Enterprise financial management risk early warning system and method based on data analysis

Abstract

The invention provides an enterprise financial management risk early warning system and method based on data analysis, and relates to the technical field of financial management, and the method comprises the steps: obtaining a meal fee reimbursement voucher and a meal video; based on the dining video, using a video processing model to determine a plurality of dining dish segmentation images and a plurality of pieces of dining dish information; based on the plurality of dining dish segmentation images and the plurality of pieces of dining dish information, using a generative adversarial network to generate a simulated high-definition image of each dining dish; determining an estimated price range of each meal dish by using a price determination model based on the simulated high-definition image of each meal dish; determining the risk degree of the meal fee reimbursement voucher based on the estimated price range of each meal dish and the meal fee reimbursement voucher; and carrying out risk early warning based on the risk degree of the meal fee reimbursement voucher. The method can accurately determine the risk degree of employee meal fee reimbursement.

Description

Technical Field

[0001] The present invention relates to the technical field of financial management, and in particular to an enterprise financial management risk early warning system and method based on data analysis. Background Art

[0002] In the financial reimbursement process, with the rapid expansion of corporate business and the frequent increase in employee travel, catering, business banquets and other activities, the complexity and data volume of the meal expense reimbursement process have increased significantly. The traditional meal expense reimbursement method is highly dependent on manual review, which has many limitations. First, manual review is inefficient, especially during the peak reimbursement period. Reviewers need to check the details of each meal expense reimbursement one by one, resulting in a lengthy review cycle and affecting employees' reimbursement experience. Secondly, manual review is prone to deviations due to fatigue, negligence or lack of experience, such as failure to promptly detect false reporting, misreporting or duplicate reimbursement, which increases the company's financial risks.

[0003] Therefore, how to accurately determine the risk level of employee meal expense reimbursement is an urgent problem that needs to be solved. Summary of the invention

[0004] The main technical problem solved by the present invention is how to accurately determine the risk level of employee meal expense reimbursement.

[0005] According to a first aspect, the present invention provides a risk warning method for enterprise financial management based on data analysis, comprising: obtaining meal reimbursement vouchers and meal videos; determining multiple meal dish segmentation images and multiple meal dish information based on the meal videos using a video processing model; generating simulated high-definition images of each meal dish based on the multiple meal dish segmentation images and the multiple meal dish information using a generative adversarial network; determining an estimated price range for each meal dish based on the simulated high-definition image of each meal dish using a price determination model; determining the risk level of the meal reimbursement vouchers based on the estimated price range of each meal dish and the meal reimbursement vouchers; and issuing a risk warning based on the risk level of the meal reimbursement vouchers.

[0006] In one possible implementation, determining the risk of a meal reimbursement voucher based on the estimated price range of each meal dish and the meal reimbursement voucher includes: determining multiple matching dish information in the meal reimbursement voucher based on the multiple meal dish information and the meal reimbursement voucher using a matching degree determination model; constructing a graph structure, the graph structure including multiple nodes and multiple edges between the multiple nodes, the multiple nodes including multiple meal dish nodes and multiple matching dish nodes, wherein each meal dish node establishes an edge with a corresponding matching dish node, the node features of the meal dish nodes include meal dish information, the estimated price range of each meal dish, the node features of the matching dish nodes include matching dish information, and the edge between the meal dish node and the matching dish node is the price difference range between the meal dish and the matching dish; and determining the risk of the meal reimbursement voucher by processing the graph structure based on a graph autoencoder.

[0007] In a possible implementation, the risk warning based on the risk level of the meal expense reimbursement voucher includes: if the risk level of the meal expense reimbursement voucher is greater than a threshold, performing a manual check.

[0008] In a possible implementation, the video processing model is a Transformer model.

[0009] According to a second aspect, the present invention provides an enterprise financial management risk warning system based on data analysis, comprising: an acquisition module for acquiring meal reimbursement vouchers and meal videos; an information determination module for determining multiple meal dish segmentation images and multiple meal dish information based on the meal video using a video processing model; an image generation module for generating a simulated high-definition image of each meal dish based on the multiple meal dish segmentation images and the multiple meal dish information using a generative adversarial network; a price estimation module for determining an estimated price range for each meal dish based on the simulated high-definition image of each meal dish using a price determination model; a risk determination module for determining the risk level of a meal reimbursement voucher based on the estimated price range of each meal dish and the meal reimbursement voucher; and a risk warning module for providing a risk warning based on the risk level of the meal reimbursement voucher.

