E-commerce design automation workflow optimization system

Abstract

The invention discloses an e-commerce design automatic workflow optimization system, which is applied to commodity display image and advertisement map design of an e-commerce platform, combines an artificial intelligence technology and an automatic workflow, and optimizes an e-commerce design flow. Comprising an intelligent template recommendation module, an adaptive design optimization module, an intelligent typesetting and custom adjustment module, a cross-platform compatibility and adaptation module, a data driving optimization module and an A / B test and intelligent evolution module. Through intelligent template matching, automatic typesetting and AI optimization, the e-commerce advertisement design efficiency is improved, the visual quality is optimized, and the user click rate and the conversion rate are improved. And by adopting intelligent size adaptation, cross-platform compatibility is realized, and the adjustment cost is reduced. An A / B test and machine learning are combined, the advertisement is optimized, and the ROI is improved. Based on user portrait analysis and personalized recommendation, the advertisement style is adjusted, and the personalized experience is enhanced. Through intelligent advertisement generation and automatic optimization, the dependence on professional design skills is reduced, and the cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of e-commerce design, and in particular to an e-commerce design automated workflow optimization system. Background Art

[0002] In the current e-commerce industry, the visual display of goods is crucial to attracting consumers. E-commerce designers need to produce a large number of product display images and advertising materials to meet the promotion needs of major e-commerce platforms. These design tasks usually involve product image processing, model image generation, advertising copywriting layout, and multi-platform adaptation, requiring designers to balance efficiency and visual consistency while maintaining creativity. In recent years, artificial intelligence (AI) technology has made significant progress in the field of image generation and optimization, and some e-commerce companies have tried to use AI-assisted tools to improve design efficiency. However, due to the complexity of design tasks, existing tools are still difficult to fully meet the actual needs of e-commerce design, resulting in designers still needing to invest a lot of time and energy in manual adjustments.

[0003] Existing e-commerce design solutions still have many shortcomings. First, the design process is fragmented, and there is a lack of effective integration between different tools. Designers need to switch between multiple software, which affects the fluency of work; second, the degree of intelligence is limited. Although some AI tools can automatically generate images, they are still insufficient in personalized recommendations, style consistency, and detail optimization, which makes it difficult to ensure the quality of design results; in addition, multi-platform adaptation is complex, and different e-commerce platforms have different requirements for image size, format, and layout. Designers need to make manual adjustments for each platform, which is time-consuming and labor-intensive. In summary, the current industry urgently needs an integrated and intelligent e-commerce design automation system to improve design efficiency, optimize design quality, and reduce the workload of designers.

[0004] The prior art has the following deficiencies:

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide an e-commerce design automation workflow optimization system, which reduces the manual operation of designers through intelligent template matching, automatic typesetting and AI optimization, improves the efficiency of e-commerce advertising design, and enables them to focus on creative work. Computer vision analysis and intelligent color matching recommendations are used to optimize the visual quality of advertisements, improve user click-through rate and purchase conversion rate. An intelligent size adaptation algorithm is used to achieve cross-platform compatibility of advertisements and reduce adjustment costs between different e-commerce platforms. Combined with A / B testing and machine learning analysis, advertising design is optimized in real time to ensure that advertisements continue to adapt to market demand and improve ROI. Based on user portrait analysis and personalized recommendations, the advertising style is adjusted for different user groups to improve personalized experience and conversion rate. Through intelligent advertising generation and automated optimization, the dependence of e-commerce operators on professional design skills is reduced, costs are reduced, design efficiency is improved, and e-commerce operations are made more intelligent to solve the problems in the above-mentioned background technology.

[0007] In order to achieve the above-mentioned purpose, the present invention provides the following technical solutions: an e-commerce design automated workflow optimization system, which is applied to the design of product display images and advertising images on e-commerce platforms, combines artificial intelligence technology and automated workflow to optimize the e-commerce design process, including an intelligent template recommendation module, an adaptive design optimization module, an intelligent typesetting and custom adjustment module, a cross-platform compatibility and adaptation module, a data-driven optimization module, and an A / B testing and intelligent evolution module;

[0008] Intelligent template recommendation module obtains basic information of products and automatically recommends matching design templates based on the basic information using artificial intelligence models. Design template recommendations are based on e-commerce platform standards, brand historical data, and user behavior analysis to improve the pertinence and consistency of designs.

[0009] The adaptive design optimization module automatically recommends the best design elements based on the recommended template, product characteristics, target audience and brand style, and combines deep learning algorithms to make adaptive adjustments to ensure optimal visual effects;

[0010] Intelligent layout and custom adjustment module, which uses automated processing technology to intelligently adjust advertising images and supports user-defined adjustments to improve image quality and design flexibility;

[0011] Cross-platform compatibility and adaptation module, which identifies the requirements of e-commerce platforms, automatically adjusts the output format of design works, ensures compatibility and consistency on different e-commerce platforms, and automatically adapts to the display requirements of mobile and PC terminals;

[0012] The data-driven optimization module dynamically optimizes the visual elements of the ad images based on the data from the e-commerce platform, automatically adjusts the design plan based on data analysis, and continuously improves the ad performance;

[0013] The A / B testing and intelligent evolution module automatically generates multiple versions of advertising images and conducts A / B testing to compare the advertising effects of different versions, select the best version and optimize future design strategies to achieve continuous evolution of design solutions.

[0014] Preferably, the commodity information parsing step further includes commodity image analysis, specifically:

[0015] Obtain the original image data of the product to be designed, and use computer vision technology to analyze the image to extract the product's outline, color distribution, texture characteristics and lighting conditions; classify the products through a deep learning model to identify their categories, such as clothing, home appliances, food, digital products, etc., and assign different design weights according to category labels; combine the brand style library to analyze whether the product's design style is consistent with the brand's existing visual image. If not, generate optimization suggestions for designers to refer to; at the same time, based on the target audience's historical interaction data, optimize the product's display angle, background style and recommended color scheme to maximize the product's visual appeal.

[0016] Preferably, the intelligent template matching step further includes template adaptive adjustment, specifically:

[0017] After selecting a matching template, the adaptive image transformation algorithm is used to automatically adjust the aspect ratio of the template according to the requirements of the e-commerce platform, and adjust the image cropping area while ensuring visual balance; based on the product feature parameters, the template layout is dynamically adjusted, including text position, product image size, blank area and auxiliary design elements, to meet the display requirements of different products; when a product has multiple display angles or variants, the system automatically adjusts the template to ensure that all variants can be reasonably presented in the design; in addition, combined with historical click-through rate and conversion rate data, the arrangement order of template elements is optimized, so that users can quickly focus on the key features of the product when browsing the advertisement, thereby improving the attractiveness and effectiveness of the advertisement.

