Financial advertisement content auditing method and device, electronic equipment and storage medium

Abstract

The invention discloses a financial advertisement content auditing method and device, electronic equipment and a storage medium, and relates to the field of financial science and technology or other related technical fields, and the method comprises the steps: obtaining an advertisement image of a to-be-audited target financial advertisement content, and carrying out the preprocessing of the advertisement image; inputting the preprocessed advertisement image into an advertisement image recognition model, and outputting image description information of the advertisement image; inputting the image description information into a content auditing model, and outputting an auditing result; and performing risk assessment on the target financial advertisement content based on the audit result to obtain a risk assessment result, and generating an advertisement content audit report based on the risk assessment result. According to the method and the device, the technical problem of relatively low auditing efficiency of a mode of auditing financial advertisement contents based on manual auditing in related technologies is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of financial technology or other related technical fields, in particular, to a financial advertisement content auditing method and device, electronic equipment and storage medium. BACKGROUND

[0002] As a bridge for financial industry to communicate with customers, financial advertisements carry multiple functions such as spreading brand value, promoting financial products, and guiding market expectations. Under the wave of digitization and informatization, financial outlets not only serve as places for providing physical financial services, but also as important windows for financial institutions to display their products and services. On the one hand, financial advertisements can effectively convey innovative financial products, preferential policies and service concepts of financial institutions, attract potential customers and enhance the loyalty of existing customers. On the other hand, financial advertisements are also an important means for financial institutions to fulfill their social responsibilities. By providing correct financial knowledge, they help the public establish healthy consumption concepts and financial habits, thereby promoting the stable and healthy development of the financial market.

[0003] Since financial advertisements are directly aimed at the general public, the compliance of their content is directly related to the protection of consumers' rights and interests, the fair competition of the financial market, and the reputation risk of financial institutions. Non-compliant advertisements may mislead consumers, harm their interests, trigger legal lawsuits, and even be punished by regulatory agencies. Therefore, strict auditing of financial advertisement content is not only a need for internal management of financial institutions, but also an inevitable requirement for responding to laws and regulations, industry standards and maintaining market order.

[0004] In related technologies, the auditing of financial advertisement content mainly relies on manual auditing. Auditors check the content of advertisements item by item according to the internal advertising guidelines of financial institutions and external laws and regulations to ensure their compliance and accuracy. Although manual auditing can provide a certain degree of flexibility and in-depth understanding, this method is inefficient and difficult to meet the demand in the face of the current large amount of financial advertisements and the rapid iteration and update of financial advertisements.

[0005] To address the above problems, no effective solutions have been proposed so far. SUMMARY

[0006] The embodiments of the present application provide a financial advertisement content auditing method and device, electronic equipment and storage medium to at least solve the technical problem of low auditing efficiency in related technologies based on manual auditing of financial advertisement content.

[0007] According to an aspect of an embodiment of the present application, a financial advertising content review method is provided, comprising: obtaining an advertising image of target financial advertising content to be reviewed, and pre-processing the advertising image; inputting the pre-processed advertising image into an advertising image recognition model to output image description information of the advertising image, wherein the advertising image recognition model is a model pre-constructed for converting an advertising image into text information; inputting the image description information into a content review model to output a review result, wherein the content review model is a model pre-constructed for reviewing whether an advertising content is abnormal; performing risk assessment on the target financial advertising content based on the review result to obtain a risk assessment result, and generating an advertising content review report based on the risk assessment result.

[0008] Further, the step of inputting the pre-processed advertising image into an advertising image recognition model to output image description information of the advertising image comprises: inputting the pre-processed advertising image into a convolution layer of the advertising image recognition model to output each visual feature of the advertising image; inputting the visual feature into a fully connected layer of the advertising image recognition model to output type information and position information of each visual feature; inputting each visual feature, the type information and the position information of the visual feature into a recurrent neural network layer of the advertising image recognition model to output image description information of the advertising image.

[0009] Further, the step of inputting the image description information into a content review model to output a review result comprises: performing word segmentation processing and vector encoding on the image description information to obtain each word segmentation vector; inputting each word segmentation vector into the content review model, identifying and matching each word segmentation vector based on a pre-set rule library through the content review model to obtain a rule review result; inputting each word segmentation vector into the content review model, performing semantic analysis on each word segmentation vector through the content review model to obtain a semantic analysis result; obtaining the review result based on the rule review result and the semantic analysis result, and outputting the review result as output information of the content review model.

[0010] Further, the step of performing risk assessment on the target financial advertising content based on the review result to obtain a risk assessment result comprises: determining abnormal items hit by the target financial advertising content based on the review result; calculating a risk value of the target financial advertising content based on abnormal values and weight values of each abnormal item hit; comparing the risk value with a pre-set risk threshold to obtain a comparison result, determining a risk level of the target financial advertising content based on the comparison result, and obtaining the risk assessment result based on the risk value and the risk level.

[0011] Further, the step of generating an advertisement content review report based on the risk assessment result comprises: in a case where the risk assessment result indicates that the target financial advertisement content has an anomaly, retrieving a knowledge base based on the review result and the risk assessment result to obtain an anomaly adjustment strategy; determining position information of the target financial advertisement content; and generating the advertisement content review report based on the advertisement image, the image description information, the review result, the risk assessment result, the anomaly adjustment strategy, and the position information.

[0012] Further, the step of preprocessing the advertisement image comprises: denoising and image enhancement processing the advertisement image; and standardizing and normalizing the advertisement image after image enhancement processing.

[0013] Further, the step of constructing the advertisement image recognition model comprises: collecting advertisement images of historical advertisement content, preprocessing the historical advertisement content, and generating description labels for the preprocessed advertisement images; generating a training sample set based on the advertisement images and the description labels, and dividing the training sample set to obtain a training set and a test set; selecting a deep learning model architecture as an initial advertisement image recognition model, and performing parameter configuration and network layer configuration on the initial advertisement image recognition model; iteratively training the initial advertisement image recognition model based on the training set to obtain a trained advertisement image recognition model; and testing the trained advertisement image recognition model based on the test set, and obtaining a final advertisement image recognition model in a case where the testing is passed.

