Contract element input method, electronic equipment, storage medium and program product

Abstract

The embodiment of the invention provides a contract element input method, electronic equipment, a storage medium and a program product, and relates to the technical field of artificial intelligence, and the method comprises the steps: recognizing characters in a contract image of a target contract through a character recognition model, and obtaining a contract text of the target contract; inputting the contract image, the contract text and the element extraction cue word of the target contract into a contract element extraction model so as to extract contract elements of the target contract from the contract image according to the contract text and the element extraction cue word by utilizing the contract element extraction model; verifying the extracted contract elements of the target contract by using the contract text of the target contract; and inputting contract elements of the target contract in the target business system according to the verification result. According to the method, text input is provided for the model as a reference in the early stage, extracted elements are verified in the later stage, it is ensured that the extracted contract elements are highly consistent with the original contract content through double guarantee, and wrong input caused by model illusion is avoided.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of artificial intelligence technology, and in particular to a method for inputting contract elements, an electronic device, a storage medium, and a program product. Background Technology

[0002] In the online processing of banking transactions, the accurate entry of contract elements is crucial. Traditional methods rely on manual operation, requiring bank staff to manually input each contract element into the system, resulting in low efficiency. With the development of artificial intelligence technology, business contracts can be visualized, and AI models can be used to identify contract elements within the image and input them into the system, thereby improving efficiency.

[0003] However, current models often suffer from the illusion problem in practical applications, that is, generating content that does not match the original document, resulting in inaccurate extracted contract elements and failing to meet the needs of online business processing. Summary of the Invention

[0004] This invention provides a method for entering contract elements, an electronic device, a storage medium, and a program product, which can improve the accuracy of contract element identification and entry, and meet the needs of online business processing.

[0005] In a first aspect, the contract element input method provided in the embodiments of the present invention includes:

[0006] The character recognition model is used to identify the characters in the contract image of the target contract to obtain the contract text of the target contract;

[0007] Input the contract image, contract text, and element extraction prompts of the target contract into the contract element extraction model, so as to use the contract element extraction model to extract the contract elements of the target contract from the contract image based on the contract text and element extraction prompts;

[0008] The contract elements extracted by the contract element extraction model are verified using the contract text of the target contract.

[0009] Based on the verification results, the contract elements of the target contract are entered into the target business system.

[0010] Secondly, the contract element input device provided in the embodiments of the present invention includes:

[0011] The character recognition module is used to identify characters in the contract image of the target contract using a character recognition model, thereby obtaining the contract text of the target contract;

[0012] The element extraction module is used to input the contract image, contract text, and element extraction prompts of the target contract into the contract element extraction model, so as to use the contract element extraction model to extract the contract elements of the target contract from the contract image based on the contract text and element extraction prompts.

[0013] The verification module is used to verify the contract elements of the target contract extracted by the contract element extraction model using the contract text of the target contract.

[0014] The data entry module is used to enter the contract elements of the target contract into the target business system based on the verification results.

[0015] Thirdly, the electronic device provided in the embodiments of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the contract element input method as described in any embodiment of the present invention.

[0016] Fourthly, the computer-readable storage medium provided in the embodiments of the present invention stores a computer program thereon, which, when executed by a processor, implements the contract element entry method as described in any embodiment of the present invention.

[0017] Fifthly, the computer program product provided in the embodiments of the present invention includes a computer program that, when executed by a processor, implements the contract element input method as described in any embodiment of the present invention.

[0018] In this embodiment of the invention, a character recognition model is used to identify characters in the contract image of the target contract to obtain the contract text. The character recognition model focuses on extracting text content from the image, providing a relatively accurate and complete text foundation. This provides additional text input for the contract element extraction model, avoiding the illusion problem that may occur when directly using the model due to image feature misleading information. Simultaneously, this contract text serves as a key reference for subsequent verification of the accuracy of extracted elements. After the contract element extraction model extracts the contract elements based on the contract image, contract text, and element extraction prompts, these elements are verified again using the previously extracted contract text to determine whether the extracted elements match the original document content. This verification... The process can promptly identify and correct potential illusions in the contract element extraction model, ensuring that the contract elements ultimately entered into the business system are highly consistent with the original contract content, thereby improving the accuracy and reliability of data entry. Specifically, it provides additional text input to the model in the early stages and rigorously verifies the elements extracted by the model in the later stages, providing dual protection to ensure a high degree of consistency between the extracted contract elements and the original contract content, effectively avoiding erroneous entry caused by model illusions. Utilizing character recognition and contract element extraction models, it achieves automatic identification, extraction, and entry of contract elements from contract images, greatly reducing the workload and time required for manual identification and entry, improving the efficiency of business processing, and better adapting to the needs of the rapid online development of banking services. Attached Figure Description

