Intelligent filling device and intelligent filling method for contract

Abstract

The invention discloses an intelligent filling device and method for a contract, and the device comprises a data input unit which is used for obtaining a contract uploaded by a user and a question input by the user; the OCR optical character recognition unit is used for performing text recognition on the contract to obtain text content of the contract; and the question and answer implementation unit is used for extracting a target field according to the question input by the user, obtaining the text content of the target field according to the text content of the contract, and intelligently filling the contract field corresponding to the contract management system according to the text content of the target field to obtain an intelligent filling result. Visibly, the data input unit sends the contract uploaded by the user to the OCR optical character recognition unit, the OCR optical character recognition unit extracts the text content in the picture, and then the question and answer implementation unit performs intelligent filling on the contract field corresponding to the contract management system according to the text content of the target field, so that intelligent filling of the contract content is realized; and the contract filling efficiency and accuracy are improved.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to an intelligent filling device and an intelligent filling method for a contract. Background Art

[0002] In today's digital age, contract management has become increasingly important, especially for transactions between businesses and individuals. A contract is a legal document agreed upon by two parties that records the terms, responsibilities, and rights of a transaction.

[0003] However, the creation and management of contracts is often a time-consuming and tedious task. Especially when filling out contracts digitally, people often need to process a large amount of text information and ensure that various details are entered accurately, which may involve a lot of manual operations and prone to errors, resulting in low efficiency and accuracy in contract filling. Summary of the invention

[0004] In view of this, the present application provides a smart contract filling device and a smart contract filling method, aiming to improve the efficiency and accuracy of contract filling.

[0005] In a first aspect, the present application provides a smart filling device for a contract, which is applied to a contract management system, and the device includes a data input unit, an OCR optical character recognition unit, and a question-answering implementation unit;

[0006] The data input unit is used to obtain the contract uploaded by the user and the question input by the user, wherein the contract is a digital image contract obtained by converting a paper contract through a scanning device or a camera device;

[0007] The OCR optical character recognition unit is used to perform text recognition on the contract to obtain the text content of the contract;

[0008] The question-answering implementation unit is used to extract the target field according to the question input by the user, and obtain the text content of the target field according to the text content of the contract;

[0009] The question-and-answer implementation unit is also used to intelligently fill in the contract field corresponding to the contract management system according to the text content of the target field to obtain an intelligent filling result.

[0010] Optionally, if the user does not input a question, the question and answer implementation unit is further used to obtain the text content of the default field based on the text content of the contract, and the default field includes the contract number, contract name, date, amount and contracting party information.

[0011] Optionally, the OCR optical character recognition unit includes an OCR optical character recognition model, and the OCR optical character recognition model is used to perform text recognition on the contract to obtain the text content of the contract.

[0012] Optionally, the training process of the OCR optical character recognition model includes:

[0013] Obtaining a training data set, wherein the training data set includes historical contracts and font files generated by an OCR optical character recognition engine tool;

[0014] Using the OCR optical character recognition engine tool to convert the format of the training data set to obtain a converted training data set;

[0015] The OCR optical character recognition model is trained using the converted training data set to obtain a trained OCR optical character recognition model.

[0016] Optionally, the question-answering implementation unit includes a question-answering big model, and the question-answering big model is used to extract the target field according to the question input by the user.

[0017] Optionally, the training process of the large question-answering model includes:

[0018] Acquire a training sample set, wherein the training sample set includes a target text and a historical question input by a user, wherein the target text is a text above a confidence threshold after the text content of the historical contract is cleaned according to the confidence level;

[0019] Performing text enhancement on the training sample set by using a hybrid text enhancement method for original text enhancement and for representation text enhancement to obtain an enhanced training sample set;

[0020] Using the enhanced training sample set, training the question-answering large model to obtain a trained question-answering large model;

[0021] The model parameters of the trained question-answering model are adjusted using evaluation indicators, wherein the evaluation indicators include accuracy, recall, and the harmonic mean of the accuracy and recall.