[0014] According to another aspect of the embodiment of the present application, a financial advertisement content review device is also provided, comprising: an acquisition unit configured to acquire an advertisement image of target financial advertisement content to be reviewed, and to preprocess the advertisement image; an identification unit configured to input the preprocessed advertisement image into an advertisement image recognition model, and to output image description information of the advertisement image, wherein the advertisement image recognition model is a model pre-constructed for converting an advertisement image into text information; a review unit configured to input the image description information into a content review model, and to output a review result, wherein the content review model is a model pre-constructed for reviewing whether an advertisement content has an anomaly; and a generation unit configured to perform risk assessment on the target financial advertisement content based on the review result to obtain a risk assessment result, and to generate an advertisement content review report based on the risk assessment result.

[0015] Further, the recognition unit comprises: a first output module, configured to input the preprocessed advertisement image into a convolution layer of the advertisement image recognition model, and output respective visual features of the advertisement image; a second output module, configured to input the visual features into a fully connected layer of the advertisement image recognition model, and output type information and position information of each of the visual features; and a third output module, configured to input each of the visual features, the type information and the position information of the visual features into a recurrent neural network layer of the advertisement image recognition model, and output image description information of the advertisement image.

[0016] Further, the auditing unit comprises: a first processing module, configured to perform word segmentation processing and vector encoding on the image description information, to obtain respective segmentation vectors; a first auditing module, configured to input each of the segmentation vectors into the content auditing model, and perform identification and matching on each of the segmentation vectors based on a preset rule library through the content auditing model, to obtain a rule auditing result; a second auditing module, configured to input each of the segmentation vectors into the content auditing model, and perform semantic analysis on each of the segmentation vectors through the content auditing model, to obtain a semantic analysis result; and a fourth output module, configured to obtain the auditing result based on the rule auditing result and the semantic analysis result, and output the auditing result as output information of the content auditing model.

[0017] Further, the generation unit comprises: a first determination module, configured to determine, based on the auditing result, an abnormal item hit by the target financial advertisement content; a first calculation module, configured to calculate a risk value of the target financial advertisement content based on an abnormal value and a weight value of each of the abnormal items hit; and a first comparison module, configured to compare the risk value with a preset risk threshold, to obtain a comparison result, determine a risk level of the target financial advertisement content based on the comparison result, and obtain the risk evaluation result based on the risk value and the risk level.

[0018] Further, the generation unit further comprises: a first retrieval module, configured to, in a case where the risk evaluation result indicates that the target financial advertisement content has an abnormality, retrieve a knowledge base based on the auditing result and the risk evaluation result, and acquire an abnormality adjustment strategy; a second determination module, configured to determine position information of the target financial advertisement content; and a first generation module, configured to generate the advertisement content auditing report based on the advertisement image, the image description information, the auditing result, the risk evaluation result, the abnormality adjustment strategy and the position information.

[0019] Further, the acquisition unit comprises: a second processing module, configured to perform denoising processing and image enhancement processing on the advertisement image; and a third processing module, configured to perform standardization processing and normalization processing on the advertisement image after the image enhancement processing.

[0020] Further, the financial advertisement content auditing apparatus further comprises: a first acquisition module, configured to acquire an advertisement image of historical advertisement content, and perform preprocessing on the historical advertisement content to generate a description label for the preprocessed advertisement image; a second generation module, configured to generate a training sample set based on the advertisement image and the description label, and divide the training sample set to obtain a training set and a test set; a first selection module, configured to select a deep learning model architecture as an initial advertisement image recognition model, and perform parameter configuration and network layer configuration on the initial advertisement image recognition model; a first training module, configured to perform iterative training on the initial advertisement image recognition model based on the training set to obtain a trained advertisement image recognition model; and a first test module, configured to test the trained advertisement image recognition model based on the test set, and obtain a final advertisement image recognition model in the case of passing the test.

[0021] According to another aspect of the embodiments of the present application, a computer readable storage medium is also provided, which comprises a stored computer program, wherein the computer readable storage medium controls a device where the computer readable storage medium is located to perform any of the above financial advertisement content auditing methods when the computer program runs.

[0022] According to another aspect of the embodiments of the present application, an electronic device is also provided, which comprises one or more processors and a memory, and the memory is configured to store one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement any of the above financial advertisement content auditing methods.

[0023] In the present application, by the following steps: acquiring an advertisement image of target financial advertisement content to be audited, and preprocessing the advertisement image, then inputting the preprocessed advertisement image into an advertisement image recognition model to output image description information of the advertisement image, wherein the advertisement image recognition model is a model pre-constructed for converting the advertisement image into text information, then inputting the image description information into a content auditing model to output an auditing result, wherein the content auditing model is a model pre-constructed for auditing whether the advertisement content is abnormal, and finally performing risk assessment on the target financial advertisement content based on the auditing result to obtain a risk assessment result, and generating an advertisement content auditing report based on the risk assessment result.

[0024] In the present application, the advertisement image of the financial advertisement content is collected in real time, and based on the pre-trained advertisement image recognition model, the advertisement image is converted into textual description information, realizing the accurate recognition of various key elements in the advertisement image, and then the description information is taken as the input data of the content review model to automatically check whether the advertisement content meets the requirements. The purpose of automatically reviewing the financial advertisement content is achieved, and the technical effect of improving the efficiency of the financial advertisement content review is achieved. Further, the technical problem of low review efficiency in the related art based on manual review of the financial advertisement content is solved. BRIEF DESCRIPTION OF DRAWINGS

[0025] The drawings described herein are used to provide a further understanding of the present application, constitute a part of the present application, the schematic embodiments of the present application and the description thereof are used to explain the present application, and do not constitute an improper limitation on the present application. In the drawings:

[0026] Figure 1 A hardware structure block diagram of a computer terminal (or mobile device) for implementing a financial advertisement content review method is shown;

[0027] Figure 2 is a flowchart of an optional financial advertisement content review method according to an embodiment of the present application;

[0028] Figure 3 is a schematic diagram of an optional financial advertisement content review process according to an embodiment of the present application;

[0029] Figure 4 is a schematic diagram of an optional financial advertisement content review device according to an embodiment of the present application;

[0030] Figure 5 is a hardware structure block diagram of an electronic device (or mobile device) for executing a financial advertisement content review method according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to make the person in the art better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by the person in the art without creative labor should belong to the scope of protection of the present application.

