Advertisement creative generation method based on AI

Abstract

The invention belongs to the technical field of advertisement creativity, and discloses an AI-based advertisement creativity generation method. The method specifically comprises the following steps: a, consumer data acquisition and portrait construction; b, content generation model construction; c, style migration model construction; d, multi-modal content generation model construction; according to the advertisement creative generation method based on the AI, the time period of creative output is remarkably shortened, leap improvement of the working efficiency is achieved, and stable improvement of the creative quality is ensured by continuously optimizing the trained model; the revolution also releases human resources, so that the system can concentrate on higher-level innovation and strategy planning, and meanwhile, the gap of the creative talent market is effectively made up.

Description

Technical Field

[0001] The present invention belongs to the technical field of advertising creativity, and specifically is a method for generating advertising creativity based on AI. Background Art

[0002] At present, the mainstream method of generating advertising creative ideas on the market is the AI ​​advertising template generation method that is not based on the NPL neural network large model;

[0003] The specific method is as follows: by providing several sets of advertising templates, and splitting the advertising templates into different types of paragraphs, different paragraphs are matched and assembled from the material library according to the delivery content to form advertisements, where the picture materials come from the product library, and according to a certain algorithm, one is extracted from each of the appearance, body details, seats, and central control instruments, and assembled into three pictures. The title is assembled according to a certain algorithm based on the current existing activities of the delivery model and the different types of descriptive words entered in advance for the model to generate advertising copy, and the advertising copy and advertising pictures are assembled into an advertising creative and delivered to the C-end;

[0004] Its defects are: the high degree of templateization leads to limited innovation, and the material preparation is cumbersome and lacks customization; the system generation capability is limited to preset templates and lacks adaptability to new advertising formats; at the same time, the system learning ability is limited, and it is difficult to optimize through feedback, and it is impossible to conduct deep semantic understanding and multimodal information integration, which increases the difficulty of controlling the creative direction.

[0005] To sum up, traditional Internet advertising has a mature operating model, but is limited by its innovation and precise targeting capabilities, resulting in many limitations in the advertising creative generation method, which cannot be better adapted to current development needs, and therefore needs to be improved. Summary of the invention

[0006] The purpose of the present invention is to provide an AI-based advertising creative generation method to solve the problems raised in the above-mentioned background technology.

[0007] In order to achieve the above-mentioned purpose, the present invention provides the following technical solution: a method for generating advertising creativity based on AI, the method specifically comprises the following steps:

[0008] a. Consumer data collection and portrait construction: Collect user attribute data, user behavior data, and user content data of the target audience through market research, web tracking, and APP data analysis channels, and use data analysis technology to construct consumer portraits that include user attribute labels, behavior patterns, interest preferences, and other dimensions;

[0009] b. Content generation model construction: Use Seq2Seq and BERT natural language processing (NLP) models to train a text generation model that can receive user features as input and output personalized creative text that meets user preferences, such as ad titles and descriptions;

[0010] c. Style transfer model construction: Collect brand visual assets, such as logos, color schemes, design elements, etc., and use CycleGAN deep learning technology to train a style transfer model that can automatically apply the brand visual style to the generated images or videos;

[0011] d. Construction of multimodal content generation model: Combine the text generation model with the image / video generation model to build a multimodal content generation model. Through entity recognition and semantic matching technology, deep integration of text and image / video is achieved to automatically generate personalized image and video advertising ideas.

[0012] e. Rich media advertising production: Based on the generated creative text and images / videos, complete rich media advertising is produced by automated or manual methods;

[0013] f. User segmentation and precision delivery: Based on user portrait data, use algorithms such as cluster analysis to segment the target audience into different groups, and deliver customized advertising ideas to different groups;

[0014] g. Effect analysis and feedback optimization: Collect effect data after advertising, such as click-through rate, conversion rate, etc., evaluate the effect of advertising creativity through data analysis, and feed the effect data back to the generation model for model iteration and optimization;

[0015] h. Review and adjustment of human-machine collaborative mode: After AI generates advertising creatives, a manual review and adjustment process is introduced. Creative or operational personnel review the AI-generated creatives based on actual needs and make adjustments when necessary.

