prompt words corresponding to the field vector information, which may include the following steps S21 to S22:

[0072] S21. Based on the field vector information corresponding to each audit indicator, determine the prompt word type corresponding to the audit indicator; where the prompt word type includes static prompt word type and dynamic prompt word type.

[0073] Optionally, this step includes: determining the type of prompt word corresponding to the interest rate clause as a static prompt word type based on the field vector information corresponding to the interest rate clause; determining the type of prompt word corresponding to the repayment period as a static prompt word type based on the field vector information corresponding to the repayment period; determining the type of prompt word corresponding to the risk warning as a dynamic prompt word type based on the field vector information corresponding to the risk warning; determining the type of prompt word corresponding to the guarantee clause as a dynamic prompt word type based on the field vector information corresponding to the guarantee clause; and determining the type of prompt word corresponding to the default clause as a dynamic prompt word type based on the field vector information corresponding to the default clause.

[0074] S22. Call the preset prompt word template to generate the audit prompt word corresponding to the prompt word type.

[0075] In some embodiments, based on the field vector information corresponding to the interest rate terms, the prompt word type corresponding to the interest rate terms is determined to be a static prompt word type. A preset prompt word template is called to generate the audit prompt word corresponding to the static prompt word type as follows: Check whether the document contains the annual loan interest rate and determine whether the annual loan interest rate is within the specified range (e.g., determine whether the annual loan interest rate is lower than 24%).

[0076] Optionally, based on the field vector information corresponding to the repayment period, the prompt word type corresponding to the repayment period is determined to be a static prompt word type. A preset prompt word template is called to generate the audit prompt word corresponding to the static prompt word type as: Check whether the repayment period is within the specified range (e.g., check whether the repayment period does not exceed 60 months).

[0077] In some embodiments, the step of calling a preset prompt word template to generate an audit prompt word corresponding to a dynamic prompt word type includes: extracting field vector information through a preprocessing module, filling the field vector information into the prompt word template, and generating a complete prompt word.

[0078] For example, the prompt template is: Generate an inspection task based on {clause text}, determine whether the description of the penalty for breach of contract in this clause is clear, and assess its reasonableness.

[0079] The input terms state that if the borrower fails to repay on time, the penalty will be 10% of the outstanding amount.

[0080] The generated audit prompt is: Does the description of the penalty for breach of contract meet the requirements? Please determine whether "the penalty is 10% of the outstanding amount" is within a reasonable range.

[0081] In some embodiments, the prompt word model includes a prompt word model corresponding to a first risk level, a prompt word model corresponding to a second risk level, and a prompt word model corresponding to a third risk level. Accordingly, generating review prompt words corresponding to the prompt word type by calling a preset prompt word template includes: obtaining business information corresponding to the loan business document; determining the risk level information corresponding to the loan business document based on the business type and application information in the business information; calling a prompt word template matching the risk level information; and generating review prompt words corresponding to the prompt word type.

[0082] For example, the prompt word model corresponding to the first risk level is a high-risk prompt word model, the prompt word model corresponding to the second risk level is a medium-risk prompt word model, and the prompt word model corresponding to the third risk level is a low-high-risk prompt word model. It should be noted that in some embodiments, the high-risk prompt word model is applicable to loan customers with higher risk or loan scenarios with more uncertainties. Correspondingly, the review prompt words emphasize detailed checks and rigorous compliance analysis. For example, the review prompt words include "complex guarantee structure." The medium-risk prompt word model is applicable to customer groups with lower risk but still requiring attention. Correspondingly, the review prompt words emphasize the checking of key fields and compliance verification. For example, the review prompt words include "interest rate range check." The low-risk prompt word model is applicable to high-quality customers or low-risk scenarios. Correspondingly, the review prompt words emphasize efficiency, with simple and direct prompts. For example, the review prompt words include "fast review" and "high-quality customer."

[0083] S203. Input the field vector information corresponding to at least one audit indicator and the audit prompt words corresponding to each field vector information into the artificial intelligence audit model to generate the audit result information corresponding to the loan business document.

[0084] In this embodiment, a rule database corresponding to each audit indicator can be pre-set. The field vector information and audit prompts corresponding to each audit indicator are input into the AI ​​audit model. This AI audit model can call the rule database corresponding to the audit indicator to generate audit result information for each field vector information. Combining the audit result information for each field vector information, the audit result information corresponding to the loan business document is generated.

