Landing contract evidence chain authenticity identification system

Abstract

The invention belongs to the technical field of information, and particularly relates to a loan contract evidence chain authenticity identification system, which comprises an evidence acquisition module, a data processing and analysis module, an authenticity identification module and a result output module, according to the loan contract evidence chain authenticity identification system provided by the invention, the OCR algorithm, the convolutional neural network algorithm and the recurrent neural network algorithm are integrated, so that the system can automatically identify and process text and image information in a loan contract, the time and cost of manual auditing are greatly reduced, and the identification efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the field of information technology, and specifically refers to a loan contract evidence chain authenticity identification system. Background Art

[0002] The loan contract evidence chain refers to a complete system of interrelated and mutually corroborated evidence materials in the process of financial lending, which is used to prove the authenticity and legality of the loan relationship. Usually, these evidence chains include the text of the loan contract, the handwritten signatures of both parties, transaction records, communication records, and possible witness testimony. These evidences support each other to form a closed chain of evidence, which is of great significance for resolving loan disputes and protecting the legitimate rights and interests of the parties.

[0003] At present, the authenticity of the evidence chain of loan contracts is mainly verified by manual review, which is not only inefficient but also easily affected by human factors, resulting in errors in the verification results. Although relevant technical literature has proposed solutions based on blockchain technology to improve the immutability and transparency of the evidence chain, these solutions are often costly to implement and difficult to popularize, and are unlikely to be widely used in the short term. Summary of the invention

[0004] In view of the above situation, in order to solve the problem of authenticity identification of the loan contract evidence chain, it mainly relies on manual review. This method is not only inefficient, but also easily affected by human factors, resulting in errors in the identification results. The present invention provides a loan contract evidence chain authenticity identification system. By integrating OCR algorithm, convolutional neural network algorithm and recurrent neural network algorithm, the system can automatically identify and process text and image information in the loan contract, greatly reducing the time and cost of manual review and improving the identification efficiency.

[0005] The technical solution adopted by the present invention is as follows: The present invention provides a loan contract evidence chain authenticity identification system, comprising:

[0006] Evidence collection module: responsible for collecting the original data of the loan contract from the user interface;

[0007] Data processing and analysis module: extract key features, perform data comparison and analysis;

[0008] Authenticity identification module: outputs identification results according to preset judgment criteria;

[0009] Result output module: output the identification results in the form of a report, and support electronic archiving and online query.

[0010] Furthermore, the evidence collection module uses an OCR algorithm to automatically identify the content of the contract text and verify the authenticity of the signature. Furthermore, the data processing and analysis module uses a convolutional neural network algorithm and a recurrent neural network algorithm to perform natural language processing on the text content in the loan contract, extract keywords and phrases, perform image processing on the signature and seal, and extract edge, color and texture features.

[0011] Furthermore, the authenticity identification module ensures the performance of the model on unknown data through cross-validation and hyperparameter tuning.

[0012] Furthermore, the report contains identification results, analysis process, algorithms used and standard information.

[0013] Furthermore, the report is in PDF format for easy storage and sharing.

[0014] Furthermore, the formula of the OCR algorithm is as follows:

[0015] Classification:

[0016] Grayscale: Convert a color image to a grayscale image;

[0017] I gray (x, y) = 0.2989·R(x, y) + 0.5870·G(x, y) + 0.1140·B(x, y); Binarization: convert the grayscale image into a black and white image;

[0018] T = argmax T ((w0·μ0+w1·μ1) 2 );

[0019] Where w0 and w1 are the weights of the two types of pixels, μ0 and μ1 are the means of the two types of pixels;

[0020] Feature extraction:

[0021] HOG feature extraction;

[0022]

[0023] Among them G x and G y is the gradient of the image in the x and y directions;

[0024] Classification:

[0025] Character classification using convolutional neural networks;

[0026] O = f(W*I+b);

[0027] Where O is the output, W is the convolution kernel, I is the input image, b is the bias, and f is the activation function; post-processing:

[0028] Use n-gram to correct the recognition results to improve the recognition accuracy.