[0019] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating a contract element input method provided in an embodiment of the present invention;

[0021] Figure 2 This is another flowchart illustrating the contract element input method provided in this embodiment of the invention;

[0022] Figure 3a This is an example diagram of the contract element extraction method provided in an embodiment of the present invention;

[0023] Figure 3b This is an example diagram of the contract element verification method provided in an embodiment of the present invention;

[0024] Figure 4 This is a schematic diagram of a contract element input device provided in an embodiment of the present invention;

[0025] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0026] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention 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 the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0028] Figure 1This is a flowchart illustrating a contract element input method provided in an embodiment of the present invention. This method is applicable to scenarios where online financial transactions require the input of contract elements. Examples of online financial transactions include credit, bill transactions, insurance, and asset management. For instance, when a customer applies for a loan from a bank, they need to input elements from the loan contract into the system for online approval. These elements may include customer information, loan amount, interest rate, and repayment plan. Similarly, when a customer applies for bill discounting from a bank, the bank needs to input elements from the bill discounting contract, including the bill number, drawer, payee, bill amount, discount rate, and discount period. When processing group insurance, elements from the group insurance contract need to be input, including policyholder information, insured list, insurance type, insured amount, and insurance period. When a bank sells wealth management products, elements from the wealth management contract need to be input, including investor information, investment amount, product term, expected rate of return, and risk level. For ease of description, the following explanation uses credit transactions as an example. The contract element input method can be executed by the contract element input device provided in this embodiment of the invention, which can be implemented in software and / or hardware. In a specific embodiment, the device can be integrated into an electronic device, such as a computer or server. The following embodiment illustrates the integration of the contract element input device into an electronic device. See also... Figure 1 The contract element input method in this embodiment may include the following steps:

[0029] Step 101: Use a character recognition model to identify the characters in the contract image of the target contract to obtain the contract text of the target contract.

[0030] A character recognition model is a software or tool used to convert the text content in an image into editable and processable text. Character recognition models can include traditional optical character recognition models, end-to-end text generation models, deep learning sequence models, etc. The target contract is a paper contract, which refers to a contract signed by a customer with a bank or other relevant parties during the online process of financial business and currently needs to be entered and processed for elements. It may be a loan contract, a mortgage contract, a bank note contract, etc., and contains key business elements such as information of both parties to the contract, loan amount, interest rate, repayment method, etc. A contract image refers to a digital image file obtained by converting a paper contract through scanning, photographing, etc. When handling online financial business, a bank customer will photograph or scan the retained paper contract to generate a contract image and upload it to the bank system in a packaged form for subsequent automated processing. A character is the smallest unit that makes up text, including letters, numbers, punctuation marks, Chinese characters, etc. In a contract image, characters are presented in the form of images. The goal of a character recognition model is to recognize these character images and convert them into corresponding text characters. For example: "A", "B", "1", "2", ".", etc. are all characters. Contract text is the electronic text content obtained by recognizing the characters in a contract image through a character recognition model. It reflects the character information in the contract image and is the basis for subsequent extraction and verification of contract elements. The contract text contains all the character information in the contract and is arranged in the order and format of the original document. For example, a loan contract text may contain content such as "Borrower: ABC Company, Lender: DEF Bank, Loan Amount: 1 million yuan".

[0031] In the process of handling credit business online, the traditional manual entry of contract elements is inefficient and error-prone. To improve efficiency and accuracy, a character recognition model can be used to recognize all the characters in the contract image uploaded by the customer and generate contract text.

[0032] Currently, there are also those that directly use traditional optical character recognition models to recognize contract elements. However, traditional optical character recognition models are usually trained or configured for contracts with fixed layouts (such as standard invoices or forms). They rely on predefined rules (such as text position, font size, keyword matching) to extract elements and have poor generalization ability and are powerless for contracts with very different styles. For example, an optical character recognition model trained to recognize a "standard procurement contract" may only look for the "contract number" at a specific position (such as the upper right corner). In actual application contracts, the styles are very different, the layouts are diverse, and the formats vary greatly. Traditional optical character recognition models may not be able to locate or recognize elements, resulting in the failure of contract element extraction.