[0085] For example, a rules database may include rules terms such as interest rate terms, repayment periods, risk warnings, guarantee terms, and default terms.

[0086] Optionally, the field vector information corresponding to at least one audit indicator includes: field vector information corresponding to the interest rate clause and field vector information corresponding to the default clause. The audit prompt for the field vector information corresponding to the interest rate clause is: Determine whether the annual loan interest rate is lower than 24%. The audit prompt for the field vector information corresponding to the default clause is: Does the description of the penalty for default meet the requirements? Please determine whether "the penalty is 10% of the outstanding amount" is within a reasonable range.

[0087] The AI-generated audit model's field vector information for the interest rate clause indicates that the annual loan interest rate is below 24%. Similarly, the AI-generated audit model's field vector information for the default clause indicates that the description of the penalty for default meets the requirements, and that "the penalty is 10% of the outstanding amount" is within a reasonable range. Therefore, the audit results for the loan document are: annual loan interest rate below 24%, description of the penalty for default meets the requirements, and "the penalty is 10% of the outstanding amount" is within a reasonable range.

[0088] This application provides a method for reviewing loan business documents: First, the text information corresponding to the loan business document to be reviewed is obtained, and the text information is vectorized to obtain text vector information. Then, at least one field vector information corresponding to a review indicator is extracted from the text vector information. Based on the field vector information corresponding to each review indicator, a preset prompt word template is called to generate review prompt words corresponding to the field vector information. Finally, the field vector information corresponding to at least one review indicator and the review prompt words corresponding to each field vector information are input into an artificial intelligence review model to generate review result information corresponding to the loan business document. In this embodiment, because a preset prompt word template is called based on the field vector information corresponding to each review indicator to generate review prompt words corresponding to the field vector information, and different review prompt words are generated for different field vector information based on the prompt word engineering, accurate and effective prompt words are generated to guide the artificial intelligence review model to quickly and accurately generate review result information corresponding to the loan business document, thus improving the review efficiency of loan business documents.

[0089] Figure 5 A schematic diagram of the structure of a loan business document review device provided in this application embodiment. Figure 1 ,like Figure 5 As shown, the device includes:

[0090] The acquisition module 501 is used to acquire the text information corresponding to the loan business document to be reviewed, and to perform vectorization processing on the text information to obtain text vector information.

[0091] The generation module 502 is used to extract at least one field vector information corresponding to the audit indicator from the text vector information, and to call a preset prompt word template to generate the audit prompt word corresponding to the field vector information based on the field vector information corresponding to each audit indicator.

[0092] The review module 503 is used to input the field vector information corresponding to at least one review indicator and the review prompt words corresponding to each field vector information into the artificial intelligence review model to generate the review result information corresponding to the loan business document.

[0093] In one possible implementation, the audit indicators include one or more of the following: interest rate terms, repayment period, risk warning, guarantee terms, and default terms; correspondingly, the generation module 502 extracts field vector information corresponding to at least one audit indicator from the text vector information, including: extracting one or more of the following: field vector information corresponding to the interest rate terms, field vector information corresponding to the repayment period, field vector information corresponding to the risk warning, field vector information corresponding to the guarantee terms, and field vector information corresponding to the default terms.

[0094] In one possible implementation, the generation module 502, based on the field vector information corresponding to each audit indicator, calls a preset prompt word template to generate audit prompt words corresponding to the field vector information, including: determining the prompt word type corresponding to the audit indicator based on the field vector information corresponding to each audit indicator; wherein, the prompt word type includes static prompt word type and dynamic prompt word type; and calling the preset prompt word template to generate audit prompt words corresponding to the prompt word type.

[0095] In one possible implementation, the generation module 502 determines the prompt word type corresponding to each audit indicator based on the field vector information corresponding to each audit indicator, including: determining the prompt word type corresponding to the interest rate clause as a static prompt word type based on the field vector information corresponding to the interest rate clause; determining the prompt word type corresponding to the repayment period as a static prompt word type based on the field vector information corresponding to the repayment period; determining the prompt word type corresponding to the risk warning as a dynamic prompt word type based on the field vector information corresponding to the risk warning; determining the prompt word type corresponding to the guarantee clause as a dynamic prompt word type based on the field vector information corresponding to the guarantee clause; and determining the prompt word type corresponding to the default clause as a dynamic prompt word type based on the field vector information corresponding to the default clause.

