Advertisement dynamic generation method and device, equipment and storage medium

Abstract

This application discloses a method, apparatus, device, and storage medium for dynamically generating advertisements. The method includes: acquiring user data of users visiting an advertisement display page; determining user interest tags based on the user data; generating advertisement copy and image materials corresponding to the user interest tags based on the user interest tags and a trained generative adversarial network model; acquiring the ad placement size of the advertisement display page and the device type of the visiting user; and dynamically rendering the advertisement copy and image materials based on the ad placement size and the device type to generate a corresponding customized advertisement. This solution, by combining user interest tags and a generative adversarial network model, achieves personalized generation of advertisement content and dynamically adjusts the advertisement display effect for different ad placement sizes and user device types, thereby increasing user attention and improving the display quality and user experience of advertisements on various terminals.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus, device and storage medium for dynamically generating advertisements. Background Technology

[0002] With the rapid development of the digital advertising industry, advertising creativity is increasingly widely used in the advertising technology field, becoming an important driver for enhancing brand influence and user conversion rates. With the widespread adoption of mobile internet, social media, and big data technologies, the advertising environment is becoming increasingly complex and volatile, and advertisers' demands for personalized, diversified, and precise creative content are constantly rising.

[0003] Current advertising creative primarily relies on manual design and static rule setting. With increasingly homogenized advertising creatives and diversified user needs, traditional solutions are revealing significant problems such as inefficiency, lack of personalization, and insufficient adaptability, failing to meet the demands of efficient and precise modern advertising. Manually designing advertising creatives not only consumes substantial time and human resources, making it difficult to meet the needs of massive advertising campaigns, but also, due to a lack of in-depth understanding of user interests and behaviors, struggles to generate customized creatives, resulting in monotonous and unoriginal advertising. Furthermore, traditional design solutions have limited adaptability to different ad sizes and device types, failing to quickly respond to diverse display environments, thus reducing advertising effectiveness and user experience. Summary of the Invention

[0004] This application provides a method, apparatus, device, and storage medium for dynamically generating advertisements. It generates matching advertisement copy and image materials by accessing users' interest tags, and dynamically renders them based on the advertisement placement size and device type to generate personalized advertisement content. This method effectively improves the accuracy of advertisement targeting and display effects, and is suitable for the intelligent creative generation needs of various advertising scenarios.

[0005] Firstly, this application provides a method for dynamically generating advertisements, including: Obtain user data of users who visit the ad display page, and determine user interest tags based on the user data; Based on the user interest tags and the trained generative adversarial network model, advertising copy and image materials corresponding to the user interest tags are generated; The system obtains the ad placement size of the ad display page and the device type of the accessing user, and dynamically renders the ad copy and image materials based on the ad placement size and the device type to generate a corresponding customized ad.

[0006] Secondly, this application provides an advertising dynamic generation device, comprising: The user tag module is used to obtain user data of users who visit the advertising display page and determine user interest tags based on the user data; The advertising creative module is used to generate advertising copy and image materials corresponding to the user interest tags based on the user interest tags and the trained generative adversarial network model; The ad generation module is used to obtain the ad space size of the ad display page and the device type of the accessing user, and dynamically render the ad copy and the image material based on the ad space size and the device type to generate a corresponding customized ad.

[0007] Thirdly, this application provides an advertising dynamic generation device, comprising: One or more processors; A memory that stores one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the dynamic advertising generation method as described in the first aspect.

[0008] Fourthly, this application provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform the dynamic advertising generation method as described in the first aspect.

[0009] This application constructs an interest tag recognition mechanism based on user behavior data analysis and a generative adversarial network-driven advertising content generation strategy, achieving intelligent customization and dynamic rendering of advertising copy and image materials. The method takes user data from ad display pages as input, accurately extracts user interest tags based on multi-dimensional user behavior features, and comprehensively captures user preference information. Subsequently, combined with a trained generative adversarial network model, it generates highly matched advertising copy and image materials based on user interest tags, achieving personalized content customization. Furthermore, it obtains information on ad placement size and user device type, and dynamically adapts and renders the generated advertising copy and image materials based on this information, ensuring optimal visual performance of the advertising content on different devices and ad placements. This collaborative mechanism of user interest-driven content generation and device adaptation significantly improves the relevance and display effect of advertisements, enhances user interaction experience, and is widely applicable to various scenarios such as digital advertising platforms, intelligent marketing systems, and automated advertising delivery tools. Attached Figure Description

[0010] Figure 1 This is a flowchart of an advertising dynamic generation method provided in an embodiment of this application; Figure 2 This is a flowchart of a method for determining user interest tags provided in an embodiment of this application; Figure 3This is a flowchart of a user interest tag filtering method provided in an embodiment of this application; Figure 4 This is a flowchart illustrating a method for generating advertising copy and image materials according to an embodiment of this application; Figure 5 This is a flowchart of a customized advertisement generation method provided in an embodiment of this application; Figure 6 This is a flowchart of a method for determining element layout density provided in an embodiment of this application; Figure 7 This is a flowchart illustrating a method for adjusting the layout position of advertising materials according to an embodiment of this application; Figure 8 This is a flowchart illustrating the steps of a method for dynamically generating advertisements provided in an embodiment of this application; Figure 9 This is a structural block diagram of an advertising dynamic generation device provided in an embodiment of this application; Figure 10 This is a schematic diagram of the structure of an advertising dynamic generation device provided in an embodiment of this application. Detailed Implementation

[0011] To make the objectives, technical solutions, and advantages of this application clearer, specific embodiments of this application will be described in further detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely for explaining this application and not for limiting it. It should also be noted that, for ease of description, only the parts relevant to this application are shown in the drawings, not all of them. Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe operations (or steps) as being processed sequentially, many of these operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. A process can be terminated when its operation is completed, but it may also have additional steps not included in the drawings. A process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0012] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0013] Currently, advertising creativity is playing an increasingly important role in the advertising technology field. However, traditional advertising creative design methods still heavily rely on manual design and static rule settings. While traditional design methods meet advertisers' needs to some extent, their limitations are becoming increasingly prominent in the face of severe homogenization of advertising creativity and increasingly diverse user needs. Although traditional solutions have the advantages of flexible design, ease of understanding and deployment, they suffer from significant shortcomings such as low design efficiency and difficulty in achieving personalization and dynamic adaptation, resulting in overall advertising effectiveness failing to meet targets. Existing advertising creative design faces multiple technical challenges. First, the manual design process is time-consuming and labor-intensive, making it difficult to meet the urgent need for efficient creative output in large-scale advertising campaigns. Second, the lack of personalized generation mechanisms for different user interests and preferences makes it difficult for advertising content to accurately reach the target audience. Third, market advertising creativity is generally homogenized and lacks innovation, making it difficult to stimulate user clicks and interaction. In addition, manually designed creatives cannot quickly adapt to diverse ad space sizes and device types, affecting the visual effect of ad display and user experience.