[0032] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0033] It should be noted that the financial advertising content review method and device in this application can be used in the field of financial technology to automatically review the content of financial advertisements, and can also be used in any field other than the field of financial technology to automatically review the content of financial advertisements. This application does not limit the application field of the financial advertising content review method and device.

[0034] It should be noted that the collected information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in this application are information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation portals for users to choose to authorize or refuse. For example, an interface is set up between this system and relevant users or institutions to provide users with corresponding operation portals for users to choose to agree or refuse the automated decision-making results; if the user chooses to refuse, the expert decision-making process will be entered.

[0035] The following embodiments of the present invention can be applied to various financial advertising content review systems, applications, and devices. This invention uses an image recognition model to convert advertisement images into textual descriptions, and then uses a content review model to analyze these textual descriptions to determine whether the financial advertisement content is compliant. This enables automated review of financial advertisement content, improves review efficiency, and reduces compliance risks.

[0036] The present invention will be described in detail below with reference to various embodiments.

[0037] Example 1

[0038] According to the embodiments of the present application, an embodiment of a financial advertising content auditing method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0039] The method embodiment provided by the first embodiment of the present application can be executed in a mobile terminal, a computer terminal or similar computing device. Figure 1 A hardware structure block diagram of a computer terminal (or mobile device) for implementing the financial advertising content auditing method is shown. As shown in the figure, Figure 1 The computer terminal 10 (or mobile device) can include one or more processors 102 (the processor 102 can include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it can also include a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which can be included as one of the ports in the BUS bus), a network interface, a power supply and / or a camera. Those skilled in the art can understand that, Figure 1 The structure shown is only schematic, which does not limit the structure of the above-mentioned electronic device. For example, the computer terminal 10 can include more or less components than those shown in Figure 1 or have a different configuration than that shown in Figure 1 .

[0040] It should be noted that the one or more processors 102 and / or other data processing circuits described above can be referred to herein as "data processing circuits" in general. The data processing circuit can be embodied in whole or in part as software, hardware, firmware or any other combination. In addition, the data processing circuit can be a single independent processing module, or any one of the other elements combined into the computer terminal 10 (or mobile device) in whole or in part. As referred to in the embodiments of the present application, the data processing circuit serves as a processor control (for example, selection of a variable resistance terminal path connected to an interface).

[0041] The memory 104 can be used to store software programs of application software and modules, such as program instructions / data storage devices corresponding to the financial advertising content auditing method in the embodiments of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, implements the financial advertising content auditing method described above. The memory 104 can include a high-speed random access memory, and can also include a non-volatile memory, such as one or more magnetic storage devices, flash memories, or other non-volatile solid-state memories. In some examples, the memory 104 can further include memories remotely arranged with respect to the processor 102, which can be connected to the computer terminal 10 through a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0042] The transmission device 106 is used to receive or send data via a network. Specific examples of the above-mentioned network can include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to be able to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (Radio Frequency, RF) module, which is used to communicate with the Internet in a wireless manner.

[0043] The display can be, for example, a touch screen type liquid crystal display (LCD), which can enable a user to interact with the user interface of the computer terminal 10 (or mobile device).

[0044] Under the above-mentioned operating environment, the present application provides a financial advertising content auditing method as shown in Figure 2 The implementation subject of the method is a financial advertising content auditing system.

[0045] Figure 2 is a flowchart of an optional financial advertising content auditing method according to an embodiment of the present application, as shown in Figure 2 The method includes the following steps:

[0046] In step S201, an advertisement image of target financial advertising content to be audited is obtained, and the advertisement image is preprocessed.

[0047] In the above-mentioned step S201, the financial advertising is generally deployed at a specific location such as a financial site, and the financial advertising image of the target area is captured by the image acquisition device at a regular time. The advertisement image of the target financial advertising content to be audited is obtained based on the obtained image, and the image is preprocessed, which can specifically include image enhancement processing, denoising processing, and standardization processing.

[0048] Further, the step of pre-processing the advertisement image includes: performing denoising processing and image enhancement processing on the advertisement image; performing standardization processing and normalization processing on the advertisement image after the image enhancement processing.

[0049] Specifically, the pre-processing operation includes denoising processing, which can use median filtering, bilateral filtering or deep learning-based denoising algorithms to reduce random noise and interference in the advertisement image. The purpose of denoising processing is to improve image quality and ensure that the model can accurately identify key information and avoid recognition errors caused by noise. At the same time, histogram equalization, adaptive contrast enhancement and other technologies are used to optimize the brightness, contrast and color of the image. The purpose of image enhancement is to make the details in the image more obvious and improve the recognition of text and graphical elements, which is convenient for subsequent image description generation. The enhanced advertisement image is adjusted to a preset size and format, for example, uniformly adjusted to 224x224 pixels in RGB format. Standardization processing ensures that all images have consistency when inputting the model, so that the model can more effectively process image data and avoid performance fluctuations caused by differences in image size or format. Normalization processing is used to scale the pixel values of image data to a fixed range, such as between 0 and 1, to standardize the distribution of input data. This helps to improve the stability of model training and the accuracy of recognition, and prevents learning bias caused by differences in data distribution.

[0050] In some optional embodiments, image recognition technology and image segmentation technology are used to preliminarily identify the image, identify the text area, logo area, character image area, product image area, etc. in the advertisement image, thereby segmenting the advertisement image to obtain image areas containing different types of elements.

[0051] In step S202, the pre-processed advertisement image is input into the advertisement image recognition model, and the image description information of the advertisement image is output.

[0052] The advertisement image recognition model is a model pre-constructed for converting the advertisement image into text information.

[0053] In the above step S202, the pre-processed advertisement image is input as input data into the advertisement image recognition model, and the advertisement image recognition model is a model pre-constructed for converting the advertisement image into text information. The pre-processed advertisement image is identified using a deep learning architecture, such as a convolutional neural network combined with a recurrent neural network or a neural network model based on natural language processing. The advertisement image recognition model is trained on a large amount of image data and can understand and describe various elements contained in the image, such as text, graphics, product information, etc. The model first extracts features from the image, then generates a descriptive text sequence through the decoder part, and finally outputs detailed image description information.

[0054] The information output by the advertisement image recognition model can be one or more texts describing the content of the advertisement image, including product names, slogans, character activities, and other key details appearing in the image. These description information will be the direct basis for subsequent content review, ensuring direct conversion from image to text description, and providing clear and structured text input for subsequent review.