[0033] In this embodiment, when using an optical character recognition (OCR) model, to handle contracts with vastly different styles, the OCR model is no longer used to directly extract contract elements (i.e., not to directly locate or identify specific elements). Instead, the OCR model is used to extract the "all text content" of the contract image, that is, to extract the text string of the entire document. For example, the OCR model can output all the text in the contract, including the title, body text, footer, etc. The core capability of the OCR model is "character recognition," which can convert image pixels into text regardless of the contract layout. Even if the OCR model is not perfect in recognizing contracts with vastly different styles, it can still provide most of the text content, providing additional input and constraints for the subsequent contract element extraction model.

[0034] Step 102: Input the contract image, contract text, and element extraction prompts of the target contract into the contract element extraction model, so as to use the contract element extraction model to extract the contract elements of the target contract from the contract image based on the contract text and element extraction prompts.

[0035] Element extraction prompts are keywords, phrases, and instructions used to guide contract element extraction models in extracting specific contract elements, helping the model accurately locate and extract the required information. In credit transactions, prompts might include phrases like "loan amount," "interest rate," and "repayment period." These prompts help the model identify and extract corresponding values ​​and information from the contract image. A contract element extraction model is a multimodal artificial intelligence model capable of automatically identifying and extracting key elements from an input contract image, contract text, and element extraction prompts. Contract element extraction models can be pre-trained using training data. A trained contract element extraction model can understand contract text and image content and extract key information such as the loan amount and interest rate based on the prompts. Contract elements are key information items extracted from the target contract; these elements are the focus of credit transaction processing. For example, contract elements in credit transactions might include the loan amount, interest rate, repayment period, loan purpose, and collateral.

[0036] Images provide authenticity verification for text, text provides semantic interpretation for images, and prompts constrain the output; these three elements form a closed-loop reliable system. Through three input channels—contract scan images, text extracted by a character recognition model, and business instructions—a multimodal AI model is driven to accurately extract the key business data required by the bank.

[0037] Step 103: Validate the contract elements of the target contract extracted by the contract element extraction model using the contract text of the target contract.

[0038] The extracted contract elements can be compared with the contract text to check whether the extracted elements are consistent with the content in the text, thereby ensuring the accuracy and authenticity of the extraction results.

[0039] For example, a contract element extraction model can extract the element "Loan Amount: 1 million yuan" from a loan contract. This can be achieved by using a fuzzy matching algorithm to search for descriptions related to "Loan Amount" in the contract text and calculating the similarity between the extracted element and the text content. The similarity between the extracted element and the text content can be measured by the edit distance between them, or by vectorizing them and measuring cosine similarity. If the similarity exceeds a preset threshold (e.g., 80%), the extracted element is considered accurate, and the verification passes; conversely, if the similarity is below the threshold, the extracted element may contain errors or misleading information, and the verification fails.

[0040] This verification mechanism can effectively reduce extraction errors caused by model illusions or other errors, ensuring that the contract elements entered into the business system are highly consistent with the original contract content, thereby improving the accuracy and security of business processing.

[0041] Step 104: Based on the verification results, enter the contract elements of the target contract into the target business system.

[0042] The verification result refers to the conclusion after verifying the contract elements extracted by the contract element extraction model, that is, determining whether the extracted elements are consistent with the contract text of the target contract. For example, the verification result shows that the element "Loan amount: 1 million yuan" is consistent with the contract text and the verification passes; while the element "Interest rate: 5%" does not match the contract text and the verification fails.

[0043] The target business system refers to the system used by a bank or financial institution to process and manage financial business, such as a credit management system or a customer relationship management system. For example, a bank's credit management system is used to record and manage information such as customer credit applications, approval status, and loan disbursement.

[0044] During the online processing of credit transactions, the system inputs the extracted contract elements into the target business system based on the verification results. Specifically, verified contract elements are entered into the corresponding locations in the target business system, while unverified elements are left blank or marked as requiring manual review and correction. This approach ensures the accuracy and completeness of the contract elements recorded in the target business system, providing reliable data support for subsequent credit approval, loan disbursement, and post-loan management.

[0045] For example, if the system verifies that the elements "Loan Amount: 1 million yuan" and "Repayment Term: 2 years" pass the verification, they can be entered into the corresponding loan amount and repayment term fields in the credit management system. However, if the element "Interest Rate: 5%" fails the verification, an empty value will be entered in the interest rate field or it will be marked as suspicious, prompting credit personnel to conduct manual review and correction to ensure that the interest rate information recorded in the system is accurate.

[0046] The character recognition model in this embodiment can employ a traditional optical character recognition (OCR) model. The weaknesses of the OCR model can be compensated for by a multimodal artificial intelligence (AI) model (because the multimodal AI model can handle layout variations). Furthermore, the text output of the OCR model is used for post-processing (fuzzy matching) to constrain the multimodal AI model. In this way, the role of the OCR model is downgraded from "feature extractor" to "text provider," and the combination of the multimodal AI model and fuzzy matching enhances robustness. Ultimately, the solution in this embodiment can reduce illusions while adapting to contracts with diverse styles.