[0014] Based on the above challenges, there is an urgent need to build an intelligent advertising creative method with efficient and automated creative generation capabilities, supporting personalized customization and dynamic adaptation across multiple devices, in order to break through the bottlenecks of traditional design, improve the novelty and delivery effectiveness of advertising creatives, and meet the actual needs of advertising platforms, content creators, and intelligent marketing systems for high-quality and diversified advertising content.

[0015] To address the issues of insufficient personalization and low adaptation efficiency in existing advertising content generation, this embodiment provides a dynamic advertising generation method. This method integrates user behavior data analysis with a content customization mechanism driven by generative adversarial networks (GANs) to achieve accurate generation and dynamic rendering of advertising copy and image materials, meeting the adaptation requirements of diverse user needs and multi-terminal display environments. Taking user data from the ad display page as input, the method first accurately identifies user interest tags based on multi-dimensional user behavior characteristics, constructing a user interest profile. Then, combined with a trained GAN model, it automatically generates highly matched advertising copy and image materials based on user interest tags, achieving intelligent creation of personalized advertising content. Furthermore, it obtains information on the ad placement size and the device type of the accessing user on the ad display page, and dynamically adapts and renders the generated advertising copy and image materials based on these parameters, ensuring optimal visual performance of the advertising content on different devices and ad placements. This solution, through the collaborative processing of user profile construction, content generation, and multi-device dynamic rendering, significantly improves the relevance and display effect of advertising creatives, possesses good scalability and application value, and is suitable for various scenarios such as digital advertising platforms, intelligent marketing systems, and automated advertising delivery tools.

[0016] The dynamic ad generation method provided in this embodiment can be executed by a dynamic ad generation device, which can be implemented through software and / or hardware. This dynamic ad generation device can consist of two or more physical entities, or it can consist of a single physical entity. For example, the dynamic ad generation device can be an operation and maintenance server used to generate ads.

[0017] The ad dynamic generation device is equipped with at least one type of operating system, including but not limited to Android, Linux, and Windows. The ad dynamic generation device can install at least one application based on the operating system; this application can be a built-in application of the operating system or an application downloaded from a third-party device or server. In this embodiment, the ad dynamic generation device has at least one application capable of executing the ad dynamic generation method.

[0018] For ease of understanding, this embodiment uses the operation and maintenance server as the main body for executing the dynamic advertising generation method as an example for description.

[0019] Figure 1 A flowchart of a method for dynamically generating advertisements according to an embodiment of this application is provided. (Reference) Figure 1 The specific methods for dynamically generating advertisements include: S110. Obtain user data of users who visit the advertising display page, and determine user interest tags based on the user data.

[0020] In one embodiment, user data of users visiting the ad display page is first obtained. Here, user data refers to relevant information representing the visiting user, and the ad display page refers to the interface through which the user browses the ad content. After obtaining the user data, user interest tags are determined based on the user data. These user interest tags are tags representing user interests and preferences, used to assist in personalized ad recommendations.

[0021] In one embodiment, user interest tags can be determined by analyzing and mining the user's browsing history, click behavior, and historical purchase data to extract the user's interest features and generate corresponding tags.

[0022] In one embodiment, user interest tags can be determined by using a machine learning model to classify and cluster user data to obtain category tags for user interests.

[0023] In one embodiment, the method for determining user interest tags may be: based on a rule engine, matching preset interest keywords with user behavior data to infer the corresponding interest tags.

[0024] In one embodiment, the generated user interest tags can be dynamically updated and adjusted in real time as user behavior changes, thereby improving the relevance of ad recommendations and user experience.

[0025] Optionally, Figure 2 A flowchart of a user interest tag determination method provided in an embodiment of this application is given. (Reference) Figure 2 The method for determining user interest tags specifically includes: S1101. Extract browsing history, search history and purchase history from the user data.

[0026] For example, browsing history, search history, and purchase history can be extracted from user data. Browsing history refers to data representing the user's browsing behavior on advertising display pages and related pages, search history refers to data representing the user's input search keywords and related search behavior, and purchase history refers to information representing the purchase transactions completed by the user.

[0027] In one embodiment, browsing history can be extracted by parsing user access logs to obtain information such as the URL of the accessed page, access time, and duration of stay.

[0028] In one embodiment, the search history can be extracted by filtering the keywords entered by the user and the search time from the user's search history database.

[0029] In one embodiment, the purchase record can be extracted by obtaining relevant information such as the user's order details and transaction time from the e-commerce transaction system.

[0030] S1102. Perform keyword segmentation and keyword vectorization on the browsing history, the search history and the purchase history respectively to obtain the corresponding browsing feature vector, search feature vector and purchase feature vector.

[0031] For example, browsing history, search history, and purchase history are processed by keyword segmentation and keyword vectorization to obtain corresponding browsing feature vectors, search feature vectors, and purchase feature vectors. Keyword segmentation refers to using natural language processing technology to divide text content into semantically meaningful keywords; keyword vectorization refers to converting the segmented keywords into numerical vectors for easier subsequent calculation and analysis; the browsing feature vector, search feature vector, and purchase feature vector are numerical vectors representing the user's interest features in the corresponding records.

[0032] In one embodiment, keyword segmentation can be performed by using a statistical or rule-based segmentation algorithm to break down text data into word units.

[0033] In one embodiment, keyword vectorization can be achieved by using a word embedding model to convert keywords into fixed-dimensional vector representations.

[0034] In one embodiment, the feature vector can be generated by weighting the keyword vectors or by extracting the overall feature vector through a deep learning model.

[0035] S1103. Determine user interest tags based on the browsing feature vector, the search feature vector, and the purchase feature vector.

[0036] For example, user interest tags are determined based on browsing feature vectors, search feature vectors, and purchase feature vectors. Here, user interest tags refer to category tags used to characterize user interests and preferences, assisting in personalized recommendations and precise advertising.