[0022] Optionally, the data input unit is further used to score the smart filling result to obtain a scoring result of the smart filling result.

[0023] Optionally, the device further comprises a data storage unit;

[0024] The data storage unit is used to store the smart filling result and the scoring result in a database.

[0025] Optionally, the data input unit includes a visual interface for guiding the user to perform operations and interactions.

[0026] In a second aspect, the present application provides a smart filling method for a contract, which is applied to the smart filling device described in the first aspect, and the method includes:

[0027] Obtaining a contract uploaded by a user and a question input by the user, wherein the contract is a digital image contract obtained by converting a paper contract through a scanning device or a camera device;

[0028] Performing text recognition on the contract to obtain the text content of the contract;

[0029] Extracting a target field according to the question input by the user, and acquiring text content of the target field according to the text content of the contract;

[0030] The contract field corresponding to the contract management system is intelligently filled according to the text content of the target field to obtain an intelligent filling result.

[0031] In a third aspect, an embodiment of the present application provides an electronic device, the electronic device comprising:

[0032] A memory for storing one or more programs;

[0033] Processor; when the one or more programs are executed by the processor, the intelligent filling method of the contract described in the second aspect above is implemented.

[0034] In a fourth aspect, an embodiment of the present application provides a computer storage medium, in which a program is stored. When the program is executed by a processor, the intelligent filling method of the contract described in the second aspect is implemented.

[0035] The above technical solution has the following beneficial effects:

[0036] The present application provides a smart filling device and method for a contract, wherein the smart filling device includes a data input unit, an OCR optical character recognition unit and a question-answering implementation unit; wherein the data input unit is used to obtain a contract uploaded by a user and a question input by the user, wherein the contract is a digital image contract obtained by converting a paper contract through a scanning device or a camera device; the OCR optical character recognition unit is used to perform text recognition on the contract and obtain the text content of the contract; the question-answering implementation unit is used to extract a target field according to the question input by the user and obtain the text content of the target field according to the text content of the contract; the question-answering implementation unit is also used to perform smart filling on the contract field corresponding to the contract management system according to the text content of the target field to obtain a smart filling result. Through the above technical solution, the data input unit sends the contract uploaded by the user to the OCR optical character recognition unit, the OCR optical character recognition unit extracts the text content in the picture, and then the question-answering implementation unit performs smart filling on the contract field corresponding to the contract management system according to the text content of the target field, thereby realizing rapid recognition and smart filling of the contract content, thereby improving the efficiency and accuracy of contract filling. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in this embodiment or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 An architectural diagram of a smart contract filling device provided in an embodiment of the present application;

[0039] Figure 2 A structural diagram of a smart filling device provided in an embodiment of the present application;

[0040] Figure 3 Another structural schematic diagram of the intelligent filling device provided in an embodiment of the present application;

[0041] Figure 4 A flow chart of a method for intelligently filling a contract provided in an embodiment of the present application. DETAILED DESCRIPTION

[0042] The technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0043] In order to facilitate further understanding of the technical solution provided by this application, the background technology involved in this application will be explained below.

[0044] With the development of informatization, paper contracts need to be entered into the contract information management system for management after signing so as to be used for subsequent contract execution.

[0045] However, the creation and management of contracts is often a time-consuming and tedious task. Especially when filling out contracts digitally, people often need to process a large amount of text information and ensure that various details are entered accurately, which may involve a lot of manual operations and prone to errors, resulting in low efficiency and accuracy in contract filling.