[0010] In one possible implementation, the risk determination module is also used to: determine multiple matching dish information in the meal reimbursement voucher based on the multiple meal dish information and the meal reimbursement voucher using a matching determination model; construct a graph structure, the graph structure including multiple nodes and multiple edges between the multiple nodes, the multiple nodes including multiple meal dish nodes and multiple matching dish nodes, wherein each meal dish node establishes an edge with a corresponding matching dish node, the node features of the meal dish nodes include meal dish information and an estimated price range of each meal dish, the node features of the matching dish nodes include matching dish information, and the edge between the meal dish node and the matching dish node is the price difference range between the meal dish and the matching dish; and determine the risk of the meal reimbursement voucher by processing the graph structure based on the graph autoencoder.

[0011] In a possible implementation, the risk warning based on the risk level of the meal expense reimbursement voucher includes: if the risk level of the meal expense reimbursement voucher is greater than a threshold, performing a manual check.

[0012] In a possible implementation, the video processing model is a Transformer model.

[0013] According to a third aspect, an embodiment of the present invention provides an electronic device, comprising: a processor; a memory; and a computer program; wherein the computer program is stored in the memory and configured to be executed by the processor to implement a method as described above, the method comprising: obtaining a meal reimbursement voucher and a meal video; determining a plurality of meal dish segmentation images and a plurality of meal dish information based on the meal video using a video processing model; generating a simulated high-definition image of each meal dish based on the plurality of meal dish segmentation images and the plurality of meal dish information using a generative adversarial network; determining an estimated price range for each meal dish based on the simulated high-definition image of each meal dish using a price determination model; determining the risk level of the meal reimbursement voucher based on the estimated price range of each meal dish and the meal reimbursement voucher; and providing a risk warning based on the risk level of the meal reimbursement voucher.

[0014] According to the fourth aspect, the present embodiment provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned enterprise financial management risk warning method based on data analysis, the method comprising: obtaining meal reimbursement vouchers and meal videos; determining a plurality of meal dish segmentation images and a plurality of meal dish information based on the meal videos using a video processing model; generating a simulated high-definition image of each meal dish based on the plurality of meal dish segmentation images and the plurality of meal dish information using a generative adversarial network; determining an estimated price range for each meal dish based on the simulated high-definition image of each meal dish using a price determination model; determining a risk level of the meal reimbursement voucher based on the estimated price range of each meal dish and the meal reimbursement voucher; and issuing a risk warning based on the risk level of the meal reimbursement voucher.

[0015] The present invention provides an enterprise financial management risk warning system and method based on data analysis. The method includes obtaining meal reimbursement vouchers and meal videos; using a video processing model to determine multiple meal dish segmentation images and multiple meal dish information based on the meal videos; using a generative adversarial network to generate a simulated high-definition image of each meal dish based on the multiple meal dish segmentation images and the multiple meal dish information; using a price determination model to determine an estimated price range for each meal dish based on the simulated high-definition image of each meal dish; determining the risk level of the meal reimbursement voucher based on the estimated price range of each meal dish and the meal reimbursement voucher; and issuing a risk warning based on the risk level of the meal reimbursement voucher. The method can accurately determine the risk level of employee meal reimbursement. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of a process flow of an enterprise financial management risk early warning method based on data analysis provided by an embodiment of the present invention; Figure 2 A schematic diagram of a process for determining the risk level of a meal expense reimbursement voucher provided in an embodiment of the present invention; Figure 3 A schematic diagram of an enterprise financial management risk early warning system based on data analysis provided by an embodiment of the present invention; Figure 4 A schematic diagram of an electronic device provided by an embodiment of the present invention; DETAILED DESCRIPTION

[0017] The present invention is further described in detail below by specific embodiments in conjunction with the accompanying drawings. Wherein similar elements in different embodiments adopt associated similar element numbers. In the following embodiments, many detailed descriptions are for making the present invention better understood. However, those skilled in the art can easily recognize that some features can be omitted in different situations, or can be replaced by other elements, materials, methods. In some cases, some operations related to the present invention are not shown or described in the specification, this is to avoid the core part of the present invention being overwhelmed by too much description, and for those skilled in the art, it is not necessary to describe these related operations in detail, and they can fully understand the related operations according to the description in the specification and the general technical knowledge in the art.