[0037] In one embodiment, the user interest label can be determined by inputting various feature vectors into a pre-trained classification model and determining the user interest label through the class probability distribution output by the model.

[0038] In one embodiment, the method for determining user interest tags may be: using a clustering algorithm to aggregate and analyze multidimensional feature vectors, identifying potential user interest groups and generating corresponding interest tags.

[0039] In one embodiment, the method for determining user interest tags may be: matching and filtering feature vectors based on a rule engine, and inferring user interest categories by combining pre-defined business rules.

[0040] Optionally, Figure 3 A flowchart of a user interest tag filtering method provided in an embodiment of this application is given. (Reference) Figure 3 The user interest tag filtering method specifically includes: S1104. Based on the user data, determine a number of initial interest tags, and obtain the exposure data and feedback data of the accessing user for each of the initial interest tags within a preset statistical period.

[0041] For example, several initial interest tags are determined based on user data, where initial interest tags refer to interest categories initially identified based on user behavior data. Subsequently, the exposure data and feedback data corresponding to each initial interest tag for the visiting user are obtained within a preset statistical period. Exposure data refers to the number of times the user sees advertisements related to the interest tag within the statistical period; feedback data refers to the user's interaction with the relevant advertisements, such as clicking, saving, or purchasing.

[0042] In one embodiment, the initial interest tags can be determined by filtering out several potential interest tags based on the user's historical behavior analysis or pre-trained model prediction.

[0043] In one embodiment, exposure data and feedback data can be obtained by using an advertising delivery system to collect statistics on user ad impressions and interactions over a specified period.

[0044] In one embodiment, the preset statistical period can be one day, one week, or one month, with the specific duration set according to business needs.

[0045] S1105. Calculate the user fatigue index corresponding to the initial interest tag based on the exposure data and the feedback data.

[0046] For example, the user fatigue index corresponding to the initial interest tags is calculated based on exposure data and feedback data. The user fatigue index is a numerical value used to measure the degree of visual or psychological fatigue experienced by users towards advertising content related to specific interest tags, reflecting a decreasing trend in advertising effectiveness.

[0047] In one embodiment, the user fatigue index can be calculated as follows: based on the ratio of exposure count to user feedback rate, the fatigue index increases accordingly as the exposure count increases and the feedback rate decreases.

[0048] In one embodiment, the user fatigue index can be calculated by using a weighted function that combines exposure frequency and negative feedback data to comprehensively calculate the fatigue score.

[0049] In one embodiment, the user fatigue index can be dynamically adjusted using a time decay model to reflect changes in user interest and fatigue levels in real time.

[0050] S1106. Use the user fatigue index to determine user interest tags from the initial interest tags.

[0051] For example, user fatigue metrics are used to determine final user interest tags from initial interest tags. These user interest tags, after fatigue adjustment, more accurately reflect the user's current interests and preferences, thus improving the effectiveness of personalized recommendations.

[0052] In one embodiment, the method for determining user interest tags may be: filtering initial interest tags based on user fatigue index, removing tags whose fatigue index exceeds a preset threshold, and retaining tags with high interest activity.

[0053] In one embodiment, the method for determining user interest tags may be to combine fatigue index and interest intensity index to comprehensively evaluate the effectiveness of the tags and dynamically adjust the final set of interest tags.

[0054] In one embodiment, user interest tags can be determined by iteratively updating user interest tags through multiple rounds, and gradually optimizing tag accuracy and recommendation performance using fatigue indicators.

[0055] S120. Based on the user interest tags and the trained generative adversarial network model, generate advertising copy and image materials corresponding to the user interest tags.

[0056] In one embodiment, based on the acquired user interest tags and the trained generative adversarial network (GAN) model, advertising copy and image assets corresponding to the user interest tags are generated. Here, user interest tags refer to tags that represent user interests and preferences, used to guide the personalized generation of advertising content; the GAN model refers to a deep learning model trained adversarially, capable of generating matching advertising copy and image assets based on the input interest tags; advertising copy refers to textual content that expresses promotional information; and image assets refer to visual content used for advertising display.

[0057] In one embodiment, the method of generating advertising copy and image materials based on user interest tags and a generative adversarial network model can be as follows: user interest tags are used as input to the model, the generative network generates preliminary content, and then the discriminative network performs quality evaluation and optimization, iteratively improving the generation effect.

[0058] In one embodiment, the way to generate advertising copy and image materials based on user interest tags and generative adversarial network models can be: combining a generative adversarial network trained with multimodal data, and outputting advertising content that is highly matched with user interests by fusing interest tags and historical advertising data.

[0059] In one embodiment, the method of generating advertising copy and image materials based on user interest tags and generative adversarial network models can be: using conditional generative adversarial networks to generate customized advertising content for different interest tags by controlling the generation conditions.

[0060] Optionally, Figure 4 A flowchart of a method for generating advertising copy and image materials according to an embodiment of this application is provided. (Reference) Figure 4 The specific methods for generating the advertising copy and image materials include: S1201. Match target products according to the user interest tags and obtain product feature information of the target products. The product feature information includes product category, brand tags, function description and price range.

[0061] For example, target products are matched based on user interest tags, and product feature information of the target products is obtained. Here, user interest tags refer to a set of tags reflecting the user's current interests and preferences; target products refer to products that have a high degree of matching with user interest tags and are suitable for personalized recommendations; product feature information includes product category, brand tags, functional description, and price range, used to describe the core attributes of the target product.

[0062] In one embodiment, the method for matching target products may be: calculating the similarity between user interest tags and product tags using a tag matching algorithm, and selecting products with a similarity exceeding a threshold as target products.

[0063] In one embodiment, the way to obtain product feature information may be by reading information such as product category, brand label, function description and price range from a product database or product management system.

[0064] In one embodiment, the product category is used to represent the major or sub-category to which the product belongs, the brand label is used to represent the brand affiliation of the product, the functional description is used to describe the main functions and applicable scenarios of the product, and the price range is used to indicate the pricing range of the product.

[0065] S1202. The user interest tags and the product feature information are fused through feature encoding and vector concatenation to obtain a fused feature vector.

[0066] For example, user interest tags and product feature information are fused through feature encoding and vector concatenation to obtain a fused feature vector. Feature encoding refers to converting user interest tags and product feature information into numerical feature representations for unified processing and calculation; vector concatenation refers to merging the encoded user interest tag vector and product feature vector in dimensional order into a new high-dimensional vector; the fused feature vector is a numerical vector containing comprehensive information about user interests and product characteristics, used to drive the calculations of the subsequent advertising content generation model.