[0047] In this embodiment, a character recognition model is used to identify characters in the contract image of the target contract, obtaining the contract text. The character recognition model focuses on extracting text content from the image, providing a relatively accurate and complete text foundation. This provides additional text input for the contract element extraction model, avoiding the illusion problem that might arise from misleading image features when directly using the model. Simultaneously, this contract text serves as a crucial reference for subsequent verification of the accuracy of extracted elements. After the contract element extraction model extracts the contract elements based on the contract image, contract text, and element extraction prompts, these elements are verified again using the previously extracted contract text to determine whether the extracted elements match the original document content. This verification process... The system can promptly identify and correct potential illusions in the contract element extraction model, ensuring that the contract elements ultimately entered into the business system are highly consistent with the original contract content, thereby improving the accuracy and reliability of data entry. Specifically, it provides additional text input to the model in the early stages and rigorously verifies the extracted elements later, providing dual protection to ensure a high degree of consistency between the extracted contract elements and the original contract content, effectively avoiding erroneous entry caused by model illusions. Utilizing character recognition and contract element extraction models, it achieves automatic identification, extraction, and entry of contract elements from contract images, greatly reducing the workload and time required for manual identification and entry, improving business processing efficiency, and better adapting to the rapid online development needs of banking services.

[0048] The following examples further illustrate the contract element input method provided by the embodiments of the present invention, such as... Figure 2 As shown, Figure 2This is another flowchart illustrating the contract element entry method provided in this embodiment of the invention. The contract element entry method in this embodiment may include:

[0049] Step 201: Use a character recognition model to identify the characters in the contract image of the target contract to obtain the contract text of the target contract.

[0050] Specifically, a character recognition model can be used to identify characters in the contract image of the target contract to obtain the original text of the target contract; the original text of the target contract can be corrected to obtain the intermediate text of the target contract; and layout structure information can be added to the intermediate text of the target contract to obtain the contract text of the target contract.

[0051] The original text is the text content directly identified from the contract image by a character recognition model, and may contain some recognition errors or formatting issues. The process of correcting errors in the original text identified by the character recognition model can include correcting typos, punctuation errors, and formatting issues to improve text quality. The intermediate text is the text after error correction, and is more accurate and standardized than the original text.

[0052] Page layout information refers to the arrangement and structure of text within the original contract image, which may include heading levels, character position, font, size, color, etc. Adding page layout information can help the contract element extraction model locate key areas.

[0053] By correcting text errors and adding layout structure information, the final generated contract text is not only accurate in content, but also retains the original document's layout and formatting information, providing high-quality input for subsequent contract element extraction and verification.

[0054] Step 202: Input the contract image, contract text, and element extraction prompts of the target contract into the contract element extraction model, so as to use the contract element extraction model to extract the contract elements of the target contract from the contract image based on the contract text and element extraction prompts.

[0055] like Figure 3a As shown, Figure 3a This is an example diagram of the contract element extraction method provided in this embodiment of the invention. It can be seen that the input to the contract element extraction model includes the contract text output by the character recognition model, the original contract image, and prompt words. The contract element extraction model combines this information to extract key elements from the contract. These extracted contract elements include, for example, element 1, element 2, element 3, etc.

[0056] In this embodiment, when using a character recognition model to recognize characters in the contract image of the target contract, the position of each character in the contract image can be further recognized, and the contract text of the target contract output can include the characters and the position information of each character in the contract image of the target contract. The position information refers to the specific coordinates and positions of each character in the contract image, and can include information such as the starting coordinates, width, and height of the character. For example, the position of the character "borrow" in the contract image can be expressed as (x = 100, y = 200, width = 20, height = 20), indicating that the character is located in a specific area of the image. Such detailed position information provides richer context for subsequent contract element extraction and verification, helping to improve the accuracy and reliability of element extraction.

[0057] In a specific embodiment, the prompt words can be improved according to the contract text of the target contract. That is, the original prompt words of the target contract can be obtained. The ways to obtain the original prompt words of the target contract can include: obtaining the prompt words input by the user for the element extraction of the target contract to obtain the original prompt words of the target contract; or identifying the type of the target contract to obtain the target type, and obtaining the prompt words preset for the element extraction of the contract of the target type to obtain the original prompt words of the target contract. Query the contract text of the target contract based on the original prompt words to determine the positioning information of the contract elements of the target contract according to the position information of each character in the contract image of the target contract; add the positioning information of the contract elements of the target contract to the original prompt words of the target contract to obtain the prompt words for the element extraction of the target contract.