[0018] In an embodiment of the present invention, there is provided Figure 1 The enterprise financial management risk early warning method based on data analysis shown in the figure comprises steps S1 to S6: Step S1, obtaining meal expense reimbursement vouchers and meal videos; Meal reimbursement vouchers are formal documents submitted by employees when they claim reimbursement for meal expenses. They usually include invoices, receipts, etc., and record the meal time, location, amount and dish information.

[0019] Meal videos are video materials taken by employees during their meals using video acquisition equipment. Meal videos record information such as the dining scene, food display, and diners. Meal videos can be used to extract food information and analyze dining scenes.

[0020] Step S2, determining a plurality of meal dish segmentation images and a plurality of meal dish information based on the meal video using a video processing model; The video processing model is a Transformer model, the input of the video processing model is the dining video, and the output of the video processing model is a plurality of dining dish segmentation images and a plurality of dining dish information.

[0021] The Transformer model is a deep learning model based on the self-attention mechanism. The Transformer model consists of two parts: an encoder and a decoder. The encoder can learn the representation of the input sequence through the self-attention mechanism and the feedforward neural network. On this basis, the decoder introduces the multi-head attention mechanism, decodes the encoder output and generates the target sequence. The Transformer model has significant advantages in processing time series data. It can capture the correlation between different time points in the video and pay attention to the information at different positions in the input sequence at the same time, so that the key features in the video can be efficiently extracted. In some embodiments, the Transformer model can be used to process the time series information of dining videos, and can capture the correlation between different time points in the video, so that multiple dining dishes can be generated. Segmented images and multiple dining dishes information can be extracted.

[0022] The meal segmentation image is an independent image of each dish extracted after analyzing and processing the meal video through the video processing model. The segmentation image shows the appearance characteristics of the meal dishes and can preliminarily present the shape and details of each dish.

[0023] The plurality of meal dish information is a detailed description of each dish extracted from the meal video, and each meal dish information includes the name, color, gloss, quantity, size, and portion of the dish. For example, for a dish of "Kung Pao Chicken", its dish information may include the name of the dish "Kung Pao Chicken", the color "reddish brown", the gloss "shiny", the quantity "1 serving", the size "medium plate", and the portion "moderate".

[0024] The Transformer model uses a self-attention mechanism to simultaneously focus on information at different locations in the video and excels in handling long-distance dependencies. This mechanism enables the Transformer model to extract the key visual features of each dish from the dining video and generate dish segmentation images. In addition, the Transformer model's multi-head attention mechanism can capture the correlation between different time points in the video, thereby more accurately segmenting dish images and extracting dish information.

[0025] Step S3, based on the plurality of meal dish segmentation images and the plurality of meal dish information, using a generative adversarial network to generate a simulated high-definition image of each meal dish; The Generative Adversarial Network (GAN) includes a generator and a discriminator. The generator is responsible for generating new data samples, and the discriminator is responsible for evaluating the authenticity of the generated samples. The two are continuously optimized through adversarial training to ultimately generate highly realistic data. The input of the GAN is the multiple meal segmentation images and the multiple meal information, and the output of the GAN is a simulated high-definition image of each meal.

[0026] The simulated HD images of each dish are generated by the generative adversarial network. Compared with the segmented images of dishes, the simulated HD images can more clearly show the appearance, color, weight and other details of the dishes, and make up for the problem of blurred or incomplete information in the segmented images. The simulated HD images can provide more accurate data basis for the subsequent price estimation of dishes.

[0027] Through adversarial training of the generator and the discriminator, the generative adversarial network can learn the feature distribution of dish segmentation images and dish information, and can generate realistic high-definition images based on these features. The generator uses dish segmentation images and dish information as input and generates simulated high-definition images, while the discriminator can evaluate the similarity between the generated images and the real images, thereby continuously optimizing the generator's generation results. Through this mechanism, the generative adversarial network can generate highly realistic dish images, thereby providing high-quality visual data support for subsequent processes.

[0028] Step S4, determining an estimated price range for each meal dish using a price determination model based on the simulated high-definition image of each meal dish; The estimated price range is the possible price range for each dish predicted by the price determination model analysis.

[0029] The price determination model is a deep neural network model, the input of the price determination model is a simulated high-definition image of each meal dish, and the output of the price determination model is an estimated price range for each meal dish.