[0067] In one embodiment, the feature encoding method can be: using one-hot encoding, word embedding, or a pre-trained model to convert user interest tags and product feature information into vector representations respectively.

[0068] In one embodiment, the vector concatenation method can be to directly concatenate the encoded user interest tag vector with the product feature vector along the feature dimension to form a unified input vector.

[0069] In one embodiment, the fused feature vector can also be generated by weighted concatenation, that is, by assigning weights to feature vectors from different sources to highlight the impact of key information on ad generation.

[0070] S1203. Input the fused feature vector into the trained generative adversarial network model to generate advertising copy and image materials.

[0071] For example, the fused feature vector is input into a trained generative adversarial network (GAN) model to generate advertising copy and image assets. Here, the fused feature vector refers to a numerical vector containing comprehensive information about user interests and product characteristics; the GAN model is a deep learning model trained adversarially between a generator and a discriminator, used to generate highly relevant advertising content based on the input features; the advertising copy refers to textual content that conveys promotional information; and the image assets refer to visual content used for advertising display.

[0072] In one embodiment, the method for generating advertising copy and image materials can be as follows: the fused feature vector is fed as a conditional input into a conditional generative adversarial network, the generator generates preliminary copy and image materials, and then the discriminator evaluates and provides feedback for optimization until a preset quality standard is reached.

[0073] In one embodiment, the way to generate advertising copy and image assets can be by combining a multimodal generation architecture to process text and image information synchronously during the generation process, so as to ensure the semantic and visual consistency of advertising copy and image assets.

[0074] In one embodiment, the way to generate advertising copy and image materials can also be by introducing an attention mechanism during the generation process, giving higher weight to key information in the fused feature vector, thereby improving the relevance and attractiveness of the generated content.

[0075] S130. Obtain the ad space size of the ad display page and the device type of the accessing user, and dynamically render the ad copy and the image material based on the ad space size and the device type to generate a corresponding customized ad.

[0076] In one embodiment, the ad placement size and the device type of the accessing user on the ad display page are obtained, where ad placement size refers to a parameter characterizing the size of the ad display area, and device type refers to an attribute characterizing the category of the user's accessing device. Based on the ad placement size and device type, the ad copy and image materials are dynamically rendered to generate a corresponding customized ad.

[0077] In one embodiment, the method of dynamically rendering ad copy and image materials based on ad space size and device type can be: adjusting the size and layout of image materials according to the aspect ratio of the ad space and the device screen resolution to ensure that the ad content is displayed completely and aesthetically.

[0078] In one embodiment, dynamically rendering ad copy and image materials based on ad placement size and device type can be achieved by differentiating between mobile and desktop devices based on device type, and selecting appropriate font size, text layout, and interactive effects to improve user experience.

[0079] In one embodiment, the way to dynamically render ad copy and image assets based on ad placement size and device type can be by adopting a responsive design strategy and using real-time calculation and adjustment to achieve adaptive display of ad content on different devices and ad placements.

[0080] Optionally, Figure 5 A flowchart of a customized advertisement generation method provided in an embodiment of this application is given. (Reference) Figure 5 The specific methods for generating customized advertisements include: S1301. Determine the element layout area and element layout density of the advertising space based on the advertising space size and the device type.

[0081] For example, the element layout area and element layout density of an ad space are determined based on the ad space size and device type. Here, ad space size refers to parameters characterizing the width and height of the ad display area; device type refers to the type of access device used by the user, such as mobile, tablet, or desktop; element layout area refers to the space within the ad space that can be used to place ad elements; and element layout density refers to the number of ad elements or the compactness of their arrangement within a unit layout area.

[0082] In one embodiment, the method for determining the element layout area can be: dividing the ad space into several functional areas, such as the main visual area, the title area, and the interactive area, based on the width-to-height ratio and the visible range of the ad space.

[0083] In one embodiment, the element layout density can be determined by calculating the number of elements and spacing that can be accommodated per unit area based on the device screen resolution, the user's acceptable visual load, and the importance of the advertising elements.

[0084] In one embodiment, the method for determining the element layout area and density can also be: dynamically adjusting the area division and density parameters through a responsive design algorithm to ensure good advertising display effects on devices of different sizes and types.

[0085] Optionally, Figure 6 A flowchart of a method for determining element layout density according to an embodiment of this application is provided. (Reference) Figure 6 The method for determining the element layout density specifically includes: S13011. Generate a corresponding element layout object based on the size of the ad space, divide the element layout object into several element layout areas, and determine the area weight factor based on the relative position of the element layout area in the element layout object.

[0086] For example, an element layout object is generated based on the ad space size. This element layout object is then divided into several element layout regions, and a region weight factor is determined based on the relative position of each element layout region within the element layout object. Here, the element layout object refers to the layout model constructed based on the ad space size to hold the ad elements; the element layout region refers to the functional area within the element layout object used to place the ad elements; and the region weight factor is a numerical parameter used to measure the importance of each element layout region within the overall ad space, typically related to the region's visibility, user attention, and interaction potential.

[0087] In one embodiment, the method for generating element layout objects can be: constructing a two-dimensional or multi-dimensional layout model based on the aspect ratio and resolution of the ad space, as the basic framework for arranging ad elements.

[0088] In one embodiment, the way to divide the element layout area can be: based on visual design specifications, the layout object is divided into different functional areas such as the title area, the main visual area, the auxiliary information area, and the interactive button area.

[0089] In one embodiment, the method for determining the regional weighting factor may be: based on user gaze heatmap data, click behavior analysis, or industry design experience, assign higher weights to areas located in visual focal points or high-interaction positions, and assign lower weights to peripheral or low-attention areas.

[0090] S13012. Obtain the screen pixel density parameters of the accessed device according to the device type, and calculate the baseline layout density according to the screen pixel density parameters.

[0091] For example, the screen pixel density parameter of the accessing device is obtained according to the device type, and the baseline layout density is calculated based on the screen pixel density parameter. Here, device type refers to the type of hardware used by the user to access the advertising content, such as smartphone, tablet, desktop, or wearable device; screen pixel density parameter refers to the number of pixels contained in a unit physical length, usually expressed in PPI or DPI; baseline layout density refers to the element arrangement density standard that ensures the clarity and readability of advertising elements under specific pixel density conditions.