[0055] Further, the steps of constructing the advertisement image recognition model include: collecting advertisement images of historical advertisement content, and preprocessing the historical advertisement content to generate description labels for the preprocessed advertisement images; generating a training sample set based on the advertisement images and the description labels, and dividing the training sample set to obtain a training set and a test set; selecting a deep learning model architecture as an initial advertisement image recognition model, and configuring parameters and network layers for the initial advertisement image recognition model; iteratively training the initial advertisement image recognition model based on the training set to obtain a trained advertisement image recognition model; and testing the trained advertisement image recognition model based on the test set, and obtaining a final advertisement image recognition model in the case of passing the test.

[0056] Specifically, the advertisement image recognition model is pre-trained through iteration. The steps of iteration include: sample data construction, a large number of historical advertisement images are obtained from a financial database. These images should cover various types and styles of advertisements to ensure that the model can adapt to multiple scenarios. The collected historical advertisement images are preprocessed, such as denoising, image enhancement, standardization and normalization, to ensure that the image quality meets the requirements of model training and reduces the uncertainty of image processing. The preprocessed advertisement images are manually labeled to generate detailed description labels, including text, graphical elements, product information, character activities and other key information in the image. The preprocessed advertisement images are paired with the corresponding description labels to construct a training sample set for advertisement image recognition, and then the training sample set is randomly divided into a training set and a test set according to a predetermined division ratio.

[0057] Model initialization and configuration, a deep learning architecture is selected as an initial advertisement image recognition model, for example, a composite architecture combining an image feature extraction model and a text description generation model can be used. The network architecture of the initial advertisement image recognition model can include an input layer, a convolutional layer, an attention layer, a decoding layer and an output layer to ensure that the model can effectively process image data and generate text descriptions. According to the image size and description label length after preprocessing, the hyperparameters of the initial advertisement image recognition model are set, such as learning rate, batch size, optimizer type, etc., to optimize the model training process.

[0058] Model training and testing, using the training set to iteratively train the initial advertisement image recognition model, adjusting the model parameters through the back propagation algorithm to minimize the difference between the model output and the description label, and improving the recognition accuracy of the model. Based on the test set, the trained advertisement image recognition model is tested to evaluate the performance of the model on unseen image data, ensuring the generalization ability of the model. If the performance of the model on the test set meets the preset accuracy and stability standards, the model test is considered to pass, and the model can be used as the final advertisement image recognition model for actual advertisement content review.

[0059] Further, the step of inputting the preprocessed advertisement image into the advertisement image recognition model and outputting the image description information of the advertisement image includes: inputting the preprocessed advertisement image into the convolutional layer of the advertisement image recognition model to output the visual features of the advertisement image; inputting the visual features into the fully connected layer of the advertisement image recognition model to output the type information and position information of the visual features; inputting the visual features, the type information and the position information of the visual features into the recurrent neural network layer of the advertisement image recognition model to output the image description information of the advertisement image.

[0060] Specifically, when identifying based on the advertisement image recognition model, the preprocessed advertisement image is input into the convolutional layer of the advertisement image recognition model. The convolutional layer scans the image through a series of filters, automatically extracts local features in the image such as edges, textures, color patterns, etc., and forms a visual feature representation of the image. These features are crucial for understanding the image content and are the basis for subsequent recognition and description. The visual features output by the convolutional layer are passed to the fully connected layer. This layer is responsible for integrating these features, mapping each visual feature to its possible type information and position information through weight matrix operations. Type information refers to the specific category to which the feature belongs, such as "person", "number" or "product"; position information describes the relative position of the feature in the image, which is very important for understanding the content layout of the image. After combining each visual feature, type information and position information, they are sent to the recurrent neural network layer of the model. The recurrent neural network layer can generate new output based on previous output, and gradually generate natural language text describing the content of the advertisement image according to the input feature sequence.

[0061] Step S203, inputting the image description information into the content review model and outputting the review result.

[0062] In the step S203 described above, the image description information is input as input data to a content review model, which is a model pre-constructed for reviewing whether the advertising content is abnormal. The content review model reviews the target financial advertising content according to various review strategies such as keyword matching, semantic understanding, sentiment analysis, judges whether the advertising content meets the preset review standards including compliance, morality and copyright and the like, and thus obtains a review result.

[0063] Further, the step of inputting the image description information to the content review model and outputting the review result includes: performing word segmentation processing and vector encoding on the image description information to obtain each word segmentation vector; inputting each word segmentation vector to the content review model, identifying and matching each word segmentation vector based on a preset rule library through the content review model to obtain a rule review result; inputting each word segmentation vector to the content review model, performing semantic analysis on each word segmentation vector through the content review model to obtain a semantic analysis result; obtaining the review result based on the rule review result and the semantic analysis result, and outputting the review result as output information of the content review model.

[0064] Specifically, when analyzing the image description information based on the content review model, first, the text description information is subjected to word segmentation processing, and continuous text description is decomposed into single words or phrases, which helps the model to understand the structure and word meaning of the text, so as to perform more accurate semantic analysis. The words after word segmentation are further encoded into vector form, so as to embed the semantic information of the words into numerical vectors, facilitating machine learning and deep learning model processing. Word embedding models can be used for vector encoding to capture the meaning of words in context and improve the accuracy of subsequent analysis.

[0065] The encoded word segmentation vectors are input into the content review model, and the model identifies and matches the word segmentation vectors using a preset rule library. The rule library contains all rules for compliance and risk control, such as a list of prohibited words, patterns of sensitive information, legality requirements for advertising content description, etc. Through identification and matching, the model can determine whether the image description information violates any rules, thereby obtaining a rule review result.

[0066] The content review model not only performs review based on rules, but also performs deeper semantic analysis on the word segmentation vectors. That is, the internal structure of the model (such as a multi-layer neural network) is used to understand the relationship between words, contextual information and potential semantic meaning. Semantic analysis can capture content that may not directly violate rules, but still has potential risks or non-compliance implications. For example, whether the advertising content contains discriminatory content, whether it involves false propaganda, etc. Finally, the final review result is output according to the rule review result and the semantic analysis result.

[0067] Step S204, based on the audit result, the risk of the target financial advertising content is evaluated, the risk evaluation result is obtained, and the advertising content audit report is generated based on the risk evaluation result.