[0030] A deep neural network model is a machine learning model composed of multiple processing layers. Each layer contains multiple neurons. Each neuron extracts features by performing matrix transformation and nonlinear activation on the data. The parameters used in the matrix transformation are optimized through the training process, so that the model can learn potential patterns and features from complex data. Deep neural network models include deep neural networks (DNNs), which perform well in processing high-dimensional data through their powerful nonlinear fitting capabilities, thereby effectively capturing complex relationships in the data.

[0031] The deep neural network model can extract the visual features of dishes from simulated high-definition images, and based on these features, find multiple dishes similar to the target dish in the existing data. By comprehensively analyzing the prices of these similar dishes, the deep neural network model can determine the estimated price range of the target dish. This prediction method based on the price range of similar dishes not only considers the characteristics of the dishes themselves, but also combines market data. By comprehensively considering various aspects of information, the deep neural network model can more accurately determine the estimated price range.

[0032] In some embodiments, the price determination model includes a dish feature extraction layer, a similar dish matching layer, and a price range estimation layer. The dish feature extraction layer, the similar dish matching layer, and the price range estimation layer all include a deep neural network model. The input of the dish feature extraction layer is a simulated high-definition image of each dining dish, and the output of the dish feature extraction layer is color distribution, texture complexity, shape description, and main ingredient recognition results; the input of the similar dish matching layer is color distribution, texture complexity, and shape description, and the output of the similar dish matching layer is a list of similar dishes, a similarity score, and a historical price range of similar dishes; the input of the price range estimation layer is the main ingredient recognition results, a list of similar dishes, a similarity score, and a historical price range of similar dishes, and the output of the price range estimation layer is the estimated price range of each dining dish.

[0033] Through the dish feature extraction layer, key features such as color distribution, texture complexity, shape description, and main ingredient recognition results can be extracted from the simulated high-definition images of each meal dish. These key features can directly describe the appearance and composition of the dish and provide a basis for subsequent similar dish matching. The similar dish matching layer matches a list of dishes similar to the target dish from a large amount of dish data obtained from the platform based on color distribution, texture complexity, and shape description, and outputs a similarity score and the historical price range of similar dishes. The price range estimation layer combines the main ingredient recognition results, the similar dish list, the similarity score, and the historical price range of similar dishes to ultimately determine the estimated price range for each meal dish.

[0034] By dividing the price determination model into a dish feature extraction layer, a similar dish matching layer, and a price range estimation layer, it is possible to efficiently extract specific features, match similar dishes, and estimate price ranges, thereby improving the accuracy and practicality of the model. This hierarchical structure allows each layer to focus on its specific task, gradually refining and refining information, and ultimately outputting a more reliable estimated price range.

[0035] Step S5, determining the risk level of the meal expense reimbursement voucher based on the estimated price range of each meal dish and the meal expense reimbursement voucher; In some embodiments, Figure 2 A schematic diagram of a process for determining the risk level of a meal expense reimbursement voucher provided in an embodiment of the present invention, wherein determining the risk level of a meal expense reimbursement voucher comprises steps S21 to S23: Step S21, determining a plurality of matching dish information in the meal expense reimbursement voucher using a matching degree determination model based on the plurality of meal dish information and the meal expense reimbursement voucher; The multiple matching dish information in the meal expense reimbursement voucher refers to a set of dish information that is consistent with the voucher content and is determined by comparing the dish data in the meal expense reimbursement voucher with multiple meal dish information through a matching degree determination model.

[0036] Each matching dish information includes dish name, price, quantity and other information.

[0037] The matching degree determination model is a convolutional neural network (CNN), the input of the matching degree determination model is the multiple meal dish information and the meal expense reimbursement voucher, and the output of the matching degree determination model is the multiple matching dish information in the meal expense reimbursement voucher.

[0038] Convolutional neural network is a deep learning model widely used in image processing and feature extraction tasks. Convolutional neural network usually consists of multiple layers, including convolution layer (CONV), rectified linear unit (ReLU) layer, pooling layer (POOL) and fully connected layer (FC). The convolution layer extracts local features of the image through the convolution kernel, the pooling layer reduces the dimension of the features, and the fully connected layer is used to integrate the features and output the results. Convolutional neural network can gradually extract useful features from the image and learn the contextual information of the image, so as to understand and process complex image data.