[0092] In one embodiment, the screen pixel density parameter can be obtained by reading pixel density data through the device's system API, browser UA information, or accessing the terminal's hardware parameter interface.

[0093] In one embodiment, the baseline layout density can be calculated by proportionally converting the number of elements that each layout area can accommodate and the minimum spacing based on the screen pixel density and the size of the ad space, thereby determining a suitable baseline density value.

[0094] In one embodiment, the baseline layout density can also be calculated by combining a visual comfort model to match pixel density with the user's visual discrimination ability, thereby optimizing the compactness and visualization effect of the ad elements.

[0095] S13013. Determine the element layout density corresponding to each of the element layout regions based on the region weight factor and the baseline layout density.

[0096] For example, the element layout density of each element layout area is determined based on the regional weight factor and the baseline layout density. The regional weight factor is a numerical parameter used to measure the importance of an element layout area within an ad placement; the baseline layout density is the standard for element arrangement density that ensures the clarity and readability of ad elements under specific pixel density conditions; and the element layout density refers to the number of ad elements or the recommended compactness of their arrangement within a unit layout area.

[0097] In one embodiment, the layout density of each element's layout area can be determined by multiplying the baseline layout density by the corresponding area weight factor to obtain the final layout density value of that area.

[0098] In one embodiment, the layout density of each element's layout area can be determined by weighting the baseline layout density according to the importance of the area, so that the layout density of high-weight areas is higher than that of low-weight areas, thereby highlighting the display effect of key information.

[0099] In one embodiment, the method for determining the layout density of each element's layout area can also be: introducing a density balance coefficient to visually optimize high-density areas, prevent information overload, and ensure the information integrity of low-density areas.

[0100] S1302. Adjust the layout position of the advertising copy and the image material according to the element layout area and its corresponding element layout density.

[0101] For example, the layout positions of ad copy and image assets are adjusted based on the element layout area and its corresponding element layout density. Here, the element layout area refers to the space within the ad space used to place ad elements; the element layout density refers to the number of ad elements or the compactness of their arrangement within a unit layout area; and the layout position refers to the specific placement and relative relationship of the ad copy and image assets within the ad space.

[0102] In one embodiment, adjusting the layout can be done by prioritizing the placement of advertising copy in high-visibility areas and distributing image materials in locations that enhance visual impact, based on the functional attributes of different areas.

[0103] In one embodiment, the layout position can be adjusted by dynamically adjusting the ratio and spacing of the advertising copy and image materials based on the element layout density to avoid visual crowding or information sparseness.

[0104] In one embodiment, adjusting the layout can also be achieved by combining user device interaction habits and eye trajectory analysis results to optimize the arrangement order and display priority of advertising copy and image materials, so as to improve the click-through rate and conversion rate of the advertisement.

[0105] Optionally, Figure 7 A flowchart of a method for adjusting the layout position of advertising materials according to an embodiment of this application is provided. (Reference) Figure 7 The specific methods for adjusting the layout of this advertising material include: S13021. Generate initial image material based on the image material and the element layout object, and calculate the corresponding advertisement layout density of each element layout area in the initial image material.

[0106] For example, initial image assets are generated based on image assets and element layout objects, and the ad layout density corresponding to each element layout area in the initial image assets is calculated. Here, image assets refer to the original visual content used for ad display; element layout objects refer to the layout model constructed according to the ad space size to carry ad elements; initial image assets refer to the preliminary ad image generated after arranging or embedding the image assets according to the element layout objects; and ad layout density refers to the number of ad elements or the compactness of information arrangement within a specific layout area.

[0107] In one embodiment, the initial image material can be generated by scaling, cropping, and adjusting the position of the image material according to the area division and proportion requirements of the element layout object to generate a preliminary layout image that is adapted to the ad space.

[0108] In one embodiment, the method for calculating the ad layout density can be: counting the number of ad elements and their occupied area in each element layout area, and then calculating the ratio between them and the area of ​​that area.

[0109] In one embodiment, the method for calculating the density of the advertisement layout can also be: combining computer vision technology to automatically detect the distribution of elements in the layout area, and comprehensively considering the number of elements, area ratio and relative positional relationship to obtain a more accurate density value.

[0110] S13022. Calculate the layout density deviation of each element layout area using the advertisement layout density and the element layout density.

[0111] For example, the layout density deviation of each element layout area is calculated using the ad layout density and element layout density. Here, ad layout density refers to the actual arrangement density of ad elements within each element layout area in the initial image material; element layout density refers to the ideal arrangement density determined based on the area weight factor and the baseline layout density; and layout density deviation refers to the difference between the actual layout density and the ideal layout density, used to measure the rationality of the layout and its potential for optimization.

[0112] In one embodiment, the layout density deviation can be calculated by subtracting the corresponding element layout density from the actual advertising layout density for each element layout area to obtain positive and negative deviation values.

[0113] In one embodiment, the layout density deviation can be calculated by using absolute deviation or relative deviation as a metric for layout density deviation.

[0114] In one embodiment, the method for calculating the layout density deviation can also be: combining the user's visual perception model, prioritizing the adjustment and optimization of areas with larger deviations in order to improve the overall advertising layout effect.

[0115] S13023. Determine the text layout position based on the layout density deviation, and assign the advertising text to the text layout position.

[0116] For example, the copy placement position is determined based on the layout density deviation, and the advertising copy is assigned to that placement position. Here, layout density deviation refers to the difference between the actual advertising layout density and the ideal element layout density, used to guide layout optimization; copy placement position refers to the specific area or coordinate location of the advertising copy within the advertising space; and advertising copy refers to the text content used to convey promotional information.

[0117] In one embodiment, the method for determining the placement of the copy can be: prioritizing areas with smaller or negative density deviations for placing the advertising copy, in order to avoid visual crowding and ensure that the copy is clear and readable.

[0118] In one embodiment, the method for determining the layout of the text can be: dynamically adjusting the size and position of the text area based on the length and font size of the advertising text to ensure that the content is displayed completely and coordinated with the image materials.

[0119] In one embodiment, the way to assign advertising copy to the copy layout position is to render the copy content to a certain position through a layout engine and adapt the text style to improve the overall visual effect and user experience.

[0120] S1303. Render and synthesize the adjusted advertising copy and image materials to generate the corresponding customized advertisement.

[0121] For example, the adjusted advertising copy and image materials are rendered and composited to generate corresponding customized advertisements. Rendering and compositing refers to visually processing and integrating the advertising copy and image materials according to a layout scheme to form advertising content that can be directly deployed; customized advertisements refer to advertisements generated based on user interest tags, product feature information, and device characteristics, which are visually and content-matched to the target user.