[0046] In order to overcome the above technical problems, the embodiment of the present application provides a smart filling device and smart filling method for a contract, wherein the smart filling device includes a data input unit, an OCR optical character recognition unit and a question and answer implementation unit; wherein the data input unit is used to obtain the contract uploaded by the user and the question input by the user, and the contract is a digital image contract obtained by converting a paper contract through a scanning device or a camera device; the OCR optical character recognition unit is used to perform text recognition on the contract and obtain the text content of the contract; the question and answer implementation unit is used to extract the target field according to the question input by the user, and obtain the text content of the target field according to the text content of the contract; the question and answer implementation unit is also used to perform smart filling on the contract field corresponding to the contract management system according to the text content of the target field to obtain the smart filling result. Through the above technical solution, the data input unit sends the contract uploaded by the user to the OCR optical character recognition unit, and the OCR optical character recognition unit extracts the text content in the picture. Then, the question and answer implementation unit performs smart filling on the contract field corresponding to the contract management system according to the text content of the target field, so as to realize rapid recognition and smart filling of the contract content, thereby improving the efficiency and accuracy of contract filling.

[0047] The technical terms of the embodiments of the present application are introduced as follows:

[0048] OCR (Optical Character Recognition): Optical character recognition is a technology that converts text in an image into editable text. It is usually used to scan paper documents or image files to extract the text in them for text search, editing or other text processing operations.

[0049] Question-answering large models: refers to natural language processing models trained on large amounts of data that can understand and answer questions asked by humans. These models usually use deep learning technology and are pre-trained with large amounts of text data, and can then be used for various question-answering tasks, including reading comprehension, question answering, etc.

[0050] Confidence: In the field of machine learning and artificial intelligence, confidence refers to the degree of confidence or credibility of the model's prediction results. It indicates the degree of certainty of the model's prediction results, usually expressed as a probability or score.

[0051] Regular matching: Regular expression is a tool used to describe string patterns. It can be used to search, match and replace specific patterns in text. It is the process of using regular expressions to find strings that meet specific patterns.

[0052] Deep learning: A machine learning method that uses multi-layer neural network models to learn representations and features of data to solve various tasks such as classification, regression, clustering, and generation.

[0053] Original text enhancement: A data preprocessing technique that aims to improve model performance and generalization. It involves processing the original text, such as word segmentation, stemming, and stop word removal, to reduce the complexity and noise of the data and extract useful information.

[0054] Representation-oriented text augmentation: A data augmentation technique that enhances the data by modifying the representation of the model input. This may include operations such as word embedding transformation, vocabulary replacement, or word order change to increase data diversity and model robustness.

[0055] Accuracy: refers to the proportion of samples correctly classified by the model.

[0056] Recall rate: refers to the ability of the model to successfully find all positive samples in the classification task. It is the ratio of true positive examples (samples correctly predicted by the model as positive) to all actual positive examples (including samples predicted by the model as positive and actually positive).

[0057] F1 score: The harmonic mean of accuracy and recall, used to comprehensively evaluate the performance of the model. It is the harmonic mean of the ratio of precision to recall and is often used to solve the problem of class imbalance.

[0058] Fine-tuning adjustment: refers to fine-tuning model parameters or hyperparameters based on the pre-trained model to make it better suited to specific tasks or data sets.

[0059] See also Figure 1 , Figure 1 This is an architecture diagram of the smart contract filling device provided in the embodiment of the present application. The device consists of four units: a data input unit, an OCR processing unit, a question and answer realization unit, and a data storage unit.

[0060] The data input unit includes an attachment uploading component, a data processing component, a question entry component and a user scoring component for filling results.

[0061] Among them, the attachment upload component is used to upload the contract.

[0062] The data processing component is used to process the contract data and send the processed contract to the OCR processing unit.

[0063] The question input component is used to input the questions input by the user.

[0064] The user scores the filling result component, which is used to score the smart filling result.

[0065] The OCR processing unit includes an image-to-text component for performing text recognition on the contract and obtaining the text content of the contract.

[0066] The question-answering implementation unit includes a text and question processing component, a question-answering large model component, and an answer processing component.

[0067] The question-answering model component is used to extract target fields according to the questions input by the user.

[0068] The text and question processing component is used to obtain the text content of the target field according to the text content of the contract.