[0058] Original prompts are keywords, phrases, or sentences used to guide the model in extracting specific contract elements, helping the model locate and extract the required information. In credit transactions, original prompts might include phrases like "loan amount," "interest rate," and "repayment period," helping the model extract corresponding values ​​and information from the contract text. Original prompts can be input by the user in real-time or preset. The location information of contract elements refers to the specific position of contract elements within the contract image, which may include starting coordinates, width, and height. For example, the position of the contract element "loan amount: 1 million yuan" in the contract image can be represented as (x=100, y=200, width=120, height=20), indicating that the element is located in a specific area of ​​the image. The element extraction prompts here refer to enhanced prompts generated based on the original prompts and combined with the location information of the contract elements. These prompts not only contain text content but also location information, helping the model extract contract elements more accurately. For example, the enhanced prompt could be: "Retrieves loan amount, location (x=100, y=200, width=120, height=20)". With this improved prompt, the contract element extraction model can more accurately locate and extract key elements from the contract, thereby improving the accuracy and efficiency of the extraction.

[0059] In one specific embodiment, the contract image can also be enhanced based on the contract text of the target contract. Specifically, the contract text of the target contract can be queried based on element extraction prompts, and the location information of the contract elements of the target contract can be determined based on the position information of each character in the contract image. Based on the location information of the contract elements, location markers are added to the contract image of the target contract to obtain an enhanced image of the target contract. The enhanced image of the target contract, the contract text, and the element extraction prompts are then input into the contract element extraction model.

[0060] Location markers are visual markers used to identify the positions of contract elements in a contract image, providing specific location information. Location markers can be arrows, rectangular borders, underlines, or other marking lines. An augmented image is an image created by adding location markers to the original contract image. These markers help the model locate the contract elements more accurately. For example, a yellow border can be added to "Loan Amount: 1 million yuan" in the contract image to highlight its position. Using the contract image with added location markers as the augmented image means that the augmented image not only contains the original image content but also the location markers of the contract elements, helping the model to more accurately locate and extract these elements.

[0061] Step 203: Use the contract elements of the target contract to query the contract text of the target contract and obtain the hit string.

[0062] A hit string refers to the text fragment in the contract text that matches the current contract element; that is, the corresponding content found in the contract text by searching for the current contract element. For example, if the current contract element is "Loan amount: 1 million yuan", then "Loan amount: 1 million yuan" found in the contract text is the hit string.

[0063] Step 204: Calculate the similarity between the contract elements of the target contract and the matched string.

[0064] Similarity calculation can be achieved through various methods, such as edit distance and cosine similarity.

[0065] Step 205: Determine whether the similarity exceeds the preset similarity threshold. If it does, proceed to step 208; if it does not, proceed to step 206.

[0066] If the similarity exceeds the threshold, the extracted contract elements are considered to be consistent with the content in the contract text. If the similarity does not exceed the threshold, the extracted contract elements are considered to be potentially inaccurate or to contain illusions.

[0067] Step 206: Determine if the corresponding contract element verification fails.

[0068] If the similarity does not exceed a preset threshold, the contract element will be deemed unqualified. This means that the extracted contract element is inconsistent with the content in the contract text and requires manual review or correction.

[0069] Step 207: Fill in the blanks for the element items corresponding to the target contract in the target business system.

[0070] For contract elements that fail verification, blank values ​​can be entered or the element can be marked as suspicious in the corresponding element field of the target business system (such as a credit management system). This ensures that inaccurate information is not stored in the business system and alerts relevant personnel that further processing is required.

[0071] Step 208: Confirm that the corresponding contract elements have passed verification.

[0072] If the similarity exceeds a preset threshold, the system will determine that the contract element has passed verification. This means that the extracted contract element is consistent with the content in the contract text and can be considered accurate.

[0073] Step 209: Enter the contract elements into the element items corresponding to the target contract in the target business system.

[0074] For contract elements that pass verification, they can be entered into the corresponding element field in the target business system. This ensures that the information recorded in the business system is accurate and complete, and can be used for subsequent business processing and management. Of course, for contract elements that pass verification, further manual review can be conducted. Only after passing the manual review can they be entered into the corresponding element field in the target business system to further ensure the accuracy of the entered information.