[0122] In one embodiment, the rendering composition method may be: using a graphics rendering engine to draw the advertising copy and image materials into the advertising space according to a determined layout position, while optimizing the font, color, shadow and transparency.

[0123] In one embodiment, the rendering compositing method can be: based on multi-layer compositing technology, text layers and image layers are superimposed and blended to ensure that each element remains clear and harmonious at different resolutions.

[0124] In one embodiment, the rendering and compositing method can also be: using a real-time rendering mechanism to dynamically generate the ad file format based on the user's device performance and network conditions, ensuring ad loading speed and smooth interaction.

[0125] Optionally, Figure 8 A step-by-step diagram illustrating an advertising dynamic generation method provided in an embodiment of this application is given. (Reference) Figure 8 The specific methods for dynamically generating advertisements include: S201, User interest tag reception.

[0126] For example, user interest tags can be extracted from user behavior data to construct user profiles. These user interest tags refer to category tags that represent user interests and preferences, used to guide personalized recommendations and advertising.

[0127] In one embodiment, user interest tags can be received by obtaining interest tags through user behavior analysis results uploaded by the client. In one embodiment, user interest tags can be received by obtaining them from interest tag data pushed by the server-side recommendation system. In one embodiment, the user interest tags can also be received by combining a multi-source data fusion mechanism to comprehensively acquire and update user interest tags, thereby improving the accuracy and real-time performance of the tags.

[0128] S202, GAN model generation.

[0129] For example, a generative adversarial network (GAN) model is used to generate high-quality advertising content. Here, a GAN model refers to a deep learning model trained through adversarial training between a generator and a discriminator, used to generate high-quality advertising content based on input features.

[0130] In one embodiment, the GAN model can generate initial advertising copy and image materials by taking user interest tags and related feature vectors as input and passing them through a generator network.

[0131] In one embodiment, the GAN model can generate content by using a discriminator to assess the quality of the generated content and provide feedback, thereby continuously optimizing the generator's output and improving the relevance and attractiveness of the advertising content.

[0132] In one embodiment, the GAN model can also be generated by combining a conditional generative adversarial network structure with conditional information to guide the diversification and customization of advertising content.

[0133] S203, Advertising Creative Generation.

[0134] For example, a trained generative adversarial network model can be used to generate diverse advertising copy and image materials by combining user interest tags.

[0135] Advertising creative refers to the generation of attractive and personalized advertising copy and visual content based on user interest tags and product characteristics, using algorithms.

[0136] In one embodiment, advertising creatives can be generated by combining preliminary text and image materials generated by generative adversarial networks with multimodal fusion and optimization to enhance the expressive effect of the content.

[0137] In one embodiment, advertising creative can be generated by applying natural language processing technology to polish and style the advertising copy, making the text more in line with the reading habits of the target users.

[0138] In one embodiment, ad creative generation can also be achieved by guiding the editing and layout of image materials through visual design rules and user behavior data, thereby enhancing the visual impact and click-through rate of the ad.

[0139] S204, PHP dynamic rendering.

[0140] For example, the generated advertising copy and image assets are embedded in a PHP template, and then dynamically rendered using a template engine. Dynamic PHP rendering refers to using the PHP scripting language to dynamically generate advertising page content based on input data, achieving personalized display.

[0141] In one embodiment, PHP dynamic rendering can be achieved by dynamically calling advertising copy and image materials based on user interest tags and device type to generate an advertising page layout adapted to different terminals.

[0142] In one embodiment, PHP dynamic rendering can be achieved by using a template engine to combine ad content and layout parameters to generate HTML code that conforms to design specifications in real time, thereby enabling flexible combination and display of ad elements.

[0143] In one embodiment, PHP dynamic rendering can also be achieved by combining caching mechanisms to optimize rendering efficiency, reduce server response time, and improve the user browsing experience.

[0144] S205, Adaptation for Advertising.

[0145] For example, responsive ad adaptation can be implemented based on ad placement size and device type to ensure optimal ad display. Ad adaptation refers to adjusting ad content and format according to different user devices, screen sizes, and network environments to guarantee the best display effect and user experience.

[0146] In one embodiment, ad adaptation can be achieved by automatically adjusting the ad size, resolution, and interaction method based on the type of accessing device to achieve responsive ad display.

[0147] In one embodiment, ad adaptation can be achieved by dynamically selecting image quality and content complexity based on network bandwidth and loading speed, ensuring fast ad loading and good visual effects.

[0148] In one embodiment, ad adaptation can also be achieved by combining user behavior data to optimize the presentation order and priority of ad content, thereby improving click-through rate and conversion rate.

[0149] S206, User Feedback Record.

[0150] For example, monitoring and recording user behavior feedback to advertisements, including key metrics such as click-through rate, conversion rate, and dwell time, forms a feedback dataset for user feedback recording. User feedback refers to user interaction and reactions to advertising content, including actions such as clicking, browsing duration, saving, sharing, and closing; user feedback recording refers to the process of collecting, storing, and analyzing the above behaviors.

[0151] In one embodiment, user feedback can be recorded by capturing user interaction data with advertisements in real time through a front-end event listening and log collection system.

[0152] In one embodiment, user feedback can be recorded by uploading the collected feedback data to a server and storing it in a database or big data platform for subsequent analysis and model training.

[0153] In one embodiment, user feedback can also be recorded by combining privacy protection mechanisms to ensure the security and compliance of user data and enhance user trust.

[0154] S207, GAN model optimization.

[0155] For example, by utilizing collected user feedback data, the generative adversarial network (GAN) model can be optimized in reverse through methods such as reinforcement learning or incremental learning. This involves adjusting model parameters and training data to improve the relevance and conversion rate of ad creatives. Specifically, GAN model optimization refers to improving the quality and relevance of generated ad content by adjusting model parameters, training strategies, and data input.

[0156] In one embodiment, the GAN model can be optimized by dynamically adjusting the weight parameters of the generator and discriminator based on user feedback data using reinforcement learning or online learning mechanisms.

[0157] In one embodiment, GAN model optimization can be achieved by increasing the diversity of training data and the number of samples to improve the model's generalization ability and the diversity of generated content.

[0158] In one embodiment, GAN model optimization can also be achieved by using an improved loss function or regularization method to enhance the stability of model training, reduce pattern collapse, and improve the continuity and consistency of generation results.