[0018] Preferably, the automated design optimization step further includes dynamic typesetting based on artificial intelligence, specifically:

[0019] The system uses a neural network model to optimize the layout of the advertising image to maximize user attention. First, the design draft is analyzed through a convolutional neural network (CNN), the main visual focus is identified, and the weight of each design element is calculated to ensure that the core product area maintains optimal visibility. Then, a reinforcement learning algorithm is used to adjust the text, buttons, and background elements of the advertisement to ensure the clarity of information conveyed and prevent design elements from overlapping or causing visual interference. In the final optimization stage, the system performs simulation tests on the advertising image, simulates different user perspectives, analyzes the user's attention path, and fine-tunes the position of the image elements on this basis to ensure that the advertising information can be delivered to the user in the shortest time, thereby improving the click-through rate and conversion effect of the advertisement.

[0020] Preferably, the intelligent design recommendation step further includes personalized design recommendations based on user preferences, specifically:

[0021] The system combines users' browsing and purchasing data to analyze their preference patterns and generates personalized design plans based on historical data. First, clustering algorithms are used to divide users into different consumer groups, such as fashion-preferring users, cost-effective users, high-end luxury users, etc., and corresponding design style labels are assigned to different groups. Then, based on a personalized recommendation model, the system recommends corresponding fonts, color schemes, and advertising layouts for different user groups. For example, it recommends vibrant colors to young users and simple and steady colors to business people. Finally, the system combines real-time data analysis to monitor the effects of different recommendation plans and automatically adjusts the recommendation weights to ensure that each advertisement is matched with the most appropriate target user.

[0022] Preferably, the multi-platform adaptation step further includes adaptive layout optimization, specifically:

[0023] The mathematical modeling method is used to standardize the size requirements of different e-commerce platforms and calculate the optimal size parameters of the advertising images. The optimization formula is as follows:

[0024] Assume that the set of e-commerce platforms is P = {p i}={p 1 , p 2 ,…,p n}, each platform p i With size requirements (w i ,h i ), where: w i Represents platform p i Required ad image width (unit: pixel px), h i Represents platform p i The required ad image height (unit: pixel px), the optimal adaptation size calculation formula is:

[0025] W opt =max(w 1 , w 2 ,…,w n ), H opt =max(h 1 ,h 2 ,…,h n )

[0026] , where W opt and H opt are the optimal fitting width and height respectively;

[0027] Set the adjustment coefficient α for different platforms i and β i , so that the final image adapts to the display ratio of a specific platform, the adjusted size is calculated as follows:

[0028] W i =W opt α i , H i =H opt β i

[0029] , where α i is the width scaling factor of the i-th e-commerce platform, β i is the height scaling factor of the ith e-commerce platform, W i It represents the optimized advertising image on the i-th e-commerce platform p i The final width (in pixels), H i It represents the optimized advertising image on the i-th e-commerce platform p i The final height on (unit: pixel px);

[0030] The gradient descent optimization algorithm is used to adjust the image layout so that the advertising elements are evenly distributed and meet the platform specifications. The optimization objective function is as follows:

[0031]

[0032] , where: w i represents the i-th e-commerce platform p i Required standard width of the ad image (unit: pixels px), h i represents the i-th e-commerce platform p i The required standard height of the ad image (in pixels px);

[0033] This objective function is used to measure the deviation between the optimized ad size and the platform standard size and reduce the deviation as much as possible. The goal is to minimize the size error of all platforms so that the ad image can be displayed as close to the standard size as possible on all platforms.

[0034] Preferably, the real-time optimization and feedback adjustment step further includes data-driven optimization based on A / B testing, specifically:

[0035] Set the advertising design solution set A = {a k}={a 1 , a 2 , …, a n}, where n represents the total number of advertising design schemes, and each advertising design scheme a k With an initial click rate of C k and the initial conversion rate T k , conduct A / B testing and record the number of clicks c of each advertising design on user group U k and number of purchases t k , calculate the new click rate and conversion rate, the calculation expression is:

[0036]

[0037] , where: C′ k Design a plan for advertising k The actual click rate in the test group, T′ k Design a plan for advertising k The actual conversion rate, that is, the proportion of completed purchases after clicking;

[0038] The role of the initial click-through rate and initial conversion rate is to provide benchmark data for A / B testing, so that the system can make effective comparisons and adjustments during the ad design optimization process. The initial click-through rate represents the theoretical click-through rate of the ad design before the test begins. It is usually estimated from historical data, industry benchmarks, or forecasting models, reflecting the expected click performance of the scheme in similar environments. The initial conversion rate represents the theoretical conversion rate of the ad design, that is, the proportion of users who complete purchases, registrations, or other target behaviors after clicking on the ad. It is also estimated based on historical data or market research. The main role of these two parameters is to provide reference standards for ad optimization. The system can judge whether the performance of each ad scheme meets expectations by comparing the actual click-through rate and actual conversion rate after the test, and make corresponding adjustments.

[0039] For example, if the actual click-through rate of a solution is much lower than the initial click-through rate, the system can adjust the color, layout or copy to increase user attention; if the actual conversion rate is lower than the initial expectation, it may be necessary to optimize the product display method or enhance purchase guidance. By comparing these data, the system can dynamically adjust the ad design to ensure that the final ad solution can maximize the conversion rate and ad revenue.

[0040] The optimization objective function is set so that the selected advertising design maximizes the click-through rate and conversion rate while ensuring design consistency. The expression of the optimization objective function is:

[0041]

[0042] , where λ 1 and λ 2 is the weight coefficient, which is adjusted according to marketing needs;

[0043] Through multiple rounds of iterations, design elements (such as color, font, and layout) are automatically adjusted, and ultimately converge to the optimal advertising design plan, thereby improving the effectiveness of advertising.

[0044] Preferably, the intelligent design recommendation step further includes context-aware optimization, specifically:

[0045] When recommending design solutions, the system automatically adjusts the design style based on external contextual information, such as seasons, holidays, and hot events. For example, during major promotions such as Double Eleven and Black Friday, the system automatically recommends more impactful red and orange color schemes to enhance purchasing desire; during spring promotions, it recommends fresh and bright green tones to match the seasonal atmosphere; in addition, the system can also optimize advertising based on weather information, such as adding raindrop effects when promoting waterproof products on rainy days to enhance the advertising's contextual awareness, making the advertising content more targeted and increasing users' willingness to buy.

[0046] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0047] The present invention greatly reduces the manual operations of designers in e-commerce advertising production through intelligent template matching, automated design optimization and workflow automation. Traditional advertising design usually requires designers to manually select appropriate templates, adjust the size of product images, set layout and color matching, and the whole process is time-consuming and cumbersome. The present invention uses AI-driven intelligent template matching, which can automatically recommend the best template based on product category, brand style and target users, and optimize product images, texts and interactive buttons in combination with automated typesetting technology. In this way, designers can make fine adjustments based on what the system provides, instead of designing from scratch, thereby significantly shortening the advertising production cycle, improving design efficiency, and enabling designers to focus more on creative work.