[0075] For example, such as Figure 3b As shown, Figure 3b This is an example diagram of the contract element verification method provided in this embodiment of the invention. Assuming M contract elements are extracted, each extracted contract element can be compared with the contract text recognized by the character recognition model to determine the similarity between the extracted contract element and the corresponding content in the text. If the similarity is greater than a preset threshold of 0.8, an "adopt" operation is performed, and the element is entered into the system. If the similarity is not greater than the preset threshold of 0.8, a "fill in the blank" operation is performed to fill in the blanks. The similarity judgment determines whether an element is entered into the system, ensuring the accuracy of data entry.

[0076] In this embodiment, a character recognition model is used to identify characters in the contract image of the target contract, obtaining the contract text. The character recognition model focuses on extracting text content from the image, providing a relatively accurate and complete text foundation. This provides additional text input for the contract element extraction model, avoiding the illusion problem that might arise from misleading image features when directly using the model. Simultaneously, this contract text serves as a crucial reference for subsequent verification of the accuracy of extracted elements. After the contract element extraction model extracts the contract elements based on the contract image, contract text, and element extraction prompts, these elements are verified again using the previously extracted contract text to determine whether the extracted elements match the original document content. This verification process... The system can promptly identify and correct potential illusions in the contract element extraction model, ensuring that the contract elements ultimately entered into the business system are highly consistent with the original contract content, thereby improving the accuracy and reliability of data entry. Specifically, it provides additional text input to the model in the early stages and rigorously verifies the extracted elements later, providing dual protection to ensure a high degree of consistency between the extracted contract elements and the original contract content, effectively avoiding erroneous entry caused by model illusions. Utilizing character recognition and contract element extraction models, it achieves automatic identification, extraction, and entry of contract elements from contract images, greatly reducing the workload and time required for manual identification and entry, improving business processing efficiency, and better adapting to the rapid online development needs of banking services.

[0077] Figure 4 This is a schematic diagram of a contract element input device provided in an embodiment of the present invention, as shown below. Figure 4As shown, the device may include:

[0078] The character recognition module 401 is used to recognize characters in the contract image of the target contract using a character recognition model, and obtain the contract text of the target contract.

[0079] The element extraction module 402 is used to input the contract image, contract text and element extraction prompts of the target contract into the contract element extraction model, so as to use the contract element extraction model to extract the contract elements of the target contract from the contract image based on the contract text and element extraction prompts.

[0080] The verification module 403 is used to verify the contract elements of the target contract extracted by the contract element extraction model using the contract text of the target contract.

[0081] The data entry module 404 is used to enter the contract elements of the target contract into the target business system based on the verification results.

[0082] In one embodiment, the verification module 403 verifies the contract elements of the target contract extracted by the contract element extraction model using the contract text of the target contract, including:

[0083] By using the contract elements of the target contract to query the contract text of the target contract, the matching string can be obtained;

[0084] Calculate the similarity between the contract elements of the target contract and the matched strings;

[0085] If the similarity exceeds a preset similarity threshold, the corresponding contract element is deemed to have passed verification.

[0086] If the similarity does not exceed the preset similarity threshold, the corresponding contract element is determined to have failed verification.

[0087] In one embodiment, the data entry module 404 enters the contract elements of the target contract into the target business system based on the verification results, including:

[0088] Enter the verified contract elements into the corresponding element items of the target contract in the target business system;

[0089] Fill in the blanks for the corresponding element items of the target contract in the target business system for the contract elements that fail the verification.

[0090] In one embodiment, the character recognition module 401 uses a character recognition model to recognize characters in the contract image of the target contract to obtain the contract text of the target contract, including:

[0091] The character recognition model is used to identify characters in the contract image of the target contract to obtain the original text of the target contract;

[0092] The original text of the target contract is corrected to obtain the intermediate text of the target contract;

[0093] Add layout structure information to the intermediate text of the target contract to obtain the contract text of the target contract.

[0094] In one embodiment, the contract text of the target contract includes characters and the position information of each character in the contract image of the target contract. The device further includes a prompt word acquisition module, which is used to:

[0095] Obtain the original prompts for the target contract;

[0096] The contract text of the target contract is queried based on the original prompt words, so as to determine the location information of the contract elements of the target contract according to the position information of each character in the contract image of the target contract;

[0097] By adding the location information of the contract elements of the target contract to the original prompts of the target contract, the element extraction prompts of the target contract are obtained.

[0098] In one embodiment, the prompt word acquisition module acquires the original prompt words of the target contract, including:

[0099] Obtain the prompt words input by the user for the elements of the target contract, and obtain the original prompt words of the target contract;

[0100] Alternatively, identify the type of the target contract, obtain the target type, acquire preset prompts for extracting elements of the target type contract, and obtain the original prompts for the target contract.