[0029] Furthermore, the formula of the convolutional neural network algorithm is as follows:

[0030] Convolutional Layer:

[0031] Convolution operation:

[0032]

[0033] Where I is the input image, K is the convolution kernel, and x and y are the locations of the convolution results;

[0034] Activation function:

[0035] f(x)=max(0,x);

[0036] Pooling layer:

[0037] Max Pooling:

[0038]

[0039] Where pool_region is the specified pooling region size;

[0040] Fully connected layer:

[0041] In the fully connected layer, the input is linearly transformed by the weight matrix W and the bias b, and then the activation function is used:

[0042] O = f(W·X+b);

[0043] Where O is the output and X is the input vector;

[0044] Loss function:

[0045] For classification tasks, a commonly used loss function is the cross entropy loss:

[0046]

[0047] Where C is the number of categories, y i is the true label, is the predicted probability;

[0048] Backward Propagation:

[0049] When updating weights, gradient descent is used:

[0050]

[0051] where η is the learning rate.

[0052] Furthermore, the formula of the recurrent neural network algorithm is as follows:

[0053] Basic structure: In the RNN at time step t, the input x t and the hidden state h at the previous time step t-1 is used to calculate the hidden state h at the current time step t :

[0054] h t =f(W h h t-1 +W x x t + b);

[0055] in:

[0056] W h is the weight matrix from hidden state to current hidden state;

[0057] W x is the weight matrix input to the hidden state;

[0058] b is the bias term;

[0059] f is the activation function;

[0060] Output layer: According to the current hidden state h t Calculate the output y t :

[0061] y t =W y h t +b y ;

[0062] Where W y is the weight matrix from hidden state to output layer, b y is the bias term of the output layer;

[0063] Loss function: For classification tasks, cross entropy loss is usually used:

[0064]

[0065] Where T is the sequence length, C is the number of categories, and y t，i is the true label, is the predicted probability;

[0066] Backpropagation: RNN uses backpropagation to update weights through time. The calculation of gradients involves the expansion of time steps:

[0067]

[0068]

[0069] Long Short-Term Memory Networks:

[0070] Forget Gate:

[0071] f t =σ(W f ·[h t-1 , x t ]+b f );

[0072] Input Gate:

[0073] i t =σ(W i ·[h t-1 , x t ]+b i );

[0074] Candidate memory cells:

[0075]

[0076] Update memory unit:

[0077]

[0078] Output Gate:

[0079] o t =σ(W o ·[h t-1 , x t ]+b o )

[0080] Hidden state:

[0081] h t =o t tanh(C t ).

[0082] This scheme also discloses an operation method of a loan contract evidence chain authenticity identification system, which mainly includes the following steps:

[0083] Step A1: The user uploads the original data of the loan contract through the user interface;

[0084] Step A2: The system automatically calls the evidence collection module and uses the OCR algorithm to perform text recognition on the uploaded document to extract the contract text content and related information;

[0085] Step A3: passing the extracted text content to the data processing and analysis module;

[0086] Step A4: Use convolutional neural network algorithms and recurrent neural network algorithms to perform natural language processing on the text content to extract keywords and phrases, and perform image processing on the signatures and seals in the contract to extract edge, color and texture features to identify potential signs of forgery;

[0087] Step A5: The extracted features are input into the authenticity identification module. The module uses the trained model to judge the authenticity of the contract according to the preset judgment criteria. Through cross-validation and hyperparameter tuning, the model's performance on unknown data is ensured, and the identification result is output;

[0088] Step A6: The system generates an appraisal report and outputs it in PDF format for users to save and share.

[0089] The beneficial effects achieved by the present invention using the above structure are as follows: The present invention provides a loan contract evidence chain authenticity identification system, which achieves the following beneficial effects:

[0090] (1) By integrating OCR algorithm, convolutional neural network algorithm and recurrent neural network algorithm, the system can automatically identify and process text and image information in loan contracts, greatly reducing the time and cost of manual review and improving identification efficiency.