[0016] Preferably, the consumer data collection step includes obtaining user data through various means such as market research, web page tracking, and APP data analysis, and constructing a user portrait including user attribute labels, behavioral data, and interest and hobby dimensions.

[0017] Preferably, in the content generation model construction step, a BERT pre-trained language model is used to fine-tune the model for different industries and categories to generate creative texts that meet specific needs.

[0018] Preferably, in the style transfer model construction step, the CycleGAN deep learning technology is used to transfer the brand visual style to the generated image or video to ensure that the advertising creativity is consistent with the brand image.

[0019] Preferably, in the step of constructing the multimodal content generation model, entity recognition and semantic matching technology are used to achieve deep fusion of text and image / video to generate highly personalized advertising creativity.

[0020] Preferably, in the rich media advertisement production step, automation and manual processing are combined to ensure that the advertisement content is consistent with the brand tone and can attract the target audience.

[0021] Preferably, in the user grouping and precision delivery step, the target audience is segmented into multiple groups through algorithms such as cluster analysis, and customized advertising ideas are delivered to different groups to achieve precision marketing.

[0022] Preferably, in the effect analysis and feedback optimization step, a monitoring and labeling system for click and conversion data is established, and the data analysis results are fed back to the generation model to achieve continuous optimization of the model.

[0023] Preferably, the human-machine collaborative mode review and adjustment is performed by creative or operational personnel after AI generates advertising creativity to ensure the accuracy and attractiveness of the advertising creativity.

[0024] The beneficial effects of the present invention are as follows:

[0025] The present invention not only significantly shortens the time cycle of creative output and achieves a leap in work efficiency by adopting an AI-based advertising creative generation method, but also ensures the steady improvement of creative quality by continuously optimizing the trained model. The precise push of personalized creativity, based on in-depth insights into user data, greatly enhances the satisfaction of user experience. At the same time, the automated generation process greatly reduces the cost of traditional manual creation, making the output of advertising creativity more economical and efficient. More importantly, the precise targeting and personalized strategies assisted by AI directly promote the improvement of advertising conversion rate and optimize advertising effect. In addition, the continuous iterative optimization capability of the AI ​​model ensures that creative design always keeps pace with the times and continues to improve. This change also releases human resources, enabling them to focus on higher-level innovation and strategic planning, while effectively filling the gap in the creative talent market. In the end, the continuous output of high-quality, personalized advertising creativity has injected new vitality into the brand and significantly enhanced the brand's market influence and competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 The figure is a simplified flow chart of the method for generating advertising creativity of the present invention. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] like Figure 1 As shown, an embodiment of the present invention provides a method for generating advertising creativity based on AI, and the method specifically comprises the following steps:

[0029] a. Consumer data collection and portrait construction: Collect user attribute data, user behavior data, and user content data of the target audience through market research, web tracking, and APP data analysis channels, and use data analysis technology to construct consumer portraits that include user attribute labels, behavior patterns, interest preferences, and other dimensions;

[0030] Comprehensively collect diverse data of target audiences through a wide range of data collection channels, including but not limited to online surveys, social media analysis, shopping behavior tracking, etc. Utilize advanced data analysis technologies, such as big data analysis, machine learning algorithms, etc., to build refined consumer portraits that not only contain basic demographic information, but also provide in-depth insights into users' psychological characteristics, consumption habits, and potential needs.

[0031] b. Content generation model construction: Use Seq2Seq and BERT natural language processing (NLP) models to train a text generation model that can receive user features as input and output personalized creative text that meets user preferences, such as ad titles and descriptions;

[0032] Using advanced NLP technologies such as BERT and its variants, these models are pre-trained on large-scale corpora and have strong language understanding and generation capabilities. For specific industries or brands, by fine-tuning model parameters, it can generate creative texts that are more in line with the interests and preferences of the target audience, including but not limited to advertising titles, text descriptions and slogans.