[0039] For meal expense reimbursement vouchers, the convolutional neural network can extract high-dimensional features in the voucher through the convolutional layer and pooling layer, such as key information such as dish name, price, quantity, etc. For multiple meal dish information, the convolutional neural network can extract the features of each dish and compare them with the features in the voucher. Through the fully connected layer and similarity calculation, the convolutional neural network can quantify the degree of match between the voucher and the meal dish information, and output the matching dish information. By training a large amount of labeled sample data, the convolutional neural network can learn the mapping relationship between voucher information and meal dishes, and accurately predict the matching dish information.

[0040] Step S22, constructing a graph structure, wherein the graph structure includes a plurality of nodes and a plurality of edges between the plurality of nodes, wherein the plurality of nodes include a plurality of meal dish nodes and a plurality of matching dish nodes, wherein each meal dish node establishes an edge with a corresponding matching dish node, the node features of the meal dish nodes include meal dish information and an estimated price range of each meal dish, the node features of the matching dish nodes include matching dish information, and the edge between the meal dish node and the matching dish node is a price difference range between the meal dish and the matching dish; A graph structure is a data structure used to represent entities and their relationships, and is composed of nodes (vertices) and edges (edges). In some embodiments, the nodes in the graph structure include multiple meal dish nodes and multiple matching dish nodes, and each meal dish node is connected to the corresponding matching dish node by an edge. The node features of the meal dish node include meal dish information and the estimated price range of each meal dish, the node features of the matching dish node include matching dish information, and the edge features represent the price difference range between the meal dish and the matching dish. The node features and edge features in the graph structure together constitute a rich information network, which helps to more comprehensively understand the correlation between meal dishes and matching dishes. By constructing a graph structure, the relationship between meal dishes and matching dishes can be systematically represented.

[0041] The price difference between the dining dish and the matching dish is the price range between the dining dish and the matching dish. For example, if the estimated price range of the dining dish is RMB 100 to RMB 120 and the price of the matching dish is RMB 150, then the price difference range is RMB 20 to RMB 50. If the price of the matching dish is within the estimated price range of the dining dish, it is considered that there is no difference.

[0042] Step S23, processing the graph structure based on the graph autoencoder to determine the risk level of the meal expense reimbursement voucher.

[0043] The risk level refers to an indicator that quantitatively evaluates the potential risk of meal expense reimbursement vouchers after processing the graph structure through the graph autoencoder. The risk level reflects the probability and severity of abnormalities or violations that may exist in the reimbursement vouchers.

[0044] Graph Autoencoder (GAE) is a deep learning model specifically designed for processing graph structure data. It can automatically extract the features of nodes and edges, and efficiently process the graph structure through encoding and decoding. The input of the graph autoencoder is the graph structure, and the output of the graph autoencoder is the risk level of the meal expense reimbursement voucher.

[0045] The graph autoencoder first extracts node and edge features through the graph neural network layer, such as meal information, matching dish information, and price difference range. It then updates the features of each node by aggregating the information of neighboring nodes, and further processes the features using fully connected layers and convolutional layers to ultimately generate the risk level of the meal expense reimbursement voucher.

[0046] The graph structure helps to assess the potential risks of meal expense reimbursement vouchers. For example, by analyzing the price difference range between the meal dish node and the matching dish node through the graph autoencoder, it can be determined whether the price in the reimbursement voucher is reasonable and whether there is any false or misreporting. If the price difference range is too large, it may indicate that the reimbursement voucher is abnormal and needs further verification. In addition, the edge features in the graph structure provide direct quantitative indicators for the model, so that the graph structure can more accurately assess the reimbursement risk.

[0047] Step S6: conducting a risk warning based on the risk level of the meal expense reimbursement voucher.

[0048] If the risk level of the meal expense reimbursement voucher is greater than the threshold, manual verification is performed. The threshold can be manually set in advance based on historical data and financial rules. For example, when the risk level exceeds 35%, the system automatically triggers the manual verification process to confirm whether there are abnormal situations such as false reporting, misreporting or duplicate reimbursement. If the risk level is less than or equal to the threshold, the system automatically approves the reimbursement application without manual verification.