[0068] In the above step S204, the audit result contains the compliance state of the advertising content, the hit rules and the existing abnormal content. According to the audit result, the target financial advertising content can be further risk evaluated, and the advertising content audit report can be generated based on the risk evaluation result. The risk evaluation can include determining whether there is a risk, a risk category, a risk level, etc. Finally, a pre-established report template is called, and the results of the audit are added to the report template to form the report of this advertising content audit.

[0069] Further, based on the audit result, the risk of the target financial advertising content is evaluated, and the risk evaluation result is obtained. The step includes: determining the abnormal items hit by the target financial advertising content based on the audit result; calculating the risk value of the target financial advertising content based on the abnormal value and the weight value of each abnormal item hit; comparing the risk value with the preset risk threshold to obtain the comparison result, determining the risk level of the target financial advertising content based on the comparison result, and obtaining the risk evaluation result based on the risk value and the risk level.

[0070] Specifically, in the risk evaluation, first, based on the audit result output by the content audit model, the abnormal items hit in the target financial advertising content are determined. These abnormal items may include but are not limited to illegal language, sensitive information, copyright issues, moral controversy, etc., which are key factors affecting the compliance and risk of the advertisement. For each abnormal item hit, a corresponding abnormal value and weight value are assigned. The abnormal value is used to quantify the severity of the abnormality, and the weight value reflects the importance of the abnormal type in risk evaluation, which is usually based on industry standards, laws and regulations, and the risk preference of the financial institution itself. According to the abnormal value and the weight value of each abnormal item hit, the overall risk value of the target financial advertising content is calculated by weighted summation. The calculation of the risk value aims to evaluate the overall compliance risk level of the advertising content. The calculated risk value is compared with the preset risk threshold, and different risk levels are divided by multiple risk thresholds. The risk threshold can be dynamically adjusted according to the risk management requirements. Based on the comparison result of the risk value and the risk threshold, the risk level of the target financial advertising content is determined. Common risk level classification includes low risk, medium risk and high risk, each level of risk corresponds to different processing strategies and subsequent review processes.

[0071] Further, the step of generating an advertisement content review report based on the risk assessment result includes: in the case that the risk assessment result indicates that the target financial advertisement content has an abnormality, retrieving a knowledge base based on the review result and the risk assessment result to obtain an abnormality adjustment strategy; determining the location information of the target financial advertisement content; and generating the advertisement content review report based on the advertisement image, the image description information, the review result, the risk assessment result, the abnormality adjustment strategy, and the location information.

[0072] Specifically, when the risk assessment result indicates that the target financial advertisement content has an abnormality, an abnormality processing program is started. The system retrieves the corresponding abnormality adjustment strategy from the pre-set knowledge base according to the specific type of the abnormality and the risk level. The knowledge base is an information base that stores various possible abnormal situations and corresponding processing schemes, such as modifying sensitive words, adjusting picture elements, adding legal statements, etc., to ensure the compliance of the advertisement content and reduce risks. In the abnormality processing process, the system also needs to determine the accurate location information of the target financial advertisement content, including the geographical location of the advertisement point, the specific advertisement position identifier, etc., which helps to accurately locate the subsequent adjustment or replacement work and avoid misoperation. The advertisement image, the image description information, the review result, the risk assessment result, and the abnormality adjustment strategy retrieved from the knowledge base are integrated with the location information of the target advertisement to form the key content of the report. The integration of these information ensures the comprehensiveness and effectiveness of the report, providing detailed data support for decision makers. Based on the integrated information, the system automatically generates an advertisement content review report, which clearly lists the abnormality of the advertisement, the risk level, the recommended adjustment strategy, and the specific location information. The structured and standardized report makes information transmission more efficient, facilitating quick understanding and response.

[0073] Through the above steps, the advertisement image of the target financial advertisement content to be reviewed is obtained, and the advertisement image is preprocessed, and then the preprocessed advertisement image is input into an advertisement image recognition model to output image description information of the advertisement image, wherein the advertisement image recognition model is a model pre-constructed for converting the advertisement image into text information, then the image description information is input into a content review model to output a review result, wherein the content review model is a model pre-constructed for reviewing whether the advertisement content has an abnormality, and finally the target financial advertisement content is risk assessed based on the review result to obtain a risk assessment result, and an advertisement content review report is generated based on the risk assessment result.

[0074] In this embodiment, the advertisement image of the financial advertisement content is collected in real time, and based on the pre-trained advertisement image recognition model, the advertisement image is converted into textual description information, the accurate recognition of various key elements in the advertisement image is realized, and then the description information is taken as the input data of the content review model to automatically check whether the advertisement content meets the requirements. The purpose of automatically reviewing the financial advertisement content is achieved, and the technical effect of improving the efficiency of the financial advertisement content review is achieved. Further, the technical problem of low efficiency of the financial advertisement content review based on manual review in the related art is solved.

[0075] The following will be described in detail in combination with another optional specific embodiment.

[0076] Figure 3 is a schematic diagram of an optional financial advertisement content review process according to an embodiment of the application, as Figure 3 shown, the financial advertisement content review process includes:

[0077] Step one, start;

[0078] Step two, image acquisition and preprocessing;

[0079] The image of the advertisement position is periodically taken by a monitoring camera or other image acquisition device. The collected image is preprocessed, including image enhancement, denoising, size standardization, etc., to improve the accuracy of subsequent identification.

[0080] Step three, image description generation;

[0081] The preprocessed image is analyzed by using a pre-constructed image recognition model to generate a detailed image description, including identifying the text, graphical elements, product information, and character image in the advertisement. The image recognition model is based on a deep learning architecture and can identify and convert the advertisement image to obtain textual description information.

[0082] Step four, image content review;

[0083] The generated image description is matched with rules and analyzed by a pre-constructed content review model to review whether the advertisement content meets business specifications, legal and regulatory requirements, moral standards, etc. For example, whether the advertisement involves false propaganda, whether it contains discriminatory content, whether it infringes intellectual property rights, etc.

[0084] Step five, risk assessment;

[0085] According to the review result, the risk level of the advertisement content is evaluated, such as low risk, medium risk, and high risk. The risk level can be set based on the severity of the violation and the possible negative impact.

[0086] Step six, review report generation;

[0087] The automatically generated advertising content delivery audit report at least includes advertising position, advertising content description, audit result, risk level, recommended measures and the like.