[0048] The present invention ensures that advertisements have stronger visual appeal through computer vision analysis, intelligent layout optimization and color recommendation engine. Traditional advertising design often relies on the personal experience of designers, which may lead to uneven aesthetics and information transmission effects of advertisements. The present invention uses AI visual recognition technology to automatically analyze the focus, color, and light and shadow effects of product images, and optimizes the layout structure of advertisements in combination with user browsing habits to ensure that products, texts, and interactive elements are arranged according to the best visual path. In addition, the present invention can automatically match the optimal color scheme to ensure that the overall color matching of the advertisement is in line with the brand style and the aesthetic preferences of the target users. Through these technologies, the present invention can effectively improve the visual quality of advertisements, making them more attractive, thereby increasing users' click-through rate and desire to buy.

[0049] The present invention ensures that advertisements are applicable to multiple e-commerce platforms such as Taobao, JD.com, and Pinduoduo through automatic size adjustment and cross-platform adaptation technology, reducing the workload of designers to repeatedly adjust advertisements between different platforms. In traditional methods, designers need to manually adjust the size, layout, and format of advertisements for different e-commerce platforms. Especially when a brand needs to advertise on multiple platforms, the adjustment cost is very high. The present invention adopts an intelligent size adaptation algorithm, which can automatically adjust the aspect ratio, text layout, image ratio and other parameters of the advertisement according to the standards of different platforms to ensure that the display effect of the advertisement remains consistent on each platform. In addition, the present invention can also automatically adjust the font size and interactive button position according to the display characteristics of different terminals to adapt to the user experience of different devices, thereby improving the adaptability and display effect of the advertisement.

[0050] The present invention can optimize the advertising design in real time through A / B testing, user behavior monitoring and machine learning analysis, so that it can continuously adapt to market demand and improve the effect of advertising. Traditional advertising optimization often relies on the experience of designers or the feedback of the marketing team, with a long adjustment cycle and it is difficult to ensure the correct optimization direction. The present invention adopts an A / B testing mechanism, which can monitor data such as click-through rate, conversion rate, user stay time, etc. in real time after the advertisement is released, and analyze the performance of different advertising versions through a machine learning model. For example, the system can automatically test the impact of different color schemes, fonts, and button sizes on user behavior, and automatically select the design with the best performance. In addition, the system can also predict the optimal design mode of advertising based on historical data, making advertising delivery more accurate and improving the ROI of advertising.

[0051] The present invention adopts user portrait analysis and personalized recommendation engine, which can automatically adjust the design of advertisements for different user groups, thereby improving the personalization of advertisements and improving conversion rates. In traditional e-commerce advertising design, the advertising content seen by all users is basically the same, and it is impossible to make personalized adjustments based on the preferences of different users. The present invention uses machine learning to analyze users' browsing history, purchasing behavior, and interest preferences, and can match the most suitable advertising design for different user groups. For example, for young users, the system may recommend more lively and bright colors and dynamic layouts; for high-end business people, the system may choose a more simple and atmospheric style. In addition, the present invention can also automatically adjust the advertising style in combination with seasons, festivals, and hot events, so that the advertisements are more in line with current market demand, thereby improving users' click-through rates and purchase conversion rates.

[0052] The present invention uses intelligent advertising generation and automated optimization to enable e-commerce operators to quickly generate high-quality advertising designs even if they do not have professional design skills, thereby reducing the threshold and cost of advertising production. In traditional e-commerce marketing, merchants often need to hire professional graphic designers or use complex design software to produce advertisements, which not only increases operating costs but also reduces the efficiency of advertising production. The present invention uses intelligent and automated design tools, so that ordinary operators only need to input product information, and the system can automatically generate high-quality advertising images that meet the brand style, and optimize and adapt to multiple platforms. In this way, merchants can reduce their dependence on professional design teams, reduce labor costs, and at the same time greatly improve the speed and quality of advertising production, making e-commerce operations more efficient and intelligent. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0054] Figure 1 A schematic diagram of the modules of the automated workflow optimization system for e-commerce design of the present invention. DETAILED DESCRIPTION

[0055] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0056] The present invention provides Figure 1The e-commerce design automation workflow optimization system shown is applied to the product display image and advertising image design of the e-commerce platform. It combines artificial intelligence technology and automated workflow to optimize the e-commerce design process, including intelligent template recommendation module, adaptive design optimization module, intelligent typesetting and custom adjustment module, cross-platform compatibility and adaptation module, data-driven optimization module, and A / B testing and intelligent evolution module.

[0057] Intelligent template recommendation module obtains basic information of products and automatically recommends matching design templates based on the basic information using artificial intelligence models. Design template recommendations are based on e-commerce platform standards, brand historical data, and user behavior analysis to improve the pertinence and consistency of designs.

[0058] The adaptive design optimization module automatically recommends the best design elements based on the recommended template, product characteristics, target audience and brand style, and combines deep learning algorithms to make adaptive adjustments to ensure optimal visual effects;

[0059] Intelligent layout and custom adjustment module, which uses automated processing technology to intelligently adjust advertising images and supports user-defined adjustments to improve image quality and design flexibility;

[0060] Cross-platform compatibility and adaptation module, which identifies the requirements of e-commerce platforms, automatically adjusts the output format of design works, ensures compatibility and consistency on different e-commerce platforms, and automatically adapts to the display requirements of mobile and PC terminals;

[0061] The data-driven optimization module dynamically optimizes the visual elements of the ad images based on the data from the e-commerce platform, automatically adjusts the design plan based on data analysis, and continuously improves the ad performance;

[0062] The A / B testing and intelligent evolution module automatically generates multiple versions of advertising images and conducts A / B testing to compare the advertising effects of different versions, select the best version and optimize future design strategies to achieve continuous evolution of design solutions.

[0063] The product information analysis step further includes product image analysis, specifically:

[0064] Obtain the original image data of the product to be designed, and use computer vision technology to analyze the image to extract the product's outline, color distribution, texture characteristics and lighting conditions; classify the products through a deep learning model to identify their categories, such as clothing, home appliances, food, digital products, etc., and assign different design weights according to category labels; combine the brand style library to analyze whether the product's design style is consistent with the brand's existing visual image. If not, generate optimization suggestions for designers to refer to; at the same time, based on the target audience's historical interaction data, optimize the product's display angle, background style and recommended color scheme to maximize the product's visual appeal.