[0049] Based on the same inventive concept, Figure 3 A schematic diagram of an enterprise financial management risk early warning system based on data analysis provided by an embodiment of the present invention, wherein the enterprise financial management risk early warning system based on data analysis includes: An acquisition module 31 is used to obtain meal expense reimbursement vouchers and meal videos; An information determination module 32, configured to determine a plurality of meal dish segmentation images and a plurality of meal dish information based on the meal video using a video processing model; An image generation module 33, configured to generate a simulated high-definition image of each meal dish using a generative adversarial network based on the plurality of meal dish segmentation images and the plurality of meal dish information; a price estimation module 34, configured to determine an estimated price range for each meal dish using a price determination model based on the simulated high-definition image of each meal dish; A risk determination module 35, configured to determine the risk of the meal expense reimbursement voucher based on the estimated price range of each meal dish and the meal expense reimbursement voucher; The risk warning module 36 is used to issue a risk warning based on the risk level of the meal expense reimbursement voucher.

[0050] Based on the same inventive concept, an embodiment of the present invention provides an electronic device, such as Figure 4 As shown, it includes: a processor 41; a memory 42; and a computer program; wherein the computer program is stored in the memory 42 and is configured to be executed by the processor 41 to implement the enterprise financial management risk warning method based on data analysis as provided above, the method including: obtaining meal reimbursement vouchers and meal videos; using a video processing model based on the meal video to determine multiple meal dish segmentation images and multiple meal dish information; based on the multiple meal dish segmentation images and the multiple meal dish information, using a generative adversarial network to generate a simulated high-definition image of each meal dish; based on the simulated high-definition image of each meal dish, using a price determination model to determine an estimated price range for each meal dish; based on the estimated price range of each meal dish and the meal reimbursement voucher, determining the risk level of the meal reimbursement voucher; and conducting a risk warning based on the risk level of the meal reimbursement voucher.

[0051] Based on the same inventive concept, this embodiment provides a computer-readable storage medium having a computer program stored thereon, which, when executed by the processor 41, implements the aforementioned enterprise financial management risk warning method based on data analysis, the method comprising: obtaining meal reimbursement vouchers and meal videos; determining multiple meal dish segmentation images and multiple meal dish information based on the meal videos using a video processing model; generating simulated high-definition images of each meal dish based on the multiple meal dish segmentation images and the multiple meal dish information using a generative adversarial network; determining an estimated price range of each meal dish based on the simulated high-definition image of each meal dish using a price determination model; determining the risk level of the meal reimbursement voucher based on the estimated price range of each meal dish and the meal reimbursement voucher; and issuing a risk warning based on the risk level of the meal reimbursement voucher.

[0052] The enterprise financial management risk warning method based on data analysis provided in the embodiment of the present application can be applied to terminal devices (such as mobile phones), tablet computers, laptops, ultra-mobile personal computers (ultra-mobile personal computers, UMPCs), handheld computers, netbooks, personal digital assistants (personal digital assistants, PDAs), wearable devices (such as smart watches, smart glasses or smart helmets, etc.), augmented reality (augmented reality, AR) \ virtual reality (virtual reality, VR) devices, smart home devices, car computers and other electronic devices, and the embodiment of the present application does not impose any restrictions on this.

[0053] The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is only for example and does not constitute a limitation of this specification. Although not explicitly stated here, those skilled in the art may make various modifications, improvements and corrections to this specification. Such modifications, improvements and corrections are suggested in this specification, so such modifications, improvements and corrections still belong to the spirit and scope of the exemplary embodiments of this specification.

[0054] At the same time, this specification uses specific words to describe the embodiments of this specification. For example, "one embodiment", "an embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. In addition, certain features, structures or characteristics in one or more embodiments of this specification can be appropriately combined.

[0055] In addition, unless explicitly stated in the claims, the order of the processing elements and sequences described in this specification, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this specification. Although the above disclosure discusses some invention embodiments that are currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the essence and scope of the embodiments of this specification. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing server or mobile device.

[0056] Similarly, it should be noted that in order to simplify the description disclosed in this specification and thus help understand one or more embodiments of the invention, in the above description of the embodiments of this specification, multiple features are sometimes combined into one embodiment, figure or description thereof. However, this disclosure method does not mean that the features required by the subject matter of this specification are more than the features mentioned in the claims. In fact, the features of the embodiments are less than all the features of the single embodiment disclosed above.