[0033] c. Style transfer model construction: Collect brand visual assets, such as logos, color schemes, design elements, etc., and use CycleGAN deep learning technology to train a style transfer model that can automatically apply the brand visual style to the generated images or videos;

[0034] Generative adversarial networks (GANs) in deep learning, especially models such as CycleGAN, can be used to accurately transfer brand visual styles. By inputting brand-specific visual elements such as color matching and pattern design, the model can automatically learn and apply them to the generated images or videos, ensuring that the advertising creativity is highly consistent with the brand image in terms of vision, enhancing brand recognition and memory points.

[0035] d. Construction of multimodal content generation model: Combine the text generation model with the image / video generation model to build a multimodal content generation model. Through entity recognition and semantic matching technology, deep integration of text and image / video is achieved to automatically generate personalized image and video advertising ideas.

[0036] It innovatively combines text and image / video generation technology. Through cross-modal learning, the model can understand the intrinsic connection between text and visual content and achieve seamless connection between the two. This multimodal fusion not only enriches the expression of advertising creativity, but also enhances the appeal and communication effect of creativity.

[0037] e. Rich media advertising production: Based on the generated creative text and images / videos, complete rich media advertising is produced by automated or manual methods;

[0038] It combines the advantages of automated generation and manual editing. Automated tools can quickly generate preliminary advertising materials, while manual editing can make fine adjustments based on this to ensure the accuracy and creativity of the advertising content. This human-machine collaborative model greatly improves the efficiency and quality of advertising production.

[0039] f. User segmentation and precision delivery: Based on user portrait data, use algorithms such as cluster analysis to segment the target audience into different groups, and deliver customized advertising ideas to different groups;

[0040] Advanced user grouping algorithms are used, such as K-means clustering, hierarchical clustering, etc. These algorithms can segment the target audience into different groups with similar characteristics based on user portrait data. Differentiated advertising strategies are formulated for different groups to achieve accurate advertising delivery and personalized recommendations.

[0041] g. Effect analysis and feedback optimization: Collect effect data after advertising, such as click-through rate, conversion rate, etc., evaluate the effect of advertising creativity through data analysis, and feed the effect data back to the generation model for model iteration and optimization;

[0042] A comprehensive advertising effect evaluation system has been established. By collecting various data indicators after advertising, such as click-through rate, conversion rate, user feedback, etc., the effect of advertising creativity is quantitatively evaluated. At the same time, the evaluation results are fed back to the generation model in a timely manner, and the model performance is continuously improved through iterative optimization algorithms to enhance the generation quality and effect of advertising creativity.

[0043] h. Review and adjustment of human-machine collaborative mode: After AI generates advertising creatives, a manual review and adjustment process is introduced. Creative or operational personnel review the AI-generated creatives based on actual needs and make adjustments when necessary.

[0044] The review and adjustment of the human-machine collaborative model plays an important role in the process of generating advertising creatives. After AI generates preliminary creatives, professional creative or operational personnel will conduct manual review and adjustments. This human-machine collaborative approach not only takes advantage of AI's advantages in data processing and creative generation, but also ensures the creativity and compliance of advertising creatives. By continuously optimizing the human-machine collaborative mechanism, the intelligent and efficient generation of advertising creatives can be achieved.

[0045] Among them, the consumer data collection step includes obtaining user data through various means such as market research, web page tracking, and APP data analysis, and constructing a user portrait that includes user attribute labels, behavioral data, and interest and hobby dimensions.

[0046] Among them, in the content generation model construction step, the BERT pre-trained language model is used to fine-tune the model for different industries and categories to generate creative texts that meet specific needs.

[0047] Among them, in the style transfer model construction step, the CycleGAN deep learning technology is used to transfer the brand visual style to the generated image or video to ensure that the advertising creativity is consistent with the brand image.

[0048] Among them, in the step of constructing the multimodal content generation model, through entity recognition and semantic matching technology, deep fusion of text and image / video is achieved to generate highly personalized advertising creativity.

[0049] Among them, in the rich media advertising production steps, automation and manual processing are combined to ensure that the advertising content is both in line with the brand tone and can attract the target audience.