[0096] In one possible implementation, the generation module 502 calls a preset prompt word template to generate an audit prompt word corresponding to the prompt word type, including: obtaining business information corresponding to the loan business document; determining the risk level information corresponding to the loan business document based on the business type and application information in the business information; and calling a prompt word template that matches the risk level information to generate an audit prompt word corresponding to the prompt word type.

[0097] In one possible implementation, the acquisition module 501 acquires text information corresponding to the loan business document to be reviewed, and performs vectorization processing on the text information to obtain text vector information, including: acquiring initial text information corresponding to the loan business document to be reviewed, deleting invalid characters and preset format characters in the initial text information to obtain text information corresponding to the loan business document to be reviewed; determining sentence vector information corresponding to each sentence text information in the text information; and combining the sentence vector information corresponding to each sentence text information to obtain the text vector information.

[0098] In one possible implementation, the acquisition module 501 determines the sentence vector information corresponding to each sentence text information in the text information, including: for each sentence text information in the text information, vectorizing each word information in the sentence text information to obtain the word vector corresponding to the word information; for each word information in the sentence text information, acquiring the word frequency parameter and inverse document frequency parameter corresponding to the word information, and combining the word frequency parameter and the inverse document frequency parameter to determine the weight value corresponding to the word information; and performing a weighted summation of the word vectors corresponding to each word information according to the weight value corresponding to each word information to obtain the sentence vector information corresponding to the sentence text information.

[0099] This application provides a loan business document review device that, based on the field vector information corresponding to each review indicator, calls a preset prompt word template to generate review prompt words corresponding to the field vector information. Based on the prompt word engineering, different review prompt words are generated for different field vector information, so as to generate accurate and effective prompt words, guiding the artificial intelligence review model to quickly and accurately generate the corresponding review result information of the loan business document, thereby improving the review efficiency of loan business documents.

[0100] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 6As shown, the electronic device provided in this embodiment includes at least one processor 601 and a memory 602. Optionally, the electronic device further includes a communication component 603. The processor 601, memory 602, and communication component 603 are connected via a bus 604.

[0101] In a specific implementation, at least one processor 601 executes computer execution instructions stored in memory 602, causing at least one processor 601 to perform the above-described method.

[0102] The specific implementation process of processor 601 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0103] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0104] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0105] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0106] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0107] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0108] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0109] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0110] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this application.

[0111] It should be further noted that although the steps in the flowchart are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowchart may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0112] It should be noted that the terms "first," "second," etc., in the claims, specification, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, products, or apparatus.

[0113] It should be understood that the above-described device embodiments are merely illustrative, and the device of this application can also be implemented in other ways. For example, the division of units / modules in the above embodiments is only a logical functional division, and there may be other division methods in actual implementation. For example, multiple units, modules, or components may be combined, or integrated into another system, or some features may be ignored or not executed.

[0114] Furthermore, unless otherwise specified, the functional units / modules in the various embodiments of this application can be integrated into one unit / module, or each unit / module can exist physically separately, or two or more units / modules can be integrated together. The integrated units / modules described above can be implemented in hardware or as software program modules.

[0115] When integrated units / modules are implemented in hardware, the hardware can be digital circuits, analog circuits, etc. The physical implementation of the hardware structure includes, but is not limited to, transistors, memristors, etc. Unless otherwise specified, the processor can be any suitable hardware processor, such as a CPU, GPU, FPGA, DSP, and ASIC, etc. Unless otherwise specified, the storage unit can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc.

[0116] If the integrated unit / module is implemented as a software program module and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0117] In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments. The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0118] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0119] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for reviewing loan business documents, characterized in that, The method includes: Obtain the text information corresponding to the loan business document to be reviewed, and perform vectorization processing on the text information to obtain text vector information; Extract at least one field vector information corresponding to the audit indicator from the text vector information. Based on the field vector information corresponding to each audit indicator, call the preset prompt word template to generate the audit prompt word corresponding to the field vector information. The audit indicator includes one or more of the following: interest rate terms, repayment period, risk warning, guarantee terms, and default terms. Input the field vector information corresponding to at least one audit indicator and the audit prompt words corresponding to each field vector information into the artificial intelligence audit model to generate the audit result information corresponding to the loan business document.

