Contract auditing method, device and equipment and computer readable storage medium

Abstract

The invention discloses a contract auditing method, apparatus and device, and a computer readable storage medium. The method comprises the steps of obtaining an image file corresponding to a contract file; inputting the image file into a multi-modal analysis model for model identification so as to obtain text information of each text area corresponding to the image file; performing screening operation on the text information based on a preset keyword to obtain target text information; and auditing the target text information by adopting a large language model based on an entity recognition rule to obtain an auditing result. According to the method, the text information is screened, the target text information is audited on the basis of the large language model, intelligent audit of the contract file is achieved, meaningless input of the large model and the data annotation process can be reduced as much as possible through keyword screening and entity recognition rule matching, and the user experience is improved. The efficiency and accuracy of contract auditing are improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a contract review method, device, equipment and computer-readable storage medium. Background Art

[0002] Reviewing contract information is a tedious and time-consuming task. The key information of the document is usually scattered in different chapters. Manual review of the contract usually takes a lot of time to find data.

[0003] At present, the manual rule-based review method requires a lot of expert knowledge to analyze and collect documents in related fields, summarize the patterns and rules in the field, and use regular expressions or other pattern matching methods to extract key information. Although relevant information can be extracted according to the pre-set rule expression method, it is impossible to accurately extract complex text data, data with diverse structures, etc., and even if the rules are changed a little bit, the relevant entity information cannot be extracted, resulting in inaccurate data extraction and the inability to effectively and accurately review the contract.

[0004] Therefore, how to improve the efficiency and accuracy of contract review is a problem that needs to be solved urgently. Summary of the invention

[0005] The main purpose of this application is to provide a contract review method, device, equipment and computer-readable storage medium, aiming to solve the technical problem of how to improve the efficiency and accuracy of contract review.

[0006] To achieve the above purpose, the present application provides a contract review method, which includes the following steps:

[0007] Obtain the image file corresponding to the contract document;

[0008] Inputting the image file into a multimodal parsing model for model recognition to obtain text information corresponding to each text area of ​​the image file;

[0009] Performing a screening operation on the text information based on preset keywords to obtain target text information;

[0010] The target text information is audited using a large language model based on entity recognition rules to obtain an audit result.

[0011] Furthermore, the step of inputting the image file into a multimodal analysis model for model recognition to obtain text information corresponding to each text area of ​​the image file includes:

[0012] Performing text recognition on the image file to obtain initial text information corresponding to each text area in the image file;

[0013] The initial text information and the image file are input into a multimodal analysis model for model recognition to obtain text information of each text area.

[0014] Furthermore, the step of performing text recognition on the image file to obtain initial text information corresponding to each text area in the image file includes:

[0015] Performing text recognition on the image file to obtain each text area in the image file;

[0016] Perform text recognition on each text area respectively to obtain initial text information corresponding to each text area.

[0017] Furthermore, the step of filtering the text information based on preset keywords to obtain target text information includes:

[0018] The text information is screened based on the preset keyword to obtain a text screening result that matches the preset keyword in the text information, and the text screening result is used as the target text information.

[0019] Furthermore, the step of reviewing the target text information using a large language model based on entity recognition rules to obtain a review result includes:

[0020] Using a large language model to perform entity extraction on the target text information based on entity recognition rules to obtain an entity recognition result;

[0021] A large language model is used to review the entity recognition result to obtain a review result.

[0022] Furthermore, the entity recognition result at least includes the contract date and the contract amount, and the step of using the large language model to review the entity recognition result and obtain the review result includes:

[0023] A large language model is used to perform a completeness audit on the contract date in the entity recognition result, and to perform an accuracy audit on the contract amount in the entity recognition result, so as to obtain the audit result.

[0024] Furthermore, the step of obtaining the image file corresponding to the contract document includes:

[0025] Based on the file format of the contract document, converting the contract document into a contract image;

[0026] The contract image is subjected to denoising, binarization and tilt correction processing in sequence to obtain the image file.

[0027] In addition, to achieve the above-mentioned purpose, the present application also provides a contract review device, an acquisition module, for acquiring an image file corresponding to a contract document;

[0028] A recognition module, used for inputting the image file into a multimodal analysis model for model recognition to obtain text information corresponding to each text area of ​​the image file;

[0029] A screening module, used to screen the text information based on preset keywords to obtain target text information;

[0030] The audit module is used to audit the target text information using a large language model based on entity recognition rules to obtain an audit result.

[0031] In addition, to achieve the above-mentioned purpose, the present application also provides a contract review device, which includes: a memory, a processor, and a contract review program stored on the memory and executable on the processor, and the contract review program implements the steps of the aforementioned contract review method when executed by the processor.