[0069] The answer processing component is used to intelligently fill the contract field corresponding to the contract management system according to the text content of the target field to obtain an intelligent filling result.

[0070] The data storage unit includes a data storage component for storing the smart filling result and the scoring result in a database.

[0071] Next, each module of the intelligent filling device is introduced in detail.

[0072] See also Figure 2 , Figure 2 A structural diagram of an intelligent filling device provided in an embodiment of the present application. The intelligent filling device is applied to a contract management system.

[0073] The device may include: a data input unit 201 , an OCR optical character recognition unit 202 , and a question-answering implementation unit 203 .

[0074] The data input unit 201 is used to obtain the contract uploaded by the user and the question input by the user. The contract is a digital image contract obtained by converting a paper contract through a scanning device or a camera device.

[0075] Specifically, the data input unit is designed with a user-friendly interface and process so that users can operate and interact intuitively. The user interface provides concise and clear instructions and prompts to guide users to complete the process of uploading contract attachments and filling in contract information. Contract smart filling is added to the main process of contract entry as an optional function. Users can use it selectively, which helps to reduce the pressure of OCR and question-and-answer large model processing. In addition to smart filling of default fields, users can also enter relevant questions about other fields they want to fill. The question-and-answer model can obtain user intentions through input questions, return the target fields involved in the user's questions, and make necessary modifications and confirmations to the generated filled content.

[0076] Specifically: Step 1: Determine the number of contracts uploaded by the user. When uploading multiple contracts at one time, compress and package the multiple contracts into one attachment;

[0077] Step 2: Determine whether the user uses the smart fill function. If not, the user needs to manually enter all key contract information into the contract management system based on the contract document;

[0078] In a possible implementation, if the user does not input a question, the question and answer implementation unit is also used to obtain the text content of the default field based on the text content of the contract, and the default field includes the contract number, contract name, date, amount and contracting party information.

[0079] Step 3: Determine whether the user has entered a question. When the user has entered a question, the question is first understood through the question-answering model to extract the target fields that the user wants to fill in intelligently.

[0080] In one possible implementation, if the user has not entered a question, the question and answer implementation unit is also used to obtain the text content of the default field based on the text content of the contract, and the default field includes the contract number, contract name, date, amount and contracting party information, and intelligently fills in the default field.

[0081] Step 4: The user scores the smart fill result and obtains the score of the smart fill result. The score result will be used in the subsequent training and optimization of the question-answering model.

[0082] The OCR optical character recognition unit 202 is used to perform text recognition on the contract to obtain the text content of the contract.

[0083] It is understandable that the OCR processing unit is built using the OCR engine, and after model training, the OCR model has better contract recognition capabilities.

[0084] In a possible implementation, the OCR optical character recognition unit includes an OCR optical character recognition model, and the OCR optical character recognition model is used to perform text recognition on the contract to obtain the text content of the contract.

[0085] The training process of the OCR optical character recognition model includes:

[0086] Obtaining a training data set, wherein the training data set includes historical contracts and font files generated by an OCR optical character recognition engine tool;

[0087] Using the OCR optical character recognition engine tool to convert the format of the training data set to obtain a converted training data set;

[0088] The OCR optical character recognition model is trained using the converted training data set to obtain a trained OCR optical character recognition model.

[0089] Specifically: Step 1: Obtain historical contracts through historical data in the contract management system;

[0090] Step 2: Verify the format of the contract file. If it is a compressed file, decompress it first, and then split the file. After the split contract, each picture only contains one page of contract information.

[0091] Step 3: Use the OCR engine's built-in tools to generate font files so that the OCR model has the ability to recognize multiple fonts;

[0092] Step 4: The split contract documents and the generated font files are combined into a training data set, and the OCR engine's built-in tool is used to convert the data set into a box format file (this format describes the location and content of the text in each image);

[0093] Step 5: Use the training tool provided by the OCR engine to train the model. This process will use the training dataset and font folder to adjust the OCR recognition model so that it can better recognize the target text in the contract;

[0094] Step 6: After evaluating the trained model based on the recognition accuracy, optimize the model by adjusting the training parameters and increasing the amount of training data;

[0095] Step 7: Integrate the trained model into the contract management system to realize the text recognition function.