[0050] Among them, in the user segmentation and precision delivery steps, the target audience is segmented into multiple groups through algorithms such as cluster analysis, and customized advertising ideas are delivered to different groups to achieve precision marketing.

[0051] Among them, in the effect analysis and feedback optimization step, a monitoring and labeling system for click and conversion data is established, and the data analysis results are fed back to the generation model to achieve continuous optimization of the model.

[0052] Among them, the human-machine collaborative mode review and adjustment is carried out by creative or operational personnel after AI generates advertising creativity to ensure the accuracy and attractiveness of the advertising creativity.

[0053] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0054] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for generating advertising creativity based on AI, characterized by: The method specifically comprises the following steps: a. Consumer data collection and portrait construction: Collect user attribute data, user behavior data, and user content data of the target audience through market research, web tracking, and APP data analysis channels, and use data analysis technology to construct consumer portraits that include user attribute labels, behavior patterns, and interest preference dimensions; b. Content generation model construction: Use Seq2Seq and BERT natural language processing (NLP) models to train a text generation model that can receive user features as input and output personalized creative text that meets user preferences; c. Style transfer model construction: Collect brand visual assets and use CycleGAN deep learning technology to train a style transfer model that can automatically apply the brand visual style to generated images or videos; d. Construction of multimodal content generation model: Combine the text generation model with the image / video generation model to build a multimodal content generation model. Through entity recognition and semantic matching technology, deep integration of text and image / video is achieved to automatically generate personalized image and video advertising ideas. e. Rich media advertising production: Based on the generated creative text and images / videos, complete rich media advertising is produced by automated or manual methods; f. User segmentation and precision delivery: Based on user portrait data, cluster analysis algorithms are used to segment the target audience into different groups, and customized advertising ideas are delivered to different groups; g. Effect analysis and feedback optimization: Collect effect data after advertising is released, evaluate the effect of advertising creativity through data analysis, and feed the effect data back to the generation model for iterative model optimization; h. Review and adjustment of human-machine collaborative mode: After AI generates advertising creatives, a manual review and adjustment process is introduced. Creative or operational personnel review the AI-generated creatives based on actual needs and make adjustments when necessary.

2. The method for generating advertising creative ideas based on AI according to claim 1, characterized in that: The consumer data collection step includes obtaining user data through various means such as market research, web page tracking, and APP data analysis, and constructing a user portrait that includes user attribute labels, behavioral data, and interest and hobby dimensions.

3. The method for generating advertising creative ideas based on AI according to claim 1, characterized in that: In the content generation model construction step, the BERT pre-trained language model is used to fine-tune the model for different industries and categories to generate creative texts that meet specific needs.

4. The method for generating advertising creative ideas based on AI according to claim 1, characterized in that: In the style transfer model construction step, the CycleGAN deep learning technology is used to transfer the brand visual style to the generated image or video to ensure that the advertising creativity is consistent with the brand image.

5. The method for generating advertising creative ideas based on AI according to claim 1, characterized in that: In the multimodal content generation model construction step, entity recognition and semantic matching technology are used to achieve deep integration of text and image / video to generate highly personalized advertising creativity.

6. The method for generating advertising creative ideas based on AI according to claim 1, characterized in that: In the rich media advertising production steps, automation and manual processing are combined to ensure that the advertising content is consistent with the brand tone and can attract the target audience.

7. The method for generating advertising creative ideas based on AI according to claim 1, characterized in that: In the user grouping and precision delivery steps, the target audience is segmented into multiple groups through a cluster analysis algorithm, and customized advertising ideas are delivered to different groups to achieve precision marketing.

8. The method for generating advertising creative ideas based on AI according to claim 1, characterized in that: In the effect analysis and feedback optimization step, a monitoring and labeling system for click and conversion data is established, and the data analysis results are fed back to the generation model to achieve continuous optimization of the model.

9. The method for generating advertising creativity based on AI according to claim 1, characterized in that: The human-machine collaborative mode review and adjustment is carried out by creative or operational personnel after AI generates advertising ideas to ensure the accuracy and attractiveness of the advertising ideas.

Images