[0032] In addition, to achieve the above-mentioned purpose, the present application also provides a computer-readable storage medium, on which a contract review program is stored, and when the contract review program is executed by a processor, the steps of the aforementioned contract review method are implemented.

[0033] The present application obtains an image file corresponding to a contract document; then inputs the image file into a multimodal parsing model for model recognition to obtain text information corresponding to each text area of ​​the image file; then screens the text information based on preset keywords to obtain target text information; then uses a large language model to review the target text information based on entity recognition rules to obtain a review result, identifies the text information of the contract through a multimodal parsing model, screens the text information, and reviews the target text information based on a large language model to achieve intelligent review of the contract document. Through keyword screening and entity recognition rule matching, the meaningless input of the large model and the data labeling process can be minimized, thereby improving the efficiency and accuracy of contract review. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0035] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0036] Figure 1 A flowchart of the first embodiment of the contract review method of this application is provided;

[0037] Figure 2 This is a schematic diagram of the network structure of LoRA in the embodiment of the present application;

[0038] Figure 3 This is a schematic diagram of the module structure of the contract review device according to an embodiment of the present application;

[0039] Figure 4 This is a schematic diagram of the module structure of the contract review device of the embodiment of the present application.

[0040] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0041] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0042] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0043] The main solution of this application is: obtain the image file corresponding to the contract document; input the image file into the multimodal parsing model for text recognition to obtain the text information of each text area corresponding to the image file; filter the text information based on preset keywords to obtain target text information; and review the target text information using a large language model based on entity recognition rules to obtain the review result.

[0044] Currently, reviewing contract information is a tedious and time-consuming task. The key information of the document is usually scattered in different chapters, and manual review of the contract usually takes a lot of time to find data.

[0045] The manual rule-based review method requires a lot of expert knowledge to analyze and collect documents in related fields, summarize the patterns and rules in the field, and use regular expressions or other pattern matching methods to extract key information. Although relevant information can be extracted according to the pre-set rule expression method, it is impossible to accurately extract complex text data, data with diverse structures, etc. Even if the rules are changed a little bit, the relevant entity information cannot be extracted, resulting in inaccurate data extraction and the inability to effectively and accurately review the contract. Therefore, how to improve the efficiency and accuracy of contract review is a problem that needs to be solved urgently.

[0046] This application uses a multimodal parsing model to identify the text information of the contract, screens the text information, and reviews the target text information based on a large language model to achieve intelligent review of the contract document. Through keyword screening and entity recognition rule matching, it can minimize the meaningless input and data labeling process of the large model, thereby improving the efficiency and accuracy of contract review.

[0047] It should be noted that the execution subject of this embodiment can be a contract review device, or a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or a contract review device capable of realizing the above functions, etc. This embodiment does not specifically limit this. The following takes the contract review device as the execution subject as an example to illustrate this embodiment and the following embodiments.

[0048] Based on this, this application proposes a first embodiment of the contract review method, please refer to Figure 1 The contract review method includes steps S101 to S104:

[0049] Step S101, obtaining an image file corresponding to the contract document;

[0050] In this embodiment, the contract provided by the user can be in multiple formats, including PDF contract, word format, PDF of contract image, etc. The input contract file is converted into an image file, and then it can support file input in multiple formats to obtain the image file corresponding to the contract file.

[0051] In a feasible implementation manner, step S101 may further include steps S1011 to S1012:

[0052] Step S1011, converting the contract document into a contract image based on the file format of the contract document;

[0053] Step S1012, performing denoising, binarization and tilt correction processing on the contract image in sequence to obtain the image file.

[0054] In this embodiment, the file format of the input contract file is first obtained, and the contract file is converted into a contract image based on the file format of the contract file. After the contract image is obtained, the contract image is sequentially subjected to denoising, binarization and tilt correction processing to obtain an image file. Specifically, the contract image is first subjected to denoising processing to obtain a denoised image file, and then the denoised image file is binarized to obtain a binarized image file. Finally, the binarized image file is tilt corrected to obtain an image file. By performing denoising, binarization and tilt correction processing on the contract image, the accuracy of file content parsing is improved, thereby improving the efficiency and accuracy of contract review.

[0055] Step S102, inputting the image file into a multimodal analysis model for text recognition to obtain text information corresponding to each text area of ​​the image file;

[0056] In this embodiment, when an image file is obtained, the image file is input into the multimodal analysis model for text recognition to obtain text information corresponding to each text area of ​​the image file. Specifically, in a feasible implementation, step S102 may also include steps S1021 to S1022:

[0057] Step S1021, performing text recognition on the image file to obtain initial text information corresponding to each text area in the image file;

[0058] Step S1022: input the initial text information and the image file into a multimodal analysis model for model recognition to obtain text information of each text area.