[0088] Step seven, abnormal early warning processing;

[0089] For high-risk advertising content, the system immediately triggers an early warning mechanism to notify relevant departments for manual review and emergency processing, such as suspending advertising delivery, modifying content, etc.

[0090] Step eight, end.

[0091] The embodiment of the application converts the advertising image into text description information through the image recognition model, and analyzes the text description information through the content audit model to determine whether the financial advertising content is in compliance, thereby realizing automatic auditing of the financial advertising content, improving the auditing efficiency, and reducing the compliance risk.

[0092] The following will be described in detail in conjunction with another embodiment.

[0093] Embodiment two

[0094] The financial advertising content auditing device provided in the embodiment includes a plurality of implementation units, each implementation unit corresponding to each implementation step in the above embodiment one, and the specific implementation manner and beneficial effects thereof can refer to the aforementioned method embodiment, which will not be described here again.

[0095] Figure 4 is a schematic diagram of an optional financial advertising content auditing device according to the embodiment of the application, as shown in Figure 4 The financial advertising content auditing device can include an acquisition unit 41, an identification unit 42, an auditing unit 43, and a generation unit 44, wherein

[0096] The acquisition unit 41 is configured to acquire an advertising image of a target financial advertising content to be audited, and to pre-process the advertising image;

[0097] The identification unit 42 is configured to input the pre-processed advertising image into an advertising image recognition model, and to output image description information of the advertising image, wherein the advertising image recognition model is a model pre-constructed for converting the advertising image into text information;

[0098] The auditing unit 43 is configured to input the image description information into a content audit model, and to output an audit result, wherein the content audit model is a model pre-constructed for auditing whether the advertising content has an abnormality;

[0099] The generating unit 44 is configured to perform risk assessment on the target financial advertisement content based on the audit result, obtain a risk assessment result, and generate an advertisement content audit report based on the risk assessment result.

[0100] The financial advertisement content audit device can obtain the advertisement image of the target financial advertisement content to be audited through the obtaining unit 41 and pre-process the advertisement image. The pre-processed advertisement image is input into an advertisement image recognition model through the recognition unit 42, and image description information of the advertisement image is output. The advertisement image recognition model is a model pre-constructed for converting the advertisement image into text information. The image description information is input into a content audit model through the audit unit 43, and an audit result is output. The content audit model is a model pre-constructed for auditing whether the advertisement content is abnormal. The generating unit 44 performs risk assessment on the target financial advertisement content based on the audit result, obtains a risk assessment result, and generates an advertisement content audit report based on the risk assessment result.

[0101] In this embodiment, the advertisement image of the financial advertisement content is collected in real time, and the advertisement image recognition model pre-trained is used to convert the advertisement image into text description information, so as to accurately identify various key elements in the advertisement image. Then, the description information is used as the input data of the content audit model to automatically check whether the advertisement content meets the requirements. The purpose of automatically auditing the financial advertisement content is achieved, and the technical effect of improving the efficiency of auditing the financial advertisement content is achieved. Furthermore, the technical problem of low auditing efficiency in the related art caused by manually auditing the financial advertisement content is solved.

[0102] Further, the recognition unit includes a first output module configured to input the pre-processed advertisement image into a convolution layer of the advertisement image recognition model and output each visual feature of the advertisement image, a second output module configured to input the visual feature into a fully connected layer of the advertisement image recognition model and output type information and position information of each visual feature, and a third output module configured to input each visual feature, the type information and the position information of the visual feature into a recurrent neural network layer of the advertisement image recognition model and output image description information of the advertisement image.

[0103] Further, the auditing unit comprises: a first processing module, configured to perform word segmentation processing and vector coding on the image description information to obtain respective word segmentation vectors; a first auditing module, configured to input the respective word segmentation vectors into the content auditing model, and perform identification and matching on the respective word segmentation vectors based on a preset rule library through the content auditing model to obtain a rule auditing result; a second auditing module, configured to input the respective word segmentation vectors into the content auditing model, and perform semantic analysis on the respective word segmentation vectors through the content auditing model to obtain a semantic analysis result; and a fourth output module, configured to obtain an auditing result based on the rule auditing result and the semantic analysis result, and output the auditing result as output information of the content auditing model.

[0104] Further, the generating unit comprises: a first determining module, configured to determine, based on the auditing result, an abnormal item hit by the target financial advertising content; a first calculating module, configured to calculate a risk value of the target financial advertising content based on abnormal values and weight values of the respective abnormal items hit; and a first comparing module, configured to compare the risk value with a preset risk threshold to obtain a comparison result, determine a risk level of the target financial advertising content based on the comparison result, and obtain a risk assessment result based on the risk value and the risk level.

[0105] Further, the generating unit further comprises: a first retrieving module, configured to retrieve a knowledge base based on the auditing result and the risk assessment result to obtain an abnormal adjustment strategy in a case where the risk assessment result indicates that the target financial advertising content has an abnormality; a second determining module, configured to determine position information of the target financial advertising content; and a first generating module, configured to generate an advertising content auditing report based on the advertising image, the image description information, the auditing result, the risk assessment result, the abnormal adjustment strategy, and the position information.

[0106] Further, the obtaining unit comprises: a second processing module, configured to perform denoising processing and image enhancement processing on the advertising image; and a third processing module, configured to perform standardization processing and normalization processing on the advertising image after the image enhancement processing.

[0107] Furthermore, the financial advertising content review device also includes: a first acquisition module, used to collect advertising images of historical advertising content, and pre-process the historical advertising content to generate description labels for the pre-processed advertising images; a second generation module, used to generate a training sample set based on the advertising images and description labels, and divide the training sample set to obtain a training set and a test set; a first selection module, used to select a deep learning model architecture as an initial advertising image recognition model, and perform parameter configuration and network layer configuration for the initial advertising image recognition model; a first training module, used to iteratively train the initial advertising image recognition model based on the training set to obtain a trained advertising image recognition model; a first testing module, used to test the trained advertising image recognition model based on the test set, and obtain the final advertising image recognition model if the test passes.