2. The auditing method according to claim 1, characterized in that, The step of extracting at least one field vector information corresponding to an audit indicator from the text vector information includes: Extract one or more of the following from the text vector information: the field vector information corresponding to the interest rate clause, the field vector information corresponding to the repayment period, the field vector information corresponding to the risk warning, the field vector information corresponding to the guarantee clause, and the field vector information corresponding to the default clause.

3. The auditing method according to claim 1, characterized in that, The step of generating audit prompts corresponding to the field vector information by calling a preset prompt template based on the field vector information for each audit indicator includes: Based on the field vector information corresponding to each audit indicator, the prompt word type corresponding to the audit indicator is determined; wherein, the prompt word type includes static prompt word type and dynamic prompt word type; The preset prompt word template is invoked to generate the review prompt word corresponding to the prompt word type.

4. The auditing method according to claim 3, characterized in that, The step of determining the prompt word type corresponding to each audit indicator based on the field vector information includes: Based on the field vector information corresponding to the interest rate terms, the prompt word type corresponding to the interest rate terms is determined to be a static prompt word type; based on the field vector information corresponding to the repayment period, the prompt word type corresponding to the repayment period is determined to be a static prompt word type; Based on the field vector information corresponding to the risk warning, the prompt word type corresponding to the risk warning is determined to be a dynamic prompt word type; based on the field vector information corresponding to the guarantee clause, the prompt word type corresponding to the guarantee clause is determined to be a dynamic prompt word type; based on the field vector information corresponding to the default clause, the prompt word type corresponding to the default clause is determined to be a dynamic prompt word type.

5. The auditing method according to claim 3, characterized in that, The step of calling a preset prompt template to generate the review prompt corresponding to the prompt type includes: Obtain the business information corresponding to the loan business document, and determine the risk level information corresponding to the loan business document based on the business type and application information in the business information. Call the prompt word template that matches the risk level information to generate the audit prompt word corresponding to the prompt word type.

6. The auditing method according to claim 1, characterized in that, The step of obtaining the text information corresponding to the loan business document to be reviewed, and vectorizing the text information to obtain text vector information includes: Obtain the initial text information corresponding to the loan business document to be reviewed, delete invalid characters and preset format characters in the initial text information, and obtain the text information corresponding to the loan business document to be reviewed; For each sentence in the text information, determine the sentence vector information corresponding to that sentence text. The text vector information is obtained by combining the sentence vector information corresponding to the text information of each sentence.

7. The auditing method according to claim 6, characterized in that, The step of determining the sentence vector information corresponding to each sentence text information in the text information includes: For each sentence text in the text information, each word in the sentence text information is vectorized to obtain the word vector corresponding to the word information; For each word in the sentence text information, obtain the word frequency parameter and inverse document frequency parameter corresponding to the word information, and combine the word frequency parameter and the inverse document frequency parameter to determine the weight value corresponding to the word information; Based on the weight value corresponding to each word, the word vectors corresponding to each word are weighted and summed to obtain the sentence vector information corresponding to the sentence text information.

8. A device for reviewing loan business documents, characterized in that, The device includes: The acquisition module is used to acquire the text information corresponding to the loan business document to be reviewed, and to perform vectorization processing on the text information to obtain text vector information. The generation module is used to extract field vector information corresponding to at least one audit indicator from the text vector information, and to call a preset prompt word template to generate audit prompt words corresponding to the field vector information based on the field vector information corresponding to each audit indicator. The audit indicators include one or more of the following: interest rate terms, repayment period, risk warning, guarantee terms, and default terms. The review module is used to input the field vector information corresponding to at least one review indicator and the review prompt words corresponding to each field vector information into the artificial intelligence review model to generate the review result information corresponding to the loan business document.

9. An electronic device, characterized in that, The electronic device includes: a memory and a processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the loan business document review method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the loan business document review method as described in any one of claims 1-7.

11. A computer program product comprising a computer program that, when executed by a processor, implements the method for reviewing loan business documents as described in any one of claims 1-7.

Images