[0065] The intelligent template matching step further includes template adaptive adjustment, specifically:

[0066] After selecting a matching template, the adaptive image transformation algorithm is used to automatically adjust the aspect ratio of the template according to the requirements of the e-commerce platform, and adjust the image cropping area while ensuring visual balance; based on the product feature parameters, the template layout is dynamically adjusted, including text position, product image size, blank area and auxiliary design elements, to meet the display requirements of different products; when a product has multiple display angles or variants, the system automatically adjusts the template to ensure that all variants can be reasonably presented in the design; in addition, combined with historical click-through rate and conversion rate data, the arrangement order of template elements is optimized, so that users can quickly focus on the key features of the product when browsing the advertisement, thereby improving the attractiveness and effectiveness of the advertisement.

[0067] The automated design optimization step further includes AI-based dynamic typesetting, specifically:

[0068] The system uses a neural network model to optimize the layout of the advertising image to maximize user attention. First, the design draft is analyzed through a convolutional neural network (CNN), the main visual focus is identified, and the weight of each design element is calculated to ensure that the core product area maintains optimal visibility. Then, a reinforcement learning algorithm is used to adjust the text, buttons, and background elements of the advertisement to ensure the clarity of information conveyed and prevent design elements from overlapping or causing visual interference. In the final optimization stage, the system performs simulation tests on the advertising image, simulates different user perspectives, analyzes the user's attention path, and fine-tunes the position of the image elements on this basis to ensure that the advertising information can be delivered to the user in the shortest time, thereby improving the click-through rate and conversion effect of the advertisement.

[0069] The intelligent design recommendation step further includes personalized design recommendations based on user preferences, specifically:

[0070] The system combines users' browsing and purchasing data to analyze their preference patterns and generates personalized design plans based on historical data. First, clustering algorithms are used to divide users into different consumer groups, such as fashion-preferring users, cost-effective users, high-end luxury users, etc., and corresponding design style labels are assigned to different groups. Then, based on a personalized recommendation model, the system recommends corresponding fonts, color schemes, and advertising layouts for different user groups. For example, it recommends vibrant colors to young users and simple and steady colors to business people. Finally, the system combines real-time data analysis to monitor the effects of different recommendation plans and automatically adjusts the recommendation weights to ensure that each advertisement is matched with the most appropriate target user.

[0071] The multi-platform adaptation step further includes adaptive layout optimization, specifically:

[0072] The mathematical modeling method is used to standardize the size requirements of different e-commerce platforms and calculate the optimal size parameters of the advertising images. The optimization formula is as follows:

[0073] Assume that the set of e-commerce platforms is P = {p i}={p 1 , p 2 ,…,p n}, each platform p i With size requirements (w i ,h i ), where: w i Represents platform p i Required ad image width (unit: pixel px), h i Represents platform p i The required ad image height (unit: pixel px), the optimal adaptation size calculation formula is:

[0074] W opt =max(w 1 , w 2 ,…,w n ), H opt =max(h 1 ,h 2 ,…,h n )

[0075] , where W opt and H opt are the optimal fitting width and height respectively;

[0076] Set the adjustment coefficient α for different platforms i and β i , so that the final image adapts to the display ratio of a specific platform, the adjusted size is calculated as follows:

[0077] Wi =W opt α i , H i =H opt β i

[0078] , where α i is the width scaling factor of the i-th e-commerce platform, β i is the height scaling factor of the ith e-commerce platform, W i It represents the optimized advertising image on the i-th e-commerce platform p i The final width (in pixels), H i It represents the optimized advertising image on the i-th e-commerce platform p i The final height on (unit: pixel px);

[0079] The gradient descent optimization algorithm is used to adjust the image layout so that the advertising elements are evenly distributed and meet the platform specifications. The optimization objective function is as follows:

[0080]

[0081] , where: w i represents the i-th e-commerce platform p i Required standard width of the ad image (unit: pixels px), h i represents the i-th e-commerce platform p i The required standard height of the ad image (in pixels px);

[0082] This objective function is used to measure the deviation between the optimized ad size and the platform standard size and reduce the deviation as much as possible. The goal is to minimize the size error of all platforms so that the ad image can be displayed as close to the standard size as possible on all platforms.

[0083] The real-time optimization and feedback adjustment steps further include data-driven optimization based on A / B testing, specifically:

[0084] Set the advertising design solution set A = {a k}={a 1 , a 2 , …, a n}, where n represents the total number of advertising design schemes, and each advertising design scheme a k With an initial click rate of C k and the initial conversion rate T k , conduct A / B testing and record the number of clicks c of each advertising design on user group U k and number of purchases t k , calculate the new click rate and conversion rate, the calculation expression is:

[0085]

[0086] , where: C′ k Design a plan for advertising k The actual click rate in the test group, T′ k Design a plan for advertising k The actual conversion rate, that is, the proportion of completed purchases after clicking;

[0087] The role of the initial click-through rate and initial conversion rate is to provide benchmark data for A / B testing, so that the system can make effective comparisons and adjustments during the ad design optimization process. The initial click-through rate represents the theoretical click-through rate of the ad design before the test begins. It is usually estimated from historical data, industry benchmarks, or forecasting models, reflecting the expected click performance of the scheme in similar environments. The initial conversion rate represents the theoretical conversion rate of the ad design, that is, the proportion of users who complete purchases, registrations, or other target behaviors after clicking on the ad. It is also estimated based on historical data or market research. The main role of these two parameters is to provide reference standards for ad optimization. The system can judge whether the performance of each ad scheme meets expectations by comparing the actual click-through rate and actual conversion rate after the test, and make corresponding adjustments.

[0088] For example, if the actual click-through rate of a solution is much lower than the initial click-through rate, the system can adjust the color, layout or copy to increase user attention; if the actual conversion rate is lower than the initial expectation, it may be necessary to optimize the product display method or enhance purchase guidance. By comparing these data, the system can dynamically adjust the ad design to ensure that the final ad solution can maximize the conversion rate and ad revenue.

[0089] The optimization objective function is set so that the selected advertising design maximizes the click-through rate and conversion rate while ensuring design consistency. The expression of the optimization objective function is:

[0090]

[0091] , where λ 1 and λ 2 is the weight coefficient, which is adjusted according to marketing needs;

[0092] Through multiple rounds of iterations, design elements (such as color, font, and layout) are automatically adjusted, and ultimately converge to the optimal advertising design plan, thereby improving the effectiveness of advertising.

[0093] The smart design recommendation step further includes context-aware optimization, specifically:

[0094] When recommending design solutions, the system automatically adjusts the design style based on external contextual information, such as seasons, holidays, and hot events. For example, during major promotions such as Double Eleven and Black Friday, the system automatically recommends more impactful red and orange color schemes to enhance purchasing desire; during spring promotions, it recommends fresh and bright green tones to match the seasonal atmosphere; in addition, the system can also optimize advertising based on weather information, such as adding raindrop effects when promoting waterproof products on rainy days to enhance the advertising's contextual awareness, making the advertising content more targeted and increasing users' willingness to buy.