[0108] It should be noted that the acquisition unit 41, identification unit 42, review unit 43, and generation unit 44 correspond to steps S201 to S204 in the first embodiment. The examples and application scenarios implemented by the above units and corresponding steps are the same, but are not limited to the contents disclosed in the first embodiment. It should be noted that the above modules or units can be hardware components or software components stored in a memory (e.g., memory 104) and processed by one or more processors (e.g., processors 102a, 102b, ..., 102n). The above modules or units can also be part of a device and can be run in the computer terminal 10 provided in the first embodiment.

[0109] The present invention is described below in conjunction with another optional embodiment.

[0110] Example 3

[0111] An embodiment of the present invention may further provide an electronic device, Figure 5 FIG. 1 is a hardware structure block diagram of an electronic device (or mobile device) for performing an optional method for reviewing financial advertisement content according to an embodiment of the present invention. Figure 5 As shown, the electronic device may include: one or more ( Figure 5 Only one is shown) processor 502, memory 504, storage controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module and display.

[0112] The memory can be configured to store software programs and modules, such as program instructions / modules corresponding to the method and device in the embodiments of the present application. The processor executes various functions and data processing by running the software programs and modules stored in the memory, that is, implements the above method. The memory can include a high-speed random access memory, and can further include a non-volatile memory, such as one or more magnetic storage devices, flash memories, or other non-volatile solid-state memories. In some examples, the memory can further include a memory remotely arranged with respect to the processor, which can be connected to the terminal through a network. Examples of the above network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0113] The processor can call information and application programs stored in the memory through the transmission device to perform the following steps: obtaining an advertisement image of a target financial advertisement content to be audited, and preprocessing the advertisement image; inputting the preprocessed advertisement image into an advertisement image recognition model to output image description information of the advertisement image, wherein the advertisement image recognition model is a model pre-constructed for converting the advertisement image into text information; inputting the image description information into a content audit model to output an audit result, wherein the content audit model is a model pre-constructed for auditing whether the advertisement content is abnormal; performing risk assessment on the target financial advertisement content based on the audit result to obtain a risk assessment result, and generating an advertisement content audit report based on the risk assessment result.

[0114] The processor can also call information and application programs stored in the memory through the transmission device to perform the following steps: inputting the preprocessed advertisement image into a convolution layer of the advertisement image recognition model to output each visual feature of the advertisement image; inputting the visual feature into a fully connected layer of the advertisement image recognition model to output type information and position information of each visual feature; inputting each visual feature, type information and position information of the visual feature into a recurrent neural network layer of the advertisement image recognition model to output image description information of the advertisement image.

[0115] The processor can also call information and application programs stored in the memory through the transmission device to perform the following steps: performing word segmentation processing and vector encoding on the image description information to obtain each word segmentation vector; inputting each word segmentation vector into the content audit model, identifying and matching each word segmentation vector based on a pre-set rule library through the content audit model to obtain a rule audit result; inputting each word segmentation vector into the content audit model, performing semantic analysis on each word segmentation vector through the content audit model to obtain a semantic analysis result; obtaining an audit result based on the rule audit result and the semantic analysis result, and outputting the audit result as output information of the content audit model.

[0116] The processor can also call information and application programs stored in the memory through the transmission device to perform the following steps: determining abnormal items hit by the target financial advertising content based on the review result; calculating a risk value of the target financial advertising content based on the abnormal values and weight values of the hit abnormal items; comparing the risk value with a preset risk threshold to obtain a comparison result, determining a risk level of the target financial advertising content based on the comparison result, and obtaining a risk assessment result based on the risk value and the risk level.

[0117] The processor can also call information and application programs stored in the memory through the transmission device to perform the following steps: in the case that the risk assessment result indicates that the target financial advertising content is abnormal, retrieving a knowledge base based on the review result and the risk assessment result to obtain an abnormal adjustment strategy; determining position information of the target financial advertising content; and generating an advertising content review report based on the advertising image, the image description information, the review result, the risk assessment result, the abnormal adjustment strategy, and the position information.

[0118] The processor can also call information and application programs stored in the memory through the transmission device to perform the following steps: performing denoising processing and image enhancement processing on the advertising image; and performing standardization processing and normalization processing on the advertising image after the image enhancement processing.

[0119] The processor can also call information and application programs stored in the memory through the transmission device to perform the following steps: collecting advertising images of historical advertising content, and pre-processing the historical advertising content to generate description labels for the pre-processed advertising images; generating a training sample set based on the advertising images and the description labels, and dividing the training sample set to obtain a training set and a test set; selecting a deep learning model architecture as an initial advertising image recognition model, and performing parameter configuration and network layer configuration for the initial advertising image recognition model; iteratively training the initial advertising image recognition model based on the training set to obtain a trained advertising image recognition model; and testing the trained advertising image recognition model based on the test set, and obtaining a final advertising image recognition model in the case that the testing is passed.

[0120] By adopting the embodiment of the present application, a financial advertising content review method is provided. The advertising image of the financial advertising content is collected in real time, and the advertising image is converted into textual description information based on a pre-trained advertising image recognition model, so as to realize accurate identification of various key elements in the advertising image. Then, the description information is taken as input data of a content review model to automatically check whether the advertising content meets the requirements. The purpose of automatically reviewing the financial advertising content is achieved, and the technical effect of improving the efficiency of the financial advertising content review is achieved. Furthermore, the technical problem of low review efficiency in the prior art that the financial advertising content is reviewed based on manual review is solved.

[0121] Those skilled in the art can understand that Figure 5 The structure shown is only schematic, and the electronic device can also be a smart phone, a tablet computer, a palm computer, a Mobile Internet Device (MID), a PAD, or the like terminal device. Figure 5 It does not limit the structure of the electronic device. For example, the electronic device can further include more or less components (such as a network interface, a display device, etc.) than those shown in the embodiment, or have a different configuration from that shown in the embodiment. Figure 5 Figure 5

[0122] Those skilled in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the terminal device related hardware through a program, and the program can be stored in a computer readable storage medium, which can include a flash disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, etc.

[0123] The application will be described below in combination with another alternative embodiment.

[0124] Embodiment Four

[0125] The embodiment of the application further provides a computer readable storage medium. Optionally, in the embodiment of the application, the computer readable storage medium can be used to save the program code executed by the financial advertising content auditing method provided in the embodiment one.

[0126] Optionally, in the embodiment of the application, the storage medium can be located in any one of the computer terminals in the computer terminal group in the computer network, or in any one of the mobile terminals in the mobile terminal group.