[0159] Based on the above embodiments, Figure 9 This is a structural block diagram of an advertising dynamic generation device provided in an embodiment of this application. (Reference) Figure 9 The advertising dynamic generation device provided in this embodiment specifically includes: a user tag module 21, an advertising material module 22, and an advertising generation module 23.

[0160] The user tagging module 21 is configured to acquire user data of users visiting the ad display page and determine user interest tags based on the user data; the ad creative module 22 is configured to generate ad copy and image creatives corresponding to the user interest tags based on the user interest tags and the trained generative adversarial network model; the ad generation module 23 is configured to acquire the ad slot size of the ad display page and the device type of the visiting user, dynamically render the ad copy and image creatives based on the ad slot size and the device type, and generate corresponding customized ads.

[0161] Based on the above embodiments, the user tag module 21 includes: a record extraction unit configured to extract browsing records, search records, and purchase records from the user data; a feature vector unit configured to perform keyword segmentation and keyword vectorization on the browsing records, search records, and purchase records respectively to obtain corresponding browsing feature vectors, search feature vectors, and purchase feature vectors; and a tag determination unit configured to determine user interest tags based on the browsing feature vectors, search feature vectors, and purchase feature vectors.

[0162] Based on the above embodiments, the user tag module 21 further includes: an exposure feedback unit, configured to determine a plurality of initial interest tags based on the user data, and obtain the exposure data and feedback data corresponding to each of the initial interest tags for the accessing user within a preset statistical period; a fatigue index unit, configured to calculate the user fatigue index corresponding to the initial interest tag based on the exposure data and the feedback data; and a tag filtering unit, configured to determine user interest tags from the initial interest tags using the user fatigue index.

[0163] Based on the above embodiments, the advertising material module 22 includes: a product feature unit, configured to match target products according to the user interest tags and obtain product feature information of the target products, the product feature information including product category, brand tags, functional description and price range; a feature fusion unit, configured to fuse the user interest tags and the product feature information through feature encoding and vector concatenation to obtain a fused feature vector; and a material generation unit, configured to input the fused feature vector into a trained generative adversarial network model to generate advertising copy and image materials.

[0164] Based on the above embodiments, the advertisement generation module 23 includes: an element layout unit configured to determine the element layout area and element layout density of the advertisement space based on the advertisement space size and the device type; a layout position unit configured to adjust the layout position of the advertisement text and the image material according to the element layout area and the corresponding element layout density; and a rendering and compositing unit configured to render and composite the adjusted advertisement text and image material to generate a corresponding customized advertisement.

[0165] Based on the above embodiments, the element layout unit includes: a region weight subunit, configured to generate a corresponding element layout object based on the ad space size, divide the element layout object into several element layout regions, and determine a region weight factor based on the relative position of the element layout region in the element layout object; a baseline density subunit, configured to obtain the screen pixel density parameter of the accessing device according to the device type, and calculate the baseline layout density according to the screen pixel density parameter; and an element density subunit, configured to determine the element layout density corresponding to each of the element layout regions based on the region weight factor and the baseline layout density.

[0166] Based on the above embodiments, the layout position unit includes: an ad density subunit, configured to generate initial image material based on the image material and the element layout object, and calculate the ad layout density corresponding to each element layout area in the initial image material; a density deviation subunit, configured to calculate the layout density deviation corresponding to each element layout area using the ad layout density and the element layout density; and a layout position subunit, configured to determine the text layout position based on the layout density deviation, and allocate the ad text to the text layout position.

[0167] The aforementioned advertising dynamic generation device provided in this application, through the construction of a structured processing system consisting of a user tagging module, an advertising material module, and an advertising generation module, achieves fully automated processing from user interest identification and advertising content generation to personalized rendering. This device can dynamically extract interest tags based on the behavioral data of visiting users, combine them with a generative adversarial network model, and generate advertising copy and image materials that highly match user interests. Furthermore, it completes adaptive rendering and display of advertising content according to the ad placement size and the type of accessing device, significantly improving the personalization and adaptability of the advertisement. Specifically, the user tagging module possesses multi-dimensional user data collection and interest tag identification capabilities, enabling it to acquire real-time user data from the ad display page. Based on big data analysis and tag mining technology, it accurately determines user interests and preferences, providing precise user profile support for advertising content generation. The advertising material module, based on user interest tags and a trained generative adversarial network model, automatically generates advertising copy and image materials highly relevant to user interests, possessing multi-modal creative generation and content style matching capabilities, enhancing the diversity and attractiveness of advertising materials. As the core unit for customized ad content, the ad generation module obtains the ad placement size and the type of the user's device on the ad display page. Based on the size and device characteristics, it dynamically adjusts the layout and presentation of ad copy and image materials to achieve adaptive rendering and high-quality output of ad content, ensuring optimal ad presentation across different terminal environments. Through the user profile-driven, creative generation, and dynamic rendering chain collaboratively constructed by the above modules, this device breaks through the limitations of traditional static ad templates, achieving deep linkage between ad content and user interests and cross-terminal adaptability. It is suitable for intelligent ad platforms, personalized marketing systems, and multi-terminal ad distribution scenarios, significantly improving ad conversion rates and user experience, becoming a core technology unit in the intelligent ad generation system.

[0168] The advertising dynamic generation device provided in this application embodiment can be used to execute the advertising dynamic generation method provided in the above embodiment, and has corresponding functions and beneficial effects.

[0169] Figure 10 This is a schematic diagram of the structure of an advertising dynamic generation device provided in an embodiment of this application, with reference to... Figure 10 The dynamic advertising generation device includes a processor 31, a memory 32, a communication device 33, an input device 34, and an output device 35. The number of processors 31 and the number of memories 32 in the dynamic advertising generation device can be one or more. The processor 31, memory 32, communication device 33, input device 34, and output device 35 of the dynamic advertising generation device can be connected via a bus or other means.

[0170] The memory 32, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as program instructions / modules corresponding to the dynamic advertising generation method in any embodiment of this application (e.g., user tag module 21, advertising material module 22, and advertising generation module 23 in the dynamic advertising generation device). The memory 32 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function; the data storage area may store data created based on the use of the device, etc. Furthermore, the memory 32 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to the device via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0171] The communication device 33 is used for data transmission.

[0172] The processor 31 executes various functional applications and data processing of the device by running software programs, instructions and modules stored in the memory 32, thereby realizing the above-mentioned dynamic advertising generation method.