[0101] In one embodiment, the contract text of the target contract includes characters and the position information of each character in the contract image of the target contract. The element extraction module 402 inputs the contract image, contract text, and element extraction prompts into the contract element extraction model, including:

[0102] Extract prompt words from the elements of the target contract to query the contract text of the target contract, and determine the location information of the contract elements of the target contract based on the position information of each character in the contract image of the target contract;

[0103] Based on the location information of the contract elements of the target contract, location markers are added to the contract elements on the contract image of the target contract to obtain an enhanced image of the target contract;

[0104] Input the enhanced image of the target contract, the contract text, and the element extraction prompts into the contract element extraction model.

[0105] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is merely an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the functional modules described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0106] The apparatus of this invention uses a character recognition model to recognize characters in a contract image of a target contract, obtaining the contract text. The character recognition model focuses on extracting text content from the image, providing a relatively accurate and complete text foundation. This provides additional text input for the contract element extraction model, avoiding the illusion problem that might arise from misleading image features when directly using the model. Simultaneously, this contract text serves as a crucial reference for subsequent verification of the accuracy of extracted elements. After the contract element extraction model extracts contract elements based on the contract image, contract text, and element extraction prompts, it again uses the previously extracted contract text to verify these elements, determining whether the extracted elements match the original document content. This verification... The verification process can promptly identify and correct potential illusions in the contract element extraction model, ensuring that the contract elements ultimately entered into the business system are highly consistent with the original contract content, thereby improving the accuracy and reliability of data entry. Specifically, providing additional text input to the model in the early stages and rigorously verifying the extracted elements in the later stages provides dual protection, ensuring a high degree of consistency between the extracted contract elements and the original contract content, effectively avoiding erroneous entry caused by model illusions. Utilizing character recognition and contract element extraction models, the system automatically identifies and extracts contract elements from contract images and inputs them, significantly reducing the workload and time required for manual identification and entry, improving business processing efficiency, and better adapting to the rapid online development needs of banking services.

[0107] The following is for reference. Figure 5 It shows a schematic diagram of the structure of a computer system 500 suitable for implementing an electronic device according to embodiments of the present invention. Figure 5 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0108] like Figure 5As shown, the computer system 500 includes a central processing unit (CPU) 501, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 502 or programs loaded from storage section 508 into random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of the computer system 500. The CPU 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0109] The following components are connected to I / O interface 505: input section 506 including keyboard, mouse, etc.; output section 507 including cathode ray tube, liquid crystal display, etc., and speakers, etc.; storage section 508 including hard disk, etc.; and communication section 509 including network interface card, such as modem, etc. Communication section 509 performs communication processing via a network such as the Internet. Drive 510 is also connected to I / O interface 505 as needed. Removable media 511, such as disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 510 as needed so that computer programs read from them can be installed into storage section 508 as needed.

[0110] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 509, and / or installed from removable medium 511. When the computer program is executed by central processing unit (CPU) 501, it performs the functions defined above in the system of this invention.

[0111] It should be noted that the computer-readable medium shown in this invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical fiber, a portable compact disk read-only memory, an optical storage device, a magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, etc., or any suitable combination thereof.

[0112] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0113] The modules and / or units described in the embodiments of the present invention can be implemented in software or hardware. The described modules and / or units can also be housed in a processor; for example, a processor can be described as including a character recognition module, an element extraction module, a verification module, and an input module. The names of these modules do not necessarily limit the functionality of the module itself.

[0114] In another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiments; or it may exist independently and not assembled into the device. The computer-readable medium carries one or more programs, which, when executed by the device, cause the device to include:

[0115] The character recognition model is used to identify characters in the contract image of the target contract to obtain the contract text. The contract image, contract text, and element extraction prompts of the target contract are input into the contract element extraction model to extract the contract elements of the target contract from the contract image based on the contract text and element extraction prompts. The contract text of the target contract is used to verify the contract elements of the target contract extracted by the contract element extraction model. Based on the verification results, the contract elements of the target contract are entered into the target business system.