[0096] It can be understood that the embodiment of the present application can continuously improve the accuracy of text recognition by the OCR optical character recognition model by continuously training and updating the OCR optical character recognition model, which is beneficial to subsequent intelligent filling.

[0097] The question-answering implementation unit 203 is used to extract the target field according to the question input by the user, and obtain the text content of the target field according to the text content of the contract;

[0098] The question-answering implementation unit 203 is also used to intelligently fill in the contract field corresponding to the contract management system according to the text content of the target field to obtain an intelligent filling result.

[0099] Among them, the question and answer implementation unit includes a question and answer big model, and the question and answer big model is used to extract the target field according to the question input by the user.

[0100] It can be understood that the deep learning framework used in the embodiment of the present application implements the question-answering big model, and the trained question-answering big model has a better understanding of contract-related issues in the banking and financial field.

[0101] The training process of the large question-answering model includes:

[0102] Acquire a training sample set, wherein the training sample set includes a target text and a historical question input by a user, wherein the target text is a text above a confidence threshold after the text content of the historical contract is cleaned according to the confidence level;

[0103] Performing text enhancement on the training sample set by using a hybrid text enhancement method for original text enhancement and for representation text enhancement to obtain an enhanced training sample set;

[0104] Using the enhanced training sample set, training the question-answering large model to obtain a trained question-answering large model;

[0105] The model parameters of the trained question-answering model are adjusted using evaluation indicators, wherein the evaluation indicators include accuracy, recall, and the harmonic mean of the accuracy and recall.

[0106] Specifically, step 1: clean the bank financial contract text data returned by the OCR processing unit according to the confidence level, filter out the text with a confidence level higher than 0.7, and then combine it with the question input by the user to form an initial training data set;

[0107] Step 2: Use a hybrid text enhancement method for original text enhancement and representation text enhancement to enhance the text of the initial dataset to obtain more training data, which is beneficial to subsequent model training;

[0108] Step 3: Considering the complexity of the contract-related information question-answering task and the need for semantic understanding, a deep learning framework is selected as the basic model, and the prepared training data set is input into the model for training;

[0109] Step 4: using evaluation indicators to adjust the model parameters of the trained question-answering model, wherein the evaluation indicators include accuracy, recall, and the harmonic mean F1 score of the accuracy and recall;

[0110] Step 5: Fine-tuning the question-answering model based on the evaluation indicators to ensure that the model better adapts to the target task;

[0111] Step 6: Integrate the fine-tuned question-and-answer model into the contract management system program to implement the contract question-and-answer function in the banking financial field.

[0112] It is understandable that the embodiment of the present application can automatically identify contract information and generate contract content through the question-and-answer large model, without the need to manually fill in each item, thereby greatly improving the efficiency of contract filling. The user only needs to provide questions or prompts, and the device can automatically generate the corresponding contract information and fill it into the system, eliminating tedious manual operations. In addition, the device has the ability to understand the contract text and can flexibly and accurately identify and extract key contract information. Moreover, through continuous interaction and feedback with users, the performance and accuracy of the question-and-answer large model are continuously improved. The device can accumulate more contextual information, improve the scope of application and universality of the device, and thus continuously improve the efficiency and accuracy of intelligent filling.

[0113] It should be noted that the embodiment of the present application is designed with a friendly front-end interface, so that users can easily interact with the device during the contract information entry process, and guide users to complete the contract filling process through an intuitive operation interface and clear prompt information. The front end splits the contract attachments and transmits them to the back-end OCR module, extracts the text content in the picture, and performs data cleaning on the extracted text according to the confidence level to obtain more accurate text content, and then submits the text content and questions to the question and answer model together. Finally, the results returned by the question and answer model are formatted and returned to the front end for automatic filling in the corresponding position of the contract system.