[0173] Input device 34 can be used to receive input digital or character information, and to generate key signal inputs related to user settings and function control of the device. Output device 35 may include display devices such as a display screen.

[0174] The aforementioned advertising dynamic generation device can be used to execute the advertising dynamic generation method provided in the above embodiments, and has corresponding functions and beneficial effects.

[0175] This application embodiment also provides a storage medium containing computer-executable instructions. When executed by a computer processor, the computer-executable instructions are used to perform a dynamic advertising generation method. The dynamic advertising generation method includes: acquiring user data of users visiting an advertising display page; determining user interest tags based on the user data; generating advertising copy and image materials corresponding to the user interest tags based on the user interest tags and a trained generative adversarial network model; acquiring the ad slot size of the advertising display page and the device type of the visiting user; and dynamically rendering the advertising copy and image materials based on the ad slot size and the device type to generate a corresponding customized advertisement.

[0176] Storage medium – any type of memory device or storage device. The term “storage medium” is intended to include: mounting media, such as CD-ROM, floppy disk, or magnetic tape devices; computer system memory or random access memory, such as DRAM, DDR RAM, SRAM, EDO RAM, Rambus RAM, etc.; non-volatile memory, such as flash memory, magnetic media (e.g., hard disk or optical storage); registers or other similar types of memory elements, etc. Storage medium may also include other types of memory or combinations thereof. Furthermore, storage medium may reside in a first computer system in which a program is executed, or it may reside in a different second computer system connected to the first computer system via a network (such as the Internet). The second computer system can provide program instructions to the first computer for execution. The term “storage medium” can include two or more storage media residing in different locations (e.g., in different computer systems connected via a network). Storage medium may store program instructions (e.g., specifically implemented as a computer program) executable by one or more processors.

[0177] Of course, the computer-executable instructions provided in the embodiments of this application are not limited to the above-mentioned dynamic advertising generation method, but can also perform related operations in the dynamic advertising generation method provided in any embodiment of this application.

[0178] The advertising dynamic generation device, storage medium, and advertising dynamic generation equipment provided in the above embodiments can execute the advertising dynamic generation method provided in any embodiment of this application. For technical details not described in detail in the above embodiments, please refer to the advertising dynamic generation method provided in any embodiment of this application.

[0179] The above description is merely a preferred embodiment and the technical principles employed in this application. This application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions that can be made by those skilled in the art will not depart from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of this application. The scope of this application is determined by the scope of the claims.

Claims

1. An advertisement dynamic generation method characterized by comprising: include: Obtain user data of users who visit the ad display page, and determine user interest tags based on the user data; Based on the user interest tags and the trained generative adversarial network model, advertising copy and image materials corresponding to the user interest tags are generated; The size of the ad placement on the ad display page and the device type of the visiting user are obtained. Based on the ad placement size and the device type, the ad copy and the image material are dynamically rendered to generate a corresponding customized ad.

2. The method for dynamically generating advertisements according to claim 1, characterized in that, The process of determining user interest tags based on the user data includes: Extract browsing history, search history, and purchase history from the user data; The browsing history, the search history, and the purchase history are respectively processed by keyword segmentation and keyword vectorization to obtain the corresponding browsing feature vector, search feature vector, and purchase feature vector. User interest tags are determined based on the browsing feature vector, the search feature vector, and the purchase feature vector.

3. The method for dynamically generating advertisements according to claim 1, characterized in that, The process of determining user interest tags based on the user data includes: Based on the user data, several initial interest tags are determined, and the exposure data and feedback data of the visiting user for each of the initial interest tags are obtained within a preset statistical period. Calculate the user fatigue index corresponding to the initial interest tag based on the exposure data and the feedback data; User interest tags are determined from the initial interest tags using the user fatigue index.

4. The method for dynamically generating advertisements according to claim 1, characterized in that, The process of generating advertising copy and image materials corresponding to the user interest tags based on the user interest tags and the trained generative adversarial network model includes: Match target products based on the user's interest tags, and obtain product feature information of the target products, including product category, brand tags, function description and price range; The user interest tags and the product feature information are fused together through feature encoding and vector concatenation to obtain a fused feature vector; The fused feature vectors are input into the trained generative adversarial network model to generate advertising copy and image materials.

5. The method for dynamically generating advertisements according to any one of claims 1-4, characterized in that, The process of dynamically rendering the advertising copy and image materials based on the ad placement size and the device type to generate a corresponding customized ad includes: The element layout area and element layout density of the ad space are determined based on the ad space size and the device type. Adjust the layout position of the advertising copy and the image material according to the element layout area and its corresponding element layout density; The adjusted advertising copy and image materials are rendered and composited to generate corresponding customized advertisements.

6. The method for dynamically generating advertisements according to claim 5, characterized in that, The step of determining the element layout area and element layout density of the ad space based on the ad space size and the device type includes: Based on the size of the ad slot, a corresponding element layout object is generated, the element layout object is divided into several element layout areas, and the area weight factor is determined based on the relative position of the element layout area in the element layout object. Obtain the screen pixel density parameters of the accessed device according to the device type, and calculate the baseline layout density according to the screen pixel density parameters; The element layout density corresponding to each element layout region is determined based on the region weighting factor and the baseline layout density.

7. The method for dynamically generating advertisements according to claim 6, characterized in that, The step of adjusting the layout position of the advertising copy and the image material according to the element layout area and its corresponding element layout density includes: An initial image material is generated based on the image material and the element layout object, and the corresponding advertisement layout density of each element layout area in the initial image material is calculated. Calculate the layout density deviation for each element layout region using the advertisement layout density and the element layout density; Based on the layout density deviation, the text layout position is determined, and the advertising text is assigned to the text layout position.

8. An advertising dynamic generation device, characterized in that, include: The user tag module is used to obtain user data of users who visit the advertising display page and determine user interest tags based on the user data; The advertising creative module is used to generate advertising copy and image materials corresponding to the user interest tags based on the user interest tags and the trained generative adversarial network model; The ad generation module is used to obtain the ad space size of the ad display page and the device type of the accessing user, and dynamically render the ad copy and the image material based on the ad space size and the device type to generate a corresponding customized ad.

9. An advertising dynamic generation device, characterized in that, include: One or more processors; A memory that stores one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the dynamic advertising generation method as described in any one of claims 1-7.

10. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are used to perform the dynamic advertising generation method as described in any one of claims 1-7.

Images