[0095] Implementation method 1: Intelligent template matching based on product information analysis;

[0096] This implementation method is aimed at the intelligent parsing and template matching optimization of product information in the process of e-commerce advertising design, aiming to improve the intelligence and matching accuracy of advertising design through artificial intelligence and automation technology, thereby enhancing the attractiveness and marketing effect of advertising. First, the system obtains the basic information of the product, including product name, category (such as clothing, home appliances, food, etc.), brand style, target audience (such as young people, business people, housewives, etc.) and color characteristics, and constructs a product feature parameter set based on this information. At the same time, the system can also extract the associated features of the product from the product database, historical order records and user feedback of the e-commerce platform, such as whether the product is a best-seller, the average browsing time of users, the purchase conversion rate, etc., to assist the subsequent intelligent matching process. The construction of the product feature parameter set adopts a machine learning method, and by analyzing a large amount of historical design data, the best design style applicable to different product categories is summarized, such as recommending simple and elegant templates for high-end clothing, and recommending colorful and cartoon templates for children's toys, so as to ensure that the matching scheme meets the industry design trend and user aesthetic habits.

[0097] During the product information parsing process, the system further uses computer vision technology to conduct in-depth analysis of product images, including extracting key visual elements such as the product's outline, color distribution, texture features, and lighting conditions. Specifically, the system extracts features from product images through deep learning models (such as convolutional neural networks (CNNs)) and uses clustering algorithms to classify the colors of products to determine suitable background colors and color schemes. For example, if the main color of a product is blue, the system will recommend a background color with a higher contrast (such as white or yellow) to enhance the visual prominence of the product; if the material of the product is relatively smooth (such as glass or metal products), the system will recommend design elements with highlights and reflections to highlight the texture of the product. In addition, the system can also identify whether the product contains a specific brand logo, and ensure that brand elements are not blocked or weakened during the template matching process to enhance brand recognition and user trust.

[0098] After completing the product information parsing, the system enters the intelligent template matching stage. The system has a built-in preset template library, which contains a variety of design templates with different styles, layouts, and colors. Each template is optimized to adapt to different product categories, advertising types (such as banner ads, social media ads, mobile ads), and e-commerce platform format requirements. The template matching process adopts a multi-layer screening mechanism. First, a group of applicable templates are screened out based on product categories, brand styles, and target audiences. Then, a secondary screening is performed based on product image features to ensure that the visual style of the selected template matches the product itself. For example, for luxury goods, the system gives priority to minimalist templates, reduces unnecessary decorative elements, and makes the product itself the visual focus; for technology products, the system will select templates with a sense of technology and futurism, and add dynamic lighting effects or gradient color backgrounds to enhance the technological atmosphere of the advertisement.

[0099] After the template matching is completed, the system will also make intelligent adjustments so that the template can adapt to the display standards of different e-commerce platforms. For example, e-commerce platforms such as Taobao, JD.com, and Pinduoduo have different requirements for the size, layout, and information level of advertising images. The system will automatically adjust the template's aspect ratio, text layout, product image size and other parameters to ensure that the advertisement can be perfectly adapted to different platforms. In addition, the system can also optimize the template layout based on historical click-through rate, conversion rate and other data. For example, if data analysis shows that a certain template structure can increase the user's browsing time, the system will give priority to recommending similar templates; if an advertisement with a certain color scheme has a low conversion rate in a specific category, the system will automatically adjust the color scheme to match the target user's aesthetic preferences.

[0100] In scenarios with multiple SKUs (such as products of different colors, sizes, and styles), the system can automatically generate multiple versions of advertising designs and ensure that the visual display of each SKU is optimized. Specifically, the system first adjusts the template layout according to the attributes of different SKUs. For example, in advertisements for clothing with multiple colors, the system will choose to display different SKUs in a color gradient to enhance the visual hierarchy; in advertisements for electronic products with multiple functions, the system will adjust the order of product arrangements to highlight the main selling points. In addition, the system also supports dynamic content generation, allowing e-commerce designers to quickly adjust the text, icons, and background elements in the template according to different market promotion needs, thereby achieving large-scale automated advertising design and improving the flexibility and efficiency of marketing activities.

[0101] Finally, designers can make personalized adjustments based on the system's recommendations, including modifying font styles, adjusting image transparency, adding special visual effects, etc., to ensure that the final ad design meets brand requirements and has unique creativity. After the design is completed, the system will also provide a design quality assessment function to score the effectiveness of the ad design based on visual aesthetic principles, user attention heat map analysis, historical data comparison and other methods, and make optimization suggestions. For example, the system may suggest adjusting the contrast of the ad copy to improve readability, or suggesting reducing the size of certain non-core elements to enhance the visual impact of the product. Through this complete intelligent template matching process, the efficiency of e-commerce ad design is greatly improved, while the attractiveness and click-through rate of the ad can also be effectively enhanced, thereby achieving a higher marketing conversion rate.

[0102] Implementation method 2: automated design optimization and intelligent typesetting;

[0103] This implementation method is aimed at automated optimization and intelligent typesetting in e-commerce advertising design, aiming to improve the visual performance, information communication efficiency and user interaction rate of advertisements through artificial intelligence and data-driven methods, thereby enhancing the attractiveness of advertisements and improving conversion effects. First, the system receives basic information about the product, including product categories (such as clothing, home appliances, food, etc.), brand style, target audience characteristics (such as age, gender, purchasing preferences) and product images, and combines this information to build the initial data set required for advertising design. After generating the preliminary design plan, the system uses artificial intelligence optimization algorithms to automatically adjust the typesetting structure, graphic layout, color matching, interactive elements, etc. in the advertisement to ensure that the final design meets the best visual effects and the standard requirements of the e-commerce platform.

[0104] In the process of intelligent typesetting optimization, the system first uses computer vision analysis to identify the main visual focus of the product image. For example, in clothing advertisements, the system can identify the model's face, the main body of the clothing and the brand logo, and ensure that these key areas are not blocked by text or other elements. In order to further optimize the readability and information transmission efficiency of the advertisement, the system predicts the user's visual attention distribution based on heat map analysis, and adjusts the arrangement of the advertising content so that users can obtain key information in the shortest time. For example, the system can automatically place price information and discount information in the area where the user's line of sight is prioritized (such as the upper left corner or the central area), and optimize the position of the CTA (call to action) button to guide users to complete the click or purchase operation. In addition, the system uses an adaptive text typesetting algorithm, which can automatically adjust the font size, line spacing, alignment, etc. according to the size of the advertising image and the length of the text to ensure that the text content is clear and readable without affecting the overall visual beauty.