[0057] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

1. The enterprise financial management risk early warning method based on data analysis is characterized by: include: Obtain meal expense reimbursement vouchers and meal videos; Determine a plurality of meal dish segmentation images and a plurality of meal dish information based on the meal video using a video processing model; Based on the plurality of meal dish segmentation images and the plurality of meal dish information, using a generative adversarial network to generate a simulated high-definition image of each meal dish; Determine an estimated price range for each meal dish using a price determination model based on the simulated high-definition image of each meal dish; Determining the risk level of the meal expense reimbursement voucher based on the estimated price range of each meal dish and the meal expense reimbursement voucher; A risk warning is issued based on the risk level of the meal expense reimbursement voucher.

2. The enterprise financial management risk early warning method based on data analysis as claimed in claim 1, characterized in that: The determining of the risk level of the meal expense reimbursement voucher based on the estimated price range of each meal dish and the meal expense reimbursement voucher includes: Determine a plurality of matching dish information in the meal expense reimbursement voucher using a matching degree determination model based on the plurality of meal dish information and the meal expense reimbursement voucher; Constructing a graph structure, wherein the graph structure includes a plurality of nodes and a plurality of edges between the plurality of nodes, the plurality of nodes include a plurality of meal dish nodes and a plurality of matching dish nodes, wherein each meal dish node establishes an edge with a corresponding matching dish node, the node features of the meal dish nodes include meal dish information and an estimated price range of each meal dish, the node features of the matching dish nodes include matching dish information, and the edge between the meal dish node and the matching dish node is a price difference range between the meal dish and the matching dish; The graph structure is processed based on the graph autoencoder to determine the risk level of the meal expense reimbursement voucher.

3. The enterprise financial management risk early warning method based on data analysis as claimed in claim 1, characterized in that: The risk warning based on the risk level of the meal expense reimbursement voucher includes: if the risk level of the meal expense reimbursement voucher is greater than a threshold, manual verification is performed.

4. The enterprise financial management risk early warning method based on data analysis as claimed in claim 1, characterized in that: The video processing model is a Transformer model.

5. An enterprise financial management risk early warning system based on data analysis, characterized in that: include: Acquisition module, used to obtain meal expense reimbursement vouchers and meal videos; An information determination module, configured to determine a plurality of meal dish segmentation images and a plurality of meal dish information based on the meal video using a video processing model; An image generation module, configured to generate a simulated high-definition image of each meal dish using a generative adversarial network based on the plurality of meal dish segmentation images and the plurality of meal dish information; a price estimation module, configured to determine an estimated price range for each meal dish using a price determination model based on the simulated high-definition image of each meal dish; A risk determination module, configured to determine the risk of the meal expense reimbursement voucher based on the estimated price range of each meal dish and the meal expense reimbursement voucher; The risk warning module is used to issue a risk warning based on the risk level of the meal expense reimbursement voucher.

6. The enterprise financial management risk early warning system based on data analysis as claimed in claim 5, characterized in that: The risk determination module is also used for: Determine a plurality of matching dish information in the meal expense reimbursement voucher using a matching degree determination model based on the plurality of meal dish information and the meal expense reimbursement voucher; Constructing a graph structure, wherein the graph structure includes a plurality of nodes and a plurality of edges between the plurality of nodes, the plurality of nodes include a plurality of meal dish nodes and a plurality of matching dish nodes, wherein each meal dish node establishes an edge with a corresponding matching dish node, the node features of the meal dish nodes include meal dish information and an estimated price range of each meal dish, the node features of the matching dish nodes include matching dish information, and the edge between the meal dish node and the matching dish node is a price difference range between the meal dish and the matching dish; The graph structure is processed based on the graph autoencoder to determine the risk level of the meal expense reimbursement voucher.

7. The enterprise financial management risk early warning system based on data analysis as claimed in claim 5, characterized in that: The risk warning based on the risk level of the meal expense reimbursement voucher includes: if the risk level of the meal expense reimbursement voucher is greater than a threshold, manual verification is performed.

8. The enterprise financial management risk early warning system based on data analysis as claimed in claim 5, characterized in that: The video processing model is a Transformer model.

9. An electronic device, characterized in that: include: processor; Memory; And a computer program; wherein, the computer program is stored in the memory and is configured to be executed by the processor to implement the enterprise financial management risk early warning method based on data analysis as described in any one of claims 1 to 4.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the enterprise financial management risk early warning method based on data analysis as described in any one of claims 1 to 4 is implemented.

Images