[0059] In this embodiment, when an image file is obtained, text recognition is performed on the image file to obtain initial text information corresponding to each text area in the image file. Specifically, an AI model in the relevant technology can be used to perform text recognition on the image file. Further, in a feasible implementation, step S1021 can also include steps a1 to a2:

[0060] Step a1, performing text recognition on the image file to obtain each text area in the image file;

[0061] Step a2: performing text recognition on each text area respectively to obtain initial text information corresponding to each text area.

[0062] In this embodiment, when an image file is obtained, text recognition is first performed on the image file to obtain various text areas in the image file. Specifically, EAST can be used to perform text recognition on the image file to detect the text areas (text blocks) in the image file to obtain the area coordinates of each text area.

[0063] After determining each text area in the image file, perform text recognition on each text area to obtain the initial text information corresponding to each text area. Specifically, for each text area, the OCR recognition engine Tesseract can be used for text recognition to obtain the text information corresponding to the text area, and the preliminary text information in the image file can be obtained. At the same time, the position information of each character in each text area can also be obtained. By locating and identifying text information in the image file, the meaninglessness of the large model can be reduced, thereby improving the efficiency and accuracy of contract review.

[0064] In this embodiment, after the initial text information of each text area is obtained, the initial text information and the image file are input into the multimodal parsing model for model recognition to obtain the text information of each text area. Among them, the multimodal large model may include Qwen-VL-Max. Qwen-VL-Max is a model designed for document understanding. It combines visual and language features and can process document images and text information therein at the same time, thereby achieving a comprehensive understanding of document images. Qwen-VL-Max fine-tunes the model through the initial text information and the original image (image file) obtained by OCR, learns the correspondence between the image file and the initial text information, optimizes the initial text information to obtain text information, improves the accuracy of the text information, and thereby improves the efficiency and accuracy of contract review.

[0065] Step S103, filtering the text information based on preset keywords to obtain target text information;

[0066] In this embodiment, considering the length of the contract, some contracts may be hundreds of pages long, but the entities to be extracted do not have relevant information on every page of the contract, but only on certain pages of the contract. If the entire contract is extracted using a large language model, it will not only fail to improve the accuracy of the extraction, but will also increase the time of the extraction and reduce the efficiency of the extraction. Therefore, after obtaining the text information of each text area, the text information is screened based on the preset keywords to obtain the target text information, wherein some rules (preset keywords) can be pre-set based on expert knowledge and experience, such as: extracting a bank account number, setting the preset keyword corresponding rule as: account number / bank account number / account; extracting the signing date, setting the preset keyword corresponding rule as: date / time; using regular expressions, that is, the preset keyword can be the keyword of each entity in the contract, searching for the preset keyword in each text information, screening the text information related to the preset keyword, removing the contract page number (text information) without a matching entity, and obtaining the target text information. The text information is screened by the preset keywords to reduce the amount of text information that needs to be identified, thereby improving the efficiency and accuracy of contract review.

[0067] In a feasible implementation manner, step S103 may further include step S1031:

[0068] Step S1031 , performing a screening operation on the text information based on the preset keyword to obtain a text screening result in the text information that matches the preset keyword, and using the text screening result as the target text information.

[0069] In this embodiment, after obtaining the text information of each text area, the text information is filtered based on the preset keywords to obtain text filtering results that match the preset keywords in the text information. Specifically, each text information can be searched through each preset keyword to obtain text information including the preset keywords, i.e., text filtering results, and the text filtering results are used as target text information, and then the text including the preset keywords is filtered out from the text information, and the text that does not match the preset keywords is removed. The text information is filtered through the preset keywords to reduce the amount of text information that needs to be identified, thereby improving the efficiency and accuracy of contract review.

[0070] Step S104: review the target text information using a large language model based on entity recognition rules to obtain a review result.

[0071] In this embodiment, after the target text information is obtained, the target text information is reviewed using a large language model based on entity recognition rules to obtain a review result. Specifically, in a feasible implementation, step S104 may also include steps S1041 to S1042:

[0072] Step S1041, using a large language model to perform entity extraction on the target text information based on entity recognition rules to obtain an entity recognition result;

[0073] Step S1042: Use a large language model to review the entity recognition result to obtain a review result.