[0091] (2) The data processing and analysis module is used to perform natural language processing on the contract content to extract keywords and phrases. At the same time, image processing is performed on the signature and seal to extract edge, color and texture features, so that traces of forgery or tampering can be more accurately identified, thereby improving the accuracy of identification.

[0092] (3) The authenticity identification module ensures the stability and reliability of the model’s performance on unknown data through cross-validation and hyperparameter tuning, further enhancing the system’s generalization ability and adaptability.

[0093] (4) The appraisal report generated by the system contains detailed appraisal results, analysis process, algorithms used and standard information, making the appraisal results highly transparent and traceable, which is easy for users to understand and accept, and also convenient for judicial authorities to review when necessary. BRIEF DESCRIPTION OF THE DRAWINGS

[0094] Figure 1 This is a structural block diagram of a loan contract evidence chain authenticity identification system proposed by the present invention;

[0095] Figure 2 A flow chart of a method for identifying the authenticity of a loan contract evidence chain proposed by the present invention;

[0096] Figure 3 This is a schematic diagram of the working principle of the evidence collection module.

[0097] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0098] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0099] Embodiment 1:

[0100] See also Figure 1-3 As shown, this embodiment is a loan contract evidence chain authenticity identification system, including an evidence collection module, a data processing and analysis module, an authenticity identification module and a result output module;

[0101] Among them, the evidence collection module is responsible for collecting the original data of the loan contract from the user interface;

[0102] Among them, the data processing and analysis module: extract key features, perform data comparison and analysis;

[0103] Among them, the authenticity identification module: outputs the identification result according to the preset judgment criteria;

[0104] Among them, the result output module: outputs the identification results in the form of a report and supports electronic archiving and online query.

[0105] Embodiment 2:

[0106] See also Figure 1-3 As shown, this embodiment is a method for using a loan contract evidence chain authenticity identification system, comprising the following steps:

[0107] Step A1: The user uploads the original data of the loan contract through the user interface;

[0108] Step A2: The system automatically calls the evidence collection module and uses the OCR algorithm to perform text recognition on the uploaded document to extract the contract text content and related information;

[0109] Step A3: passing the extracted text content to the data processing and analysis module;

[0110] Step A4: Use convolutional neural network algorithms and recurrent neural network algorithms to perform natural language processing on the text content to extract keywords and phrases, and perform image processing on the signatures and seals in the contract to extract edge, color and texture features to identify potential signs of forgery;

[0111] Step A5: The extracted features are input into the authenticity identification module. The module uses the trained model to judge the authenticity of the contract according to the preset judgment criteria. Through cross-validation and hyperparameter tuning, the model's performance on unknown data is ensured, and the identification result is output;

[0112] Step A6: The system generates an appraisal report and outputs it in PDF format for users to save and share.

[0113] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

[0114] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A loan contract evidence chain authenticity identification system, characterized by: include: Evidence collection module: responsible for collecting the original data of the loan contract from the user interface; Data processing and analysis module: extract key features, perform data comparison and analysis; Authenticity identification module: outputs identification results according to preset judgment criteria; Result output module: output the identification results in the form of a report, and support electronic archiving and online query.

2. According to claim 1, a loan contract evidence chain authenticity identification system is characterized by: The evidence collection module uses an OCR algorithm to automatically identify the content of the contract text and verify the authenticity of the signature.

3. According to claim 1, a loan contract evidence chain authenticity identification system is characterized by: The data processing and analysis module uses convolutional neural network algorithms and recurrent neural network algorithms to perform natural language processing on the text content in the loan contract, extract keywords and phrases, perform image processing on signatures and seals, and extract edge, color and texture features.

4. According to claim 1, a loan contract evidence chain authenticity identification system is characterized by: The authenticity identification module ensures the performance of the model on unknown data through cross-validation and hyperparameter tuning.

5. According to claim 1, the authenticity identification system of the loan contract evidence chain is characterized by: The report contains the identification results, analysis process, algorithms used and standard information.