[0105] In order to further enhance the attractiveness and brand consistency of advertisements, the system has a built-in intelligent color recommendation engine that can automatically generate the optimal color scheme based on the color characteristics and brand style of the product. For example, in the advertising design of fashion products, the system will give priority to color combinations that are in line with popular trends, such as high-grade gray, Morandi color, etc.; while in advertisements for electronic products, the system may recommend cool tones with a strong sense of technology, such as blue or silver. In addition, the system can also optimize color recommendations based on users' historical behavior data (such as the color scheme with the highest click-through rate in the past) to ensure that the advertising design is highly matched with the aesthetic preferences of the target audience. In terms of advertising background optimization, the system can automatically select the appropriate background style according to the type of product. For example, warm-toned backgrounds are recommended for food advertisements to enhance appetite, while simple, modern-style backgrounds are recommended for high-end home product advertisements to highlight the high-end feel of the product.

[0106] In terms of dynamic layout optimization, the system uses reinforcement learning algorithms to perform multiple rounds of iterative optimization, and constantly adjusts the layout of advertising elements to achieve the best visual effect. Specifically, the system first generates multiple layout schemes, and uses historical data to evaluate the click-through rate and conversion rate of different layout schemes. For example, the system can test different product image sizes, text positions, and button styles, and obtain actual user feedback through A / B testing. Then, the system analyzes the performance of different schemes based on the reinforcement learning algorithm and automatically adjusts the layout strategy for the next round of optimization. For example, if the experimental results show that a version with a certain button size has a higher click-through rate, the system will give priority to this size in subsequent designs; if a certain copywriting layout method causes a high user bounce rate, the system will automatically try a layout scheme that is more in line with reading habits. On this basis, the system can also perform adaptive layout in combination with user device types (such as PC, mobile, and tablet) to ensure that advertisements can achieve the best display effect on different devices.

[0107] In terms of user interaction optimization, the system supports personalized advertising design and can automatically adjust advertising content according to different user groups. For example, for young consumers, the system may add dynamic visual effects, such as gradient backgrounds, animated buttons, etc., to enhance the interactivity of advertisements; for business people, the system may choose a more concise and formal design style to highlight the professionalism of the brand. In addition, the system can also make personalized recommendations based on user behavior data (such as recently browsed products and purchase records), such as automatically adding "Guess you like" or "Hot-selling products" modules in advertisements to increase users' purchasing interest.

[0108] After the ad is released, the system will also conduct real-time data monitoring, track key indicators such as the click-through rate, conversion rate, user dwell time, and dynamically optimize based on this data. For example, if the click-through rate of a certain version of the ad is lower than expected, the system can automatically adjust the color scheme, modify the copy expression, and even try different combinations of visual elements to increase the attractiveness of the ad. In addition, the system also supports multiple rounds of A / B testing, can run multiple ad versions at the same time, and select the best plan for large-scale promotion based on the test results. For example, the system can test the red button and blue button ad versions at the same time, and after determining that the red button has a higher conversion rate, automatically use the red button version as the final delivery plan.

[0109] Implementation method three: real-time optimization and feedback adjustment based on A / B testing;

[0110] This implementation method is aimed at data-driven optimization and real-time feedback adjustment after the e-commerce advertisement is released, and aims to achieve continuous iterative optimization of advertisement design through technical means such as A / B testing, machine learning analysis, and user behavior monitoring, so as to improve click-through rate, conversion rate and overall marketing effect. In traditional e-commerce advertising design, the advertisement is usually released directly after the design is completed. Designers often rely on experience for optimization and lack scientific data analysis methods, resulting in a long adjustment cycle for advertisement design and the optimization direction is not accurate enough. This implementation method uses the A / B testing mechanism to monitor key data such as users' click behavior, browsing habits, conversion paths, etc. in real time after the advertisement is released, and uses machine learning models for intelligent analysis to dynamically optimize advertising design elements (such as color, font, layout, CTA button, background style, etc.) to ensure that the advertising plan can continuously adapt to market demand and improve the user appeal and conversion efficiency of the advertisement.

[0111] First, during the advertising delivery phase, the system will generate multiple versions of advertising designs for the same product or activity, each with slight differences in color, layout, text description, button style, etc. The system will randomly push these different versions of ads to the target user group and record the core indicators of each version, such as click-through rate, conversion rate, user dwell time, and bounce rate. For example, an ad may have two versions: version A uses a red CTA button, and version B uses a blue CTA button. The system can push these two versions to a portion of users to analyze the impact of different colors on user click-through rates. During the data collection process, the system will not only count the user's click behavior, but also analyze the user's interactive behavior on the ad page, such as whether the user quickly swipes to skip the ad, whether they stay for a long time, whether they click on the ad to complete the purchase, etc., so as to obtain more accurate user feedback data.

[0112] During the data analysis phase, the system uses machine learning algorithms to identify the strengths and weaknesses of each ad version and calculates the market performance score of each design. For example, if version A has a higher click-through rate than version B, but version B has a higher conversion rate, the system will further analyze user behavior data to determine whether version A is more attractive or version B has a higher purchase conversion. In addition, the system will conduct in-depth analysis based on user segmentation data, such as whether young users prefer ads of specific colors, and whether users of a certain age group are more likely to click on ads with discount information, thereby optimizing the personalized matching of ad content. Based on these data analysis results, the system will intelligently filter ad versions, automatically select the best ad version as the main delivery plan, and accurately push it based on user portraits, that is, recommend ad designs that best suit their preferences to different types of users.

[0113] During the ad optimization and adjustment phase, the system will make dynamic adjustments based on the A / B test results to ensure that the ad design is always in the best state. For example, if the test results show that ads with light backgrounds perform better on mobile devices, while ads with dark backgrounds have higher click-through rates on PCs, the system will automatically adjust the delivery strategy to mainly push ads with light backgrounds on mobile devices and prioritize ads with dark backgrounds on PCs. In addition, the system can also adjust ad content in real time. For example, when it is found that the click-through rate of a certain version of an ad has decreased, the system will automatically adjust the color of the CTA button, optimize the ad copy, adjust the position of the product image, and even generate a new ad version for further testing to ensure continuous optimization of the ad.

[0114] In order to further improve the efficiency of advertising, the system will also combine historical data and market trend analysis for forecasting and optimization. For example, the system can predict the possible performance of the current advertisement based on the click-through rate and conversion rate data of similar advertisements in the past, and optimize the design plan in advance. If a certain type of product has performed well in past promotions using advertisements with a specific layout or color scheme, the system can automatically refer to these historical data to generate advertisements with a similar style, and focus on testing the effectiveness of this style during the A / B testing phase. In addition, the system can also automatically adjust the visual style and copy of the advertisement in combination with external factors (such as festivals, hot events, weather changes, etc.). For example, during promotional activities such as Double Eleven and Black Friday, the system can automatically add discount labels and countdown elements, and automatically adjust the color scheme of the advertisement during holidays (such as Christmas and New Year) to make it more in line with the festive atmosphere, so as to enhance the user's emotional resonance and desire to buy.