[0116] The technical solution of this invention uses a character recognition model to recognize characters in the contract image of a target contract, obtaining the contract text. The character recognition model focuses on extracting text content from the image, providing a relatively accurate and complete text foundation. This provides additional text input for the contract element extraction model, avoiding the illusion problem that might occur when directly using the model due to image feature misleading information. Simultaneously, this contract text serves as a key reference for subsequent verification of the accuracy of extracted elements. After the contract element extraction model extracts contract elements based on the contract image, contract text, and element extraction prompts, it again uses the previously extracted contract text to verify these elements, determining whether the extracted elements match the original document content. The verification process can promptly identify and correct potential illusions in the contract element extraction model, ensuring that the contract elements ultimately entered into the business system are highly consistent with the original contract content, thereby improving the accuracy and reliability of data entry. Specifically, providing additional text input to the model in the early stages and rigorously verifying the extracted elements in the later stages provides dual protection, ensuring a high degree of consistency between the extracted contract elements and the original contract content, effectively avoiding erroneous entry caused by model illusions. Utilizing character recognition and contract element extraction models, the system automatically identifies and extracts contract elements from contract images and inputs them, significantly reducing the workload and time required for manual identification and entry, improving business processing efficiency, and better adapting to the rapid online development needs of banking services.

[0117] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the contract element input method as provided in any embodiment of this invention.

[0118] In the implementation of a computer program product, computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​as well as conventional procedural programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including local area networks (LANs) or wide area networks (WANs), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0119] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0120] It should be noted that the collection, use, storage, sharing, and transfer of user personal information involved in the technical solution of this invention all comply with the provisions of relevant laws and regulations, and require notification to the user and obtaining the user's consent or authorization. Where applicable, user personal information has undergone de-identification and / or anonymization and / or encryption technical processing. In addition, a corresponding operation entry is provided for the user to choose to agree to or reject the automated decision result; if the user chooses to reject, the process proceeds to the expert decision-making process.

[0121] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for entering contract elements, characterized in that, include: The character recognition model is used to identify the characters in the contract image of the target contract to obtain the contract text of the target contract; Input the contract image, contract text, and element extraction prompts of the target contract into the contract element extraction model, so as to use the contract element extraction model to extract the contract elements of the target contract from the contract image based on the contract text and element extraction prompts; The contract elements extracted by the contract element extraction model are verified using the contract text of the target contract. Based on the verification results, the contract elements of the target contract are entered into the target business system.

2. The method according to claim 1, characterized in that, The contract elements extracted by the contract element extraction model are validated using the contract text of the target contract, including: By using the contract elements of the target contract to query the contract text of the target contract, the matching string can be obtained; Calculate the similarity between the contract elements of the target contract and the matched strings; If the similarity exceeds a preset similarity threshold, the corresponding contract element is deemed to have passed verification. If the similarity does not exceed the preset similarity threshold, the corresponding contract element is determined to have failed verification.

3. The method according to claim 1 or 2, characterized in that, Based on the verification results, the contract elements of the target contract are entered into the target business system, including: Enter the verified contract elements into the corresponding element items of the target contract in the target business system; Fill in the blanks for the corresponding element items of the target contract in the target business system for the contract elements that fail the verification.

4. The method according to claim 1, characterized in that, The character recognition model is used to identify characters in the contract image of the target contract, thereby obtaining the contract text, including: The character recognition model is used to identify characters in the contract image of the target contract to obtain the original text of the target contract; The original text of the target contract is corrected to obtain the intermediate text of the target contract; Add layout structure information to the intermediate text of the target contract to obtain the contract text of the target contract.

5. The method according to claim 1, characterized in that, The target contract text includes characters and the position information of each character in the contract image. Feature extraction prompts are obtained using the following method: Obtain the original prompts for the target contract; The contract text of the target contract is queried based on the original prompt words, so as to determine the location information of the contract elements of the target contract according to the position information of each character in the contract image of the target contract; By adding the location information of the contract elements of the target contract to the original prompts of the target contract, the element extraction prompts of the target contract are obtained.

6. The method according to claim 5, characterized in that, Obtain the original prompts for the target contract, including: Obtain the prompt words input by the user for the elements of the target contract, and obtain the original prompt words of the target contract; Alternatively, identify the type of the target contract, obtain the target type, acquire preset prompts for extracting elements of the target type contract, and obtain the original prompts for the target contract.

7. The method according to claim 1, characterized in that, The target contract text includes characters and the position information of each character within the contract image. The contract image, contract text, and feature extraction prompts are input into the contract feature extraction model, including: Extract prompt words from the elements of the target contract to query the contract text of the target contract, and determine the location information of the contract elements of the target contract based on the position information of each character in the contract image of the target contract; Based on the location information of the contract elements of the target contract, location markers are added to the contract elements on the contract image of the target contract to obtain an enhanced image of the target contract; Input the enhanced image of the target contract, the contract text, and the element extraction prompts into the contract element extraction model.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the contract element entry method as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the contract element entry method as described in any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the contract element entry method as described in any one of claims 1 to 7.

Images