[0127] The embodiment of the application further provides a computer program product, when executed on a data processing device, is suitable for executing the steps of the financial advertising content auditing method: obtaining an advertising image of a target financial advertising content to be audited, and pre-processing the advertising image; inputting the pre-processed advertising image into an advertising image recognition model to output image description information of the advertising image, wherein the advertising image recognition model is a model pre-constructed for converting the advertising image into text information; inputting the image description information into a content auditing model to output an auditing result, wherein the content auditing model is a model pre-constructed for auditing whether the advertising content has an abnormality; performing risk assessment on the target financial advertising content based on the auditing result to obtain a risk assessment result, and generating an advertising content auditing report based on the risk assessment result.

[0128] ​​The above-mentioned sequence numbers of the embodiments of the present application are only for description, and do not represent the advantages or disadvantages of the embodiments.

[0129] In the above-mentioned embodiments of the present application, the description of each embodiment has its own focus, and the parts not described in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0130] In several embodiments provided in the present application, it should be understood that the disclosed technical contents can be implemented by other manners. Among them, the above-mentioned device embodiments are only schematic, for example, the division of the units can be a logical function division, and in actual implementation, there can be another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units or modules shown or discussed can be indirect coupling or communication connection through some interfaces, units or modules, which can be electrical or other forms.

[0131] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed to multiple units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0132] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be realized in the form of hardware or in the form of software functional unit.

[0133] The integrated unit, if realized in the form of software functional unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the whole or part of the technical solutions which essentially contribute to the prior art can be embodied in the form of software product, which is stored in a storage medium and includes a plurality of instructions for making a computer device (which can be a personal computer, a server or a network device, etc.) execute all or part of the steps of the method described in each embodiment of the present application. The foregoing storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic or optical disk and various program code storage media.

[0134] The above merely describes the preferred embodiments of the present application, and it should be pointed out that, for those skilled in the art, several improvements and refinements can be made without departing from the principles of the present application, and these improvements and refinements should also be considered as falling within the protection scope of the present application.

Claims

1. A method for reviewing financial advertisement content, characterized in that: include: Obtaining an advertisement image of the target financial advertisement content to be reviewed, and preprocessing the advertisement image; Inputting the pre-processed advertising image into an advertising image recognition model, and outputting image description information of the advertising image, wherein the advertising image recognition model is a pre-built model for converting advertising images into text information; Inputting the image description information into a content audit model and outputting an audit result, wherein the content audit model is a pre-built model for auditing whether there are any abnormalities in the advertising content; A risk assessment is performed on the target financial advertisement content based on the audit result to obtain a risk assessment result, and an advertisement content audit report is generated based on the risk assessment result.

2. The method according to claim 1, characterized in that The step of inputting the pre-processed advertising image into an advertising image recognition model and outputting image description information of the advertising image comprises: Inputting the pre-processed advertising image into the convolutional layer of the advertising image recognition model, and outputting various visual features of the advertising image; Inputting the visual features into the fully connected layer of the advertising image recognition model, and outputting type information and position information of each visual feature; Each of the visual features, the type information of the visual features, and the position information are input into a recurrent neural network layer of the advertising image recognition model, and image description information of the advertising image is output.

3. The method according to claim 1, characterized in that The steps of inputting the image description information into a content review model and outputting the review result include: Performing word segmentation processing and vector encoding on the image description information to obtain each word segmentation vector; Input each of the word segmentation vectors into the content audit model, and use the content audit model to identify and match each of the word segmentation vectors based on a preset rule library to obtain a rule audit result; Inputting each of the word segmentation vectors into the content audit model, performing semantic analysis on each of the word segmentation vectors through the content audit model, and obtaining a semantic analysis result; The audit result is obtained based on the rule audit result and the semantic analysis result, and the audit result is output as output information of the content audit model.

4. The method according to claim 1, wherein The steps of performing risk assessment on the target financial advertisement content based on the audit result and obtaining the risk assessment result include: Determining abnormal items hit by the target financial advertisement content based on the audit result; Calculating the risk value of the target financial advertisement content based on the abnormal value and weight value of each abnormal item hit; The risk value is compared with a preset risk threshold to obtain a comparison result, the risk level of the target financial advertisement content is determined based on the comparison result, and the risk assessment result is obtained based on the risk value and the risk level.

5. The method according to claim 1, wherein The steps of generating an advertising content review report based on the risk assessment results include: If the risk assessment result indicates that the target financial advertisement content is abnormal, searching a knowledge base based on the audit result and the risk assessment result to obtain an abnormality adjustment strategy; Determining location information of the target financial advertisement content; The advertising content review report is generated based on the advertising image, the image description information, the review result, the risk assessment result, the abnormality adjustment strategy, and the location information.

6. The method according to claim 1, characterized in that The step of pre-processing the advertisement image includes: performing denoising and image enhancement processing on the advertisement image; The advertisement image after image enhancement processing is subjected to standardization processing and normalization processing.

7. The method according to claim 1, characterized in that The steps of constructing the advertising image recognition model include: Collecting advertisement images of historical advertisement contents, preprocessing the historical advertisement contents, and generating description tags for the preprocessed advertisement images; Generating a training sample set based on the advertisement image and the description label, and dividing the training sample set into a training set and a test set; Selecting a deep learning model architecture as an initial advertising image recognition model, and performing parameter configuration and network layer configuration for the initial advertising image recognition model; Iteratively training the initial advertising image recognition model based on the training set to obtain a trained advertising image recognition model; The trained advertising image recognition model is tested based on the test set, and a final advertising image recognition model is obtained if the test passes.

8. A financial advertisement content review device, characterized in that: include: an acquisition unit, configured to acquire an advertisement image of a target financial advertisement content to be reviewed and pre-process the advertisement image; a recognition unit, configured to input the pre-processed advertising image into an advertising image recognition model and output image description information of the advertising image, wherein the advertising image recognition model is a pre-built model for converting advertising images into text information; an audit unit, configured to input the image description information into a content audit model and output an audit result, wherein the content audit model is a pre-built model for auditing whether there are any anomalies in the advertising content; A generating unit is configured to perform a risk assessment on the target financial advertisement content based on the audit result, obtain a risk assessment result, and generate an advertisement content audit report based on the risk assessment result.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the financial advertisement content review method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: It includes one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the financial advertising content review method described in any one of claims 1 to 7.

Images