[0074] Among them, the large language model can be the Tongyi Qianwen large model Qwen2-1.5B. Entity extraction is a sequence labeling task. Supervised learning is performed through an annotated entity extraction data set to train the Tongyi Qianwen large model Qwen2-1.5B so that the Tongyi Qianwen large model Qwen2-1.5B can adapt to the needs of specific tasks and obtain a trained large language model.

[0075] In this embodiment, after acquiring the target text information, the target text information is input into a large language model, and the large language model is used to perform entity extraction on the target text information based on entity recognition rules to obtain an entity recognition result. Specifically, after the target text information is input into the large language model, the large language model converts the target text information into a vector representation, and through a multi-layer attention mechanism, captures the semantic information in the target text information, identifies and classifies entities, and obtains an entity recognition result.

[0076] After obtaining the entity recognition result, the entity recognition result is reviewed using a large language model to obtain a review result. Specifically, in a feasible implementation, the entity recognition result at least includes a contract date and a contract amount. Step S1042 may further include step b:

[0077] Step b: using a large language model to perform integrity review on the contract date in the entity recognition result, and to perform accuracy review on the contract amount in the entity recognition result, so as to obtain the review result.

[0078] In this embodiment, after obtaining the entity recognition result, if the entity recognition result includes at least the contract date and the contract amount, the large language model performs an integrity review on the contract date in the entity recognition result and an accuracy review on the contract amount in the entity recognition result to obtain a review result. Specifically, the review rules in the large language model are preset, including the review rule for the integrity of the contract date and the review rule for the accuracy of the contract amount. Then, the Prompt preset by the Qwen large language model is used to automatically review the entity recognition result to determine whether there are errors in the data format, the expression logic of the text, etc., and a review result is obtained.

[0079] Both the fine-tuning of the Qwen-VL-Max model and the Qwen2-1.5B model adopt LoRA. LoRA is a technology for fine-tuning large-scale pre-trained models. The core idea is to effectively adapt and fine-tune large models through the method of low-rank matrix approximation. During the fine-tuning process, instead of directly updating a large number of parameters of the original model, it approximates the changes by learning low-rank matrices, thereby reducing the computational and storage requirements. As Figure 2 shown, Figure 2 is the schematic diagram of the network structure of LoRA.

[0080] Specifically, a weight matrix is selected for low-rank decomposition, and the calculation formula is: ΔW = BA = d * r + r * d;

[0081] where d is the dimension of the pre-trained layer and the trainable layer of the model. First, the dimension d is reduced to r through a fully connected layer, and then mapped back to the d dimension from r through a fully connected layer, where r << d. In this way, the number of model parameters that need to be adjusted is much lower than that of the original model. According to the above description, the calculation formula for the entire network layer is: h = W0x + ΔWx;

[0082] During initialization, the parameters of matrix A are initialized through a Gaussian function, and the weight parameters of the second matrix B are initialized as a zero matrix to ensure that during the initial training, the newly added path BA = 0 during forward propagation will not affect the model result.

[0083] The contract review method proposed in this embodiment obtains an image file corresponding to a contract document; then inputs the image file into a multimodal parsing model for model recognition to obtain text information corresponding to each text area of ​​the image file; then screens the text information based on preset keywords to obtain target text information; then uses a large language model based on entity recognition rules to review the target text information to obtain a review result, identifies the text information of the contract through a multimodal parsing model, screens the text information, and reviews the target text information based on a large language model to achieve intelligent review of the contract document, and can minimize meaningless input of the large model and data labeling process through keyword screening and entity recognition rule matching, thereby improving the efficiency and accuracy of contract review.

[0084] The present application also provides a contract review device, please refer to Figure 3 , the contract review device comprises:

[0085] An acquisition module 10 is used to acquire an image file corresponding to the contract document;

[0086] The recognition module 20 is used to input the image file into the multimodal analysis model for model recognition to obtain text information corresponding to each text area of ​​the image file;

[0087] A screening module 30, configured to screen the text information based on preset keywords to obtain target text information;

[0088] The audit module 40 is used to audit the target text information using a large language model based on entity recognition rules to obtain an audit result.

[0089] The contract review device provided in the embodiment of the present application adopts the contract review method in the above embodiment, which can solve the technical problem of how to improve the efficiency and accuracy of contract review. Compared with the prior art, the beneficial effects of the contract review device provided in the embodiment of the present application are the same as the beneficial effects of the contract review method provided in the above embodiment, and the other technical features in the contract review device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0090] The present application provides a contract review device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the contract review method in the above-mentioned embodiment one.