[0115] This implementation also supports cross-platform data integration, that is, the system can not only optimize advertising on a single e-commerce platform (such as Taobao, JD.com), but also integrate data from multiple platforms to form an omni-channel advertising optimization strategy. For example, if a certain version of an advertisement has a high conversion rate on Taobao, but a low click-through rate on JD.com, the system can analyze the user habits of different platforms and adjust the advertising content of different platforms to make it more in line with the user characteristics of each platform. In addition, the system also supports multi-terminal optimization, that is, advertising adjustments for different devices (such as PC, mobile, and APP), such as optimizing the readability of advertisements on mobile terminals (such as increasing font size and simplifying layout), and optimizing the interactivity of advertisements on PC terminals (such as adding mouse hover animation effects), to ensure that advertisements can provide the best user experience on different terminals.

[0116] During the long-term optimization process, the system will continuously accumulate and learn A / B test data and form an advertising optimization model, which can continuously improve the intelligence level of advertising design. For example, the system can automatically summarize the optimal design mode, summarize which colors, layouts, and copywriting are more likely to attract users, and automatically generate new advertising design plans based on these data, thereby reducing the manual adjustment work of designers and improving the intelligence level of advertising design. In addition, the system can also form an intelligent delivery strategy, that is, adopt different advertising design styles for different user groups, such as using more impactful advertisements for young users, more professional advertisements for business users, and more heartwarming advertisements for maternal and child users, so as to achieve more accurate advertising delivery and improve the ROI (return on investment) of advertising.

[0117] The present invention greatly reduces the manual operations of designers in e-commerce advertising production through intelligent template matching, automated design optimization and workflow automation. Traditional advertising design usually requires designers to manually select appropriate templates, adjust the size of product images, set layout and color matching, and the whole process is time-consuming and cumbersome. The present invention uses AI-driven intelligent template matching, which can automatically recommend the best template based on product category, brand style and target users, and optimize product images, texts and interactive buttons in combination with automated typesetting technology. In this way, designers can make fine adjustments based on what the system provides, instead of designing from scratch, thereby significantly shortening the advertising production cycle, improving design efficiency, and enabling designers to focus more on creative work.

[0118] The present invention ensures that advertisements have stronger visual appeal through computer vision analysis, intelligent layout optimization and color recommendation engine. Traditional advertising design often relies on the personal experience of designers, which may lead to uneven aesthetics and information transmission effects of advertisements. The present invention uses AI visual recognition technology to automatically analyze the focus, color, and light and shadow effects of product images, and optimizes the layout structure of advertisements in combination with user browsing habits to ensure that products, texts, and interactive elements are arranged according to the best visual path. In addition, the present invention can automatically match the optimal color scheme to ensure that the overall color matching of the advertisement is in line with the brand style and the aesthetic preferences of the target users. Through these technologies, the present invention can effectively improve the visual quality of advertisements, making them more attractive, thereby increasing users' click-through rate and desire to buy.

[0119] The present invention ensures that advertisements can be applied to multiple e-commerce platforms such as Taobao, JD.com, and Pinduoduo through automatic size adjustment and cross-platform adaptation technology, reducing the workload of designers to repeatedly adjust advertisements between different platforms. In traditional methods, designers need to manually adjust the size, layout, and format of advertisements for different e-commerce platforms, especially when a brand needs to advertise on multiple platforms, the adjustment cost is very high. The present invention adopts an intelligent size adaptation algorithm, which can automatically adjust the aspect ratio, text layout, image ratio and other parameters of advertisements according to the standards of different platforms to ensure that the display effect of advertisements on various platforms remains consistent. In addition, the present invention can also automatically adjust the font size and interactive button position according to the display characteristics of different terminals (PC, mobile, APP) to adapt to the user experience of different devices, thereby improving the adaptability and display effect of advertisements.

[0120] The present invention can optimize the advertising design in real time through A / B testing, user behavior monitoring and machine learning analysis, so that it can continuously adapt to market demand and improve the effect of advertising. Traditional advertising optimization often relies on the experience of designers or the feedback of marketing teams, with a long adjustment cycle and it is difficult to ensure the correct optimization direction. The present invention adopts an A / B testing mechanism, which can monitor data such as click-through rate, conversion rate, user stay time, etc. in real time after advertising is released, and analyze the performance of different advertising versions through machine learning models. For example, the system can automatically test the impact of different color schemes, fonts, and button sizes on user behavior, and automatically select the design with the best performance. In addition, the system can also predict the optimal design mode of advertising based on historical data, making advertising delivery more accurate and improving the ROI (return on investment) of advertising.

[0121] The present invention adopts user portrait analysis and personalized recommendation engine, which can automatically adjust the design of advertisements for different user groups, thereby improving the personalization of advertisements and improving conversion rates. In traditional e-commerce advertising design, the advertising content seen by all users is basically the same, and it is impossible to make personalized adjustments based on the preferences of different users. The present invention uses machine learning to analyze users' browsing history, purchasing behavior, and interest preferences, and can match the most suitable advertising design for different user groups. For example, for young users, the system may recommend more lively and bright colors and dynamic layouts; for high-end business people, the system may choose a more simple and atmospheric style. In addition, the present invention can also automatically adjust the advertising style in combination with seasons, festivals, and hot events, so that the advertisements are more in line with current market demand, thereby improving users' click-through rates and purchase conversion rates.

[0122] The present invention uses intelligent advertising generation and automated optimization to enable e-commerce operators to quickly generate high-quality advertising designs even if they do not have professional design skills, thereby reducing the threshold and cost of advertising production. In traditional e-commerce marketing, merchants often need to hire professional graphic designers or use complex design software (such as Photoshop, Illustrator) to produce advertisements, which not only increases operating costs, but also reduces the production efficiency of advertisements. The present invention uses intelligent and automated design tools, so that ordinary operators only need to input product information, and the system can automatically generate high-quality advertising images that meet the brand style, and optimize and adapt to multiple platforms. In this way, merchants can reduce their dependence on professional design teams, reduce labor costs, and greatly improve the speed and quality of advertising production, making e-commerce operations more efficient and intelligent.