[0114] It can be seen from the above technical scheme that the intelligent filling device provided by the embodiment of the present application includes a data input unit, an OCR optical character recognition unit and a question-answering implementation unit; wherein the data input unit is used to obtain the contract uploaded by the user and the question input by the user, and the contract is a digital image contract obtained by converting a paper contract through a scanning device or a camera device; the OCR optical character recognition unit is used to perform text recognition on the contract and obtain the text content of the contract; the question-answering implementation unit is used to extract the target field according to the question input by the user, and obtain the text content of the target field according to the text content of the contract; the question-answering implementation unit is also used to intelligently fill the contract field corresponding to the contract management system according to the text content of the target field to obtain an intelligent filling result. Through the above technical scheme, the data input unit sends the contract uploaded by the user to the OCR optical character recognition unit, and the OCR optical character recognition unit extracts the text content in the picture. Then, the question-answering implementation unit intelligently fills the contract field corresponding to the contract management system according to the text content of the target field, so as to realize rapid recognition and intelligent filling of the contract content, thereby improving the efficiency and accuracy of contract filling.

[0115] In one possible implementation, Figure 3 As shown, Figure 3 Another structural schematic diagram of the intelligent filling device provided in an embodiment of the present application.

[0116] The device includes a data input unit 301, an OCR optical character recognition unit 302 and a question-answering unit 303, and further includes a data storage unit 304;

[0117] The data storage unit 304 is used to store the smart filling result and the scoring result in a database.

[0118] It can be understood that after completing the intelligent filling of the contract, the user will score it based on whether the filling effect meets the requirements. This device stores the contract filling content and its scoring information in the MySQL database to expand the data set for subsequent training of the question-and-answer model, and gradually improve the ability of intelligent contract filling through multiple model training.

[0119] Correspondingly, the present application also provides a smart filling method for a contract, which is applied to the smart filling device described in the above embodiment, such as Figure 4 The figure shows a flow chart of the intelligent filling method of a contract provided in an embodiment of the present application.

[0120] The method comprises:

[0121] Step S401: Acquire the contract uploaded by the user and the question input by the user, wherein the contract is a digital image contract obtained by converting a paper contract through a scanning device or a camera device.

[0122] Step S402: Perform text recognition on the contract to obtain the text content of the contract.

[0123] Step S403: extracting a target field according to the question input by the user, and acquiring the text content of the target field according to the text content of the contract.

[0124] Step S404: intelligently fill in the contract field corresponding to the contract management system according to the text content of the target field to obtain an intelligent filling result.

[0125] It can be seen from the above technical scheme that the embodiment of the present application first obtains the contract uploaded by the user and the question input by the user, the contract is a digital image contract obtained by converting a paper contract through a scanning device or a camera device, and then performs text recognition on the contract to obtain the text content of the contract, and then extracts the target field according to the question input by the user, obtains the text content of the target field according to the text content of the contract, and finally intelligently fills the contract field corresponding to the contract management system according to the text content of the target field to obtain an intelligent filling result. Through the above technical scheme, the data input unit sends the contract uploaded by the user to the OCR optical character recognition unit, the OCR optical character recognition unit extracts the text content in the picture, and then the question and answer realization unit intelligently fills the contract field corresponding to the contract management system according to the text content of the target field, realizing rapid recognition and intelligent filling of the contract content, thereby improving the efficiency and accuracy of contract filling.

[0126] For the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the order of the actions described, because according to the present invention, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0127] An embodiment of the present application further provides an electronic device, comprising: a memory for storing one or more programs;

[0128] Processor; when the one or more programs are executed by the processor, the intelligent filling method of the contract described in the above embodiment is implemented.

[0129] An embodiment of the present application also provides a computer storage medium, in which a program is stored. When the program is executed by a processor, the intelligent filling method of the contract described in the above embodiment is implemented.