[0091] Reference below Figure 4, which shows a schematic diagram of the structure of a contract review device suitable for implementing the embodiment of the present application. The contract review device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 4 The contract review device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0092] like Figure 4 As shown, the contract review device may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the contract review device are also stored. The processing device 1001, ROM1002, and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the contract review device to communicate with other devices wirelessly or wired to exchange data. Although the figure shows a contract review device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0093] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0094] The contract review device provided by the present application adopts the contract review method in the above embodiment, which can solve the technical problem of how to improve the efficiency and accuracy of contract review. Compared with the prior art, the beneficial effects of the contract review device provided by the present application are the same as the beneficial effects of the contract review method provided by the above embodiment, and the other technical features in the contract review device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0095] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0096] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0097] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the contract review method in the above-mentioned embodiment.

[0098] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0099] The above-mentioned computer-readable storage medium may be included in the contract review device; or it may exist independently without being assembled into the contract review device.

[0100] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by the contract review device, the contract review device: obtains the image file corresponding to the contract document; inputs the image file into the multimodal parsing model for model recognition to obtain the text information of each text area corresponding to the image file; filters the text information based on preset keywords to obtain target text information; and uses a large language model to review the target text information based on entity recognition rules to obtain a review result.

[0101] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0102] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0103] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0104] The readable storage medium provided in this application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned contract review method, and can solve the technical problem of how to improve the efficiency and accuracy of contract review. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the contract review method provided in the above-mentioned embodiment, and will not be repeated here.

[0105] An embodiment of the present application provides a computer program product, including a computer program, which implements the steps of the above-mentioned contract review method when executed by a processor.

[0106] The computer program product provided in this application can solve the technical problem of how to improve the efficiency and accuracy of contract review. Compared with the prior art, the beneficial effects of the computer program product provided in the embodiment of this application are the same as the beneficial effects of the contract review method provided in the above embodiment, which will not be repeated here.

[0107] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A contract review method, characterized in that: The contract review method comprises the following steps: Obtain the image file corresponding to the contract document; Inputting the image file into a multimodal parsing model for model recognition to obtain text information corresponding to each text area of ​​the image file; Performing a screening operation on the text information based on preset keywords to obtain target text information; The target text information is audited using a large language model based on entity recognition rules to obtain an audit result.

2. The contract review method according to claim 1, characterized in that: The step of inputting the image file into a multimodal analysis model for model recognition to obtain text information corresponding to each text area of ​​the image file comprises: Performing text recognition on the image file to obtain initial text information corresponding to each text area in the image file; The initial text information and the image file are input into a multimodal analysis model for model recognition to obtain text information of each text area.

3. The contract review method according to claim 2, characterized in that: The step of performing text recognition on the image file to obtain initial text information corresponding to each text area in the image file includes: Performing text recognition on the image file to obtain each text area in the image file; Perform text recognition on each text area respectively to obtain initial text information corresponding to each text area.

4. The contract review method according to claim 1, characterized in that: The step of filtering the text information based on preset keywords to obtain target text information includes: The text information is screened based on the preset keyword to obtain a text screening result that matches the preset keyword in the text information, and the text screening result is used as the target text information.

5. The contract review method according to claim 1, characterized in that: The step of reviewing the target text information using a large language model based on entity recognition rules to obtain a review result includes: Using a large language model to perform entity extraction on the target text information based on entity recognition rules to obtain an entity recognition result; A large language model is used to review the entity recognition result to obtain a review result.

6. The contract review method according to claim 1, characterized in that: The entity recognition result at least includes the contract date and the contract amount. The step of using a large language model to review the entity recognition result and obtaining the review result includes: A large language model is used to perform a completeness audit on the contract date in the entity recognition result, and to perform an accuracy audit on the contract amount in the entity recognition result, so as to obtain the audit result.

7. The contract review method according to any one of claims 1 to 6, characterized in that: The step of obtaining the image file corresponding to the contract document comprises: Based on the file format of the contract document, converting the contract document into a contract image; The contract image is subjected to denoising, binarization and tilt correction processing in sequence to obtain the image file.

8. A contract review device, characterized in that: An acquisition module is used to acquire the image file corresponding to the contract document; A recognition module, used for inputting the image file into a multimodal analysis model for model recognition to obtain text information corresponding to each text area of ​​the image file; A screening module, used to screen the text information based on preset keywords to obtain target text information; The audit module is used to audit the target text information using a large language model based on entity recognition rules to obtain an audit result.

9. A contract review device, characterized in that: The contract review device includes: a memory, a processor, and a contract review program stored in the memory and executable on the processor. When the contract review program is executed by the processor, the steps of the contract review method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a contract review program, which, when executed by a processor, implements the steps of the contract review method according to any one of claims 1 to 7.

Images