6. According to claim 1, a loan contract evidence chain authenticity identification system is characterized by: The report is in PDF format for easy saving and sharing.

7. According to claim 2, a loan contract evidence chain authenticity identification system is characterized by: The formula of the OCR algorithm is as follows: Classification: Grayscale: Convert a color image to a grayscale image; I gray (x,y)=0.2989·R(x,y)+0.5870·G(x,y)+0.1140·B(x,y); Binarization: Convert grayscale images into black and white images; T=argmax T ((w0·μ0+w1·μ1) 2 ); Where w0 and w1 are the weights of the two types of pixels, μ0 and μ1 are the means of the two types of pixels; Feature extraction: HOG feature extraction; Among them G x and G y is the gradient of the image in the x and y directions; Classification: Character classification using convolutional neural networks; O = f(W*I+b); Where O is the output, W is the convolution kernel, I is the input image, b is the bias, and f is the activation function; Post-processing: Use n-gram to correct the recognition results to improve the recognition accuracy.

8. According to claim 3, a loan contract evidence chain authenticity identification system is characterized by: The formula of the convolutional neural network algorithm is as follows: Convolutional Layer: Convolution operation: Where I is the input image, K is the convolution kernel, and x and y are the locations of the convolution results; Activation function: f(x)=max(0,x); Pooling layer: Max Pooling: Where pool_region is the specified pooling region size; Fully connected layer: In the fully connected layer, the input is linearly transformed by the weight matrix W and the bias b, and then the activation function is used: O = f(W·X+b); Where O is the output and X is the input vector; Loss function: For classification tasks, a commonly used loss function is the cross entropy loss: Where C is the number of categories, y i is the true label, is the predicted probability; Backward Propagation: When updating weights, gradient descent is used: in η is the learning rate.

9. According to claim 3, a loan contract evidence chain authenticity identification system is characterized by: The formula of the recurrent neural network algorithm is as follows: Basic structure: In the RNN at time step t, the input x t and the hidden state h at the previous time step t-1 is used to calculate the hidden state h at the current time step t : h t =f(W hht- 1+W x x t +b); in: W h is the weight matrix from hidden state to current hidden state; W x is the weight matrix input to the hidden state; b is the bias term; f is the activation function; Output layer: According to the current hidden state h t Calculate the output y t : y t =W y h t +b y ; Where W y is the weight matrix from hidden state to output layer, b y is the bias term of the output layer; Loss function: For classification tasks, cross entropy loss is usually used: Where T is the sequence length, C is the number of categories, and y t，i is the true label, is the predicted probability; Backpropagation: RNN uses backpropagation to update weights through time. The calculation of gradients involves the expansion of time steps: Long Short-Term Memory Networks: Forget Gate: f t =σ(W f ·[h t-1 ,x t ]+b f ); Input Gate: i t =σ(W i ·[h t-1 ,x t ]+b i ); Candidate memory cells: Update memory unit: Output Gate: the t =σ(W o ·[h t-1 ,x t ]+b o ) Hidden state: h t =o t ·tanh(C t )。 10. An operating method of a loan contract evidence chain authenticity identification system, characterized by: The operation of the authenticity identification system of the loan contract evidence chain described in claim 9 mainly includes the following steps: Step A1: The user uploads the original data of the loan contract through the user interface; Step A2: The system automatically calls the evidence collection module and uses the OCR algorithm to perform text recognition on the uploaded document to extract the contract text content and related information; Step A3: passing the extracted text content to the data processing and analysis module; Step A4: Use convolutional neural network algorithms and recurrent neural network algorithms to perform natural language processing on the text content to extract keywords and phrases, and perform image processing on the signatures and seals in the contract to extract edge, color and texture features to identify potential signs of forgery; Step A5: The extracted features are input into the authenticity identification module. The module uses the trained model to judge the authenticity of the contract according to the preset judgment criteria. Through cross-validation and hyperparameter tuning, the model's performance on unknown data is ensured, and the identification result is output; Step A6: The system generates an appraisal report and outputs it in PDF format for users to save and share.

Images