[0130] The "first" and "second" in the names such as "first" and "second" (if any) mentioned in the embodiments of the present application are only used as name identifiers and do not represent the first or second in order.

[0131] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0132] Those skilled in the art can understand that the flowchart shown in the figure is only an example in which the embodiments of the present application can be implemented, and the scope of application of the embodiments of the present application is not limited by any aspect of the flowchart.

[0133] In the several embodiments provided in the present application, it should be understood that the disclosed methods, devices and equipment can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the device or unit can be electrical, mechanical or other forms.

[0134] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, the functional units in the various embodiments of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0135] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.

[0136] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A smart filling device for a contract, characterized in that, Applied to a contract management system, the device includes a data input unit, an OCR optical character recognition unit and a question-answering implementation unit; The data input unit is used to obtain the contract uploaded by the user and the question input by the user, wherein the contract is a digital image contract obtained by converting a paper contract through a scanning device or a camera device; The OCR optical character recognition unit is used to perform text recognition on the contract to obtain the text content of the contract; The question-answering implementation unit is used to extract the target field according to the question input by the user, and obtain the text content of the target field according to the text content of the contract; The question-and-answer implementation unit is also used to intelligently fill in the contract field corresponding to the contract management system according to the text content of the target field to obtain an intelligent filling result.

2. The device according to claim 1, characterized in that If the user does not input a question, the question-and-answer implementation unit is further used to obtain the text content of the default field according to the text content of the contract, and the default field includes the contract number, contract name, date, amount and contracting party information.

3. The device according to claim 1, characterized in that The OCR optical character recognition unit includes an OCR optical character recognition model, and the OCR optical character recognition model is used to perform text recognition on the contract to obtain the text content of the contract.

4. The device according to claim 3, characterized in that The training process of the OCR optical character recognition model includes: Obtaining a training data set, wherein the training data set includes historical contracts and font files generated by an OCR optical character recognition engine tool; Using the OCR optical character recognition engine tool to convert the format of the training data set to obtain a converted training data set; The OCR optical character recognition model is trained using the converted training data set to obtain a trained OCR optical character recognition model.

5. The device according to claim 1, characterized in that The question-answering implementation unit includes a question-answering big model, and the question-answering big model is used to extract the target field according to the question input by the user.

6. The device according to claim 5, characterized in that The training process of the question-answering large model includes: Acquire a training sample set, wherein the training sample set includes a target text and a historical question input by a user, wherein the target text is a text above a confidence threshold after the text content of the historical contract is cleaned according to the confidence level; Performing text enhancement on the training sample set by using a hybrid text enhancement method for original text enhancement and for representation text enhancement to obtain an enhanced training sample set; Using the enhanced training sample set, training the question-answering large model to obtain a trained question-answering large model; The model parameters of the trained question-answering model are adjusted using evaluation indicators, wherein the evaluation indicators include accuracy, recall, and the harmonic mean of the accuracy and recall.

7. The device according to claim 1, characterized in that The data input unit is also used to score the smart filling result to obtain a scoring result of the smart filling result.

8. The device according to claim 7, characterized in that The device also includes a data storage unit; The data storage unit is used to store the smart filling result and the scoring result in a database.

9. The device according to claim 1, characterized in that The data input unit includes a visual interface for guiding the user to perform operations and interactions.

10. A smart filling method for a contract, characterized in that: Applied to the smart filling device according to any one of claims 1 to 9, the method comprises: Obtaining a contract uploaded by a user and a question input by the user, wherein the contract is a digital image contract obtained by converting a paper contract through a scanning device or a camera device; Performing text recognition on the contract to obtain the text content of the contract; Extracting a target field according to the question input by the user, and acquiring text content of the target field according to the text content of the contract; The contract field corresponding to the contract management system is intelligently filled according to the text content of the target field to obtain an intelligent filling result.

Images