[0123] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. E-commerce design automated workflow optimization system, characterized by: Applied to the design of product display images and advertising images on e-commerce platforms, combining artificial intelligence technology and automated workflow to optimize the e-commerce design process, including intelligent template recommendation module, adaptive design optimization module, intelligent typesetting and custom adjustment module, cross-platform compatibility and adaptation module, data-driven optimization module, and A / B testing and intelligent evolution module; Intelligent template recommendation module obtains basic information of products and automatically recommends matching design templates based on the basic information using artificial intelligence models. Design template recommendations are based on e-commerce platform standards, brand historical data, and user behavior analysis to improve the pertinence and consistency of designs. The adaptive design optimization module automatically recommends the best design elements based on the recommended template, product characteristics, target audience and brand style, and combines deep learning algorithms to make adaptive adjustments to ensure optimal visual effects; Intelligent layout and custom adjustment module, which uses automated processing technology to intelligently adjust advertising images and supports user-defined adjustments to improve image quality and design flexibility; Cross-platform compatibility and adaptation module, which identifies the requirements of e-commerce platforms, automatically adjusts the output format of design works, ensures compatibility and consistency on different e-commerce platforms, and automatically adapts to the display requirements of mobile and PC terminals; The data-driven optimization module dynamically optimizes the visual elements of the ad images based on the data from the e-commerce platform, automatically adjusts the design plan based on data analysis, and continuously improves the ad performance; The A / B testing and intelligent evolution module automatically generates multiple versions of advertising images and conducts A / B testing to compare the advertising effects of different versions, select the best version and optimize future design strategies to achieve continuous evolution of design solutions.

2. The e-commerce design automation workflow optimization system according to claim 1 is characterized in that: The product information analysis step further includes product image analysis, specifically: Obtain the original image data of the product to be designed, and use computer vision technology to analyze the image to extract the product's outline, color distribution, texture characteristics and lighting conditions; Classify products through deep learning models, identify their categories, and assign different design weights based on category labels; Combined with the brand style library, it analyzes whether the product's design style is consistent with the brand's existing visual image. If not, it generates optimization suggestions for designers' reference. At the same time, based on the historical interaction data of the target audience, it optimizes the product's display angle, background style and recommended color scheme to maximize the product's visual appeal.

3. The e-commerce design automation workflow optimization system according to claim 1, characterized in that: The intelligent template matching step further includes template adaptive adjustment, specifically: After selecting the matching template, the adaptive image transformation algorithm is used to automatically adjust the aspect ratio of the template according to the requirements of the e-commerce platform, and adjust the image cropping area while ensuring visual balance; Based on product feature parameters, dynamically adjust the layout of the template, including text position, product image size, blank area, and auxiliary design elements, to meet the display requirements of different products; When a product has multiple display angles or variations, the template is automatically adjusted to ensure that all variations can be reasonably presented in the design; The arrangement order of template elements is optimized based on historical click-through rate and conversion rate data, so that users can quickly focus on the key features of the product when browsing the advertisement, thereby improving the attractiveness and effectiveness of the advertisement.

4. The e-commerce design automation workflow optimization system according to claim 1, characterized in that: The automated design optimization step further includes dynamic typesetting based on artificial intelligence, specifically: A neural network model is used to optimize the layout of the ad image to maximize user attention. First, the design draft is analyzed through a convolutional neural network to identify the main visual focus and calculate the weight of each design element to ensure that the core product area maintains optimal visibility. Then, a reinforcement learning algorithm is used to adjust the text, buttons, and background elements of the ad to ensure the clarity of the information conveyed and prevent design elements from overlapping or causing visual interference; In the final optimization stage, simulation tests are conducted on the advertising images to simulate different user perspectives, analyze the user's attention path, and fine-tune the position of image elements on this basis to ensure that the advertising information can be delivered to the user in the shortest time, thereby improving the click-through rate and conversion effect of the advertisement.

5. The e-commerce design automation workflow optimization system according to claim 1, characterized in that: The intelligent design recommendation step further includes personalized design recommendations based on user preferences, specifically: Combine users’ browsing and purchasing data to analyze their preference patterns and generate personalized design solutions based on historical data; first , using clustering algorithms to divide users into different consumer groups and assign corresponding design style labels to different groups; Then, based on the personalized recommendation model, corresponding fonts, color schemes and advertising layouts are recommended for different user groups; Finally, combined with real-time data analysis, the effectiveness of different recommendation schemes is monitored, and the recommendation weights are automatically adjusted to ensure that each advertising delivery matches the most appropriate target users.

6. The e-commerce design automation workflow optimization system according to claim 1, characterized in that: The multi-platform adaptation step further includes adaptive layout optimization, specifically: The mathematical modeling method is used to standardize the size requirements of different e-commerce platforms and calculate the optimal size parameters of the advertising images. The optimization formula is as follows: Assume that the set of e-commerce platforms is P = {p i }={p1,p2,…,p n }, each platform p i With size requirements (w i ,h i ), where: w i Represents platform p i Required ad image width, h i Represents platform p i The required ad image height, the optimal adaptation size calculation formula is: W opt =max(w1,w2,…,w n ),H opt =max(h1,h2,…,h n ) Where W opt and H opt are the optimal fitting width and height respectively; Set the adjustment coefficient α for different platforms i and β i , so that the final image adapts to the display ratio of a specific platform, the adjusted size is calculated as follows: W i =W opt ·a i ,H i =H opt ·b i Among them, α i is the width scaling factor of the i-th e-commerce platform, β i is the height scaling factor of the ith e-commerce platform, W i It represents the optimized advertising image on the i-th e-commerce platform p i The final width on the i It represents the optimized advertising image on the i-th e-commerce platform p i The final height on The gradient descent optimization algorithm is used to adjust the image layout so that the advertising elements are evenly distributed and meet the platform specifications. The optimization objective function is as follows: Where: w i represents the i-th e-commerce platform p i Required standard width of the ad image, h i represents the i-th e-commerce platform p i The standard height required for advertising graphics.

7. The e-commerce design automation workflow optimization system according to claim 1, characterized in that: The real-time optimization and feedback adjustment step further includes data-driven optimization based on A / B testing, specifically: Set the advertising design solution set A = {a k }={a1,a2,…,a n }, where n represents the total number of advertising design schemes, and each advertising design scheme a k With an initial click rate of C k and the initial conversion rate T k , conduct A / B testing and record the number of clicks c of each advertising design on user group U k and number of purchases t k , calculate the new click rate and conversion rate, the calculation expression is: Where: C′ k Design a plan for advertising k The actual click rate in the test group, T′ k Design a plan for advertising k The actual conversion rate, that is, the proportion of completed purchases after clicking; The optimization objective function is set so that the selected advertising design maximizes the click-through rate and conversion rate while ensuring design consistency. The expression of the optimization objective function is: Among them, λ1 and λ2 are weight coefficients, which are adjusted according to marketing needs; Through multiple rounds of iterations, the design elements are automatically adjusted and eventually converged to the optimal advertising design plan, thereby improving the effectiveness of advertising.

8. The e-commerce design automation workflow optimization system according to claim 1, characterized in that: The intelligent design recommendation step further includes context-aware optimization, specifically: When recommending design plans, it automatically adjusts the design style based on external contextual information to make the advertising content more targeted and increase users' willingness to buy.

Images