Product detail page content generation system based on AI empirical model

Abstract

The invention relates to the technical field of artificial intelligence and electronic commerce, and particularly discloses a product detail page content generation system based on an AI empirical model, which comprises a value anchoring engine, a framework arrangement center, an intelligent knowledge base, a creative generation workshop and a data iteration engine, identifying, weighting and outputting product core attributes and emotional marketing selling points; the framework arrangement center matches the detail page template according to the product classification and analyzes the structured framework; the intelligent knowledge base stores resources such as marketing models and templates; the creative generation workshop gathers multi-source data, generates copywriting and material schemes, and outputs a first draft of a detail page; a data iteration engine collects user behavior data, analyzes effects, generates optimization suggestions, and reversely optimizes the system. According to the product detail page content generation system based on the AI empirical model, efficient generation, accurate matching and dynamic optimization of the detail page content are achieved, the conversion rate is remarkably increased, and the marketing cost is reduced.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence and e-commerce technology, and in particular to a product detail page content generation system based on AI experience models. Background Technology

[0002] As the core carrier connecting brands and users in e-commerce scenarios, the product detail page still relies on the traditional model of manual content generation, which has long been a bottleneck in efficiency and effectiveness in multiple dimensions.

[0003] First, from extracting product selling points and designing page frameworks to matching and arranging materials, it requires collaboration among multiple roles in marketing, design, and operations. The production of a single product detail page often takes 3-7 days. In the event of e-commerce promotions or a large number of new product launches, manual production capacity is insufficient to meet the demand for batch generation, which can easily lead to missing the marketing window.

[0004] Secondly, the discovery of selling points relies heavily on the subjective judgment of operations personnel, making it difficult to systematically integrate multi-source data such as product technical parameters, user feedback, and differences from competitors, resulting in inaccurate delivery of core value. At the same time, high-quality marketing experience (such as high-conversion copywriting structure and scenario-based expression logic) cannot be reused at scale, and the quality of detail pages varies greatly across different products and batches, highlighting the problem of homogenization.

[0005] Furthermore, traditional manual methods struggle to deeply integrate target user profiles with the characteristics of each platform for customized content creation, often resulting in a disconnect between selling points and user needs. For example, appliance detail pages targeting young renters still use lengthy technical parameter descriptions aimed at family users, leading to short user dwell time and low conversion rates.

[0006] Furthermore, after the details page is launched, it is difficult for humans to track user behavior data in real time and quickly locate inefficient content elements, resulting in the details page remaining in a "static" state for a long time and failing to adapt to changes in market demand.

[0007] Finally, in the traditional model, marketing resources such as material libraries and high-quality content features of influencers are stored in a scattered manner, which cannot be efficiently linked with the product detail page generation process. For example, it is not possible to quickly reuse the scenario-based expressions in popular notes from social media influencers, nor is it easy to intelligently filter resources based on the matching degree between materials and selling points. This further exacerbates the inefficiency and homogenization of content generation, ultimately leading to high brand marketing costs, while the conversion effect of the product detail page has always been difficult to break through the bottleneck. Summary of the Invention

[0008] The purpose of this invention is to provide a product detail page content generation system based on an AI experience model, which achieves efficient generation, accurate matching and dynamic optimization of product detail page content, reduces overall marketing costs while ensuring content compliance and the accumulation and reuse of marketing experience.

[0009] To achieve the above objectives, this invention provides a product details page content generation system based on an AI experience model, including a value anchoring engine, a framework orchestration hub, an intelligent knowledge base, a creative generation workshop, and a data iteration engine; The value anchoring engine receives raw product data and identifies and outputs core product attributes and emotional marketing selling points based on a built-in marketing experience model. The framework orchestration hub is used to match and call the corresponding details page template from the intelligent knowledge base based on the product's classification information, and parse the structured framework of the output content; An intelligent knowledge base is used to store marketing experience models, product detail page templates, multimedia materials, compliance rules, and expert knowledge. The Creative Generation Workshop is used to gather marketing resources from the value anchoring engine, the content framework of the framework arrangement center, and the intelligent knowledge base. It generates creative copy and material matching solutions by integrating generative AI models and outputs the first draft of the product detail page. The data iteration engine is used to collect user behavior data after the details page goes live, analyze and quantify the content effect to generate optimization suggestions, and then feed these optimization suggestions back to the value anchoring engine and the creative generation workshop.

[0010] Preferred value anchoring engines include: The raw data receiving unit is used to receive raw product data, which includes objective product attribute data and marketing value-related data. The feature recognition unit receives data from the raw data receiving unit and extracts the core attributes and emotional marketing selling points of the product. When extracting the core attributes of the product, natural language processing (NLP) technology is used to process the raw text data, and entity recognition is performed through a BERT pre-trained model to extract the physical attributes, technical attributes, and performance attributes of the product. Computer vision technology is used to process the raw image data, and morphological attributes and material texture attributes are extracted through a ResNet50 model. When extracting emotional marketing selling points, the feature recognition unit is based on the core attributes of the product and combines the historical case mapping rules of the marketing experience model library in the intelligent knowledge base to construct the relationship between core attributes, user needs and emotional selling points. Then, the TF-IDF algorithm is used to supplement emotional expressions from the marketing value association data, and the positive scenario-based expressions are filtered by the TextCNN model. The weight calculation unit calculates the weights of the product's core attributes and emotional marketing selling points based on multiple predefined influencing factors. The weight calculation of emotional marketing selling points is affected by the weights of their associated product core attributes. The Selling Point Output Unit is used to transmit the combination of the product's core attributes and emotional marketing selling points with the highest weight to the Creative Generation Workshop, and simultaneously output the priority marks of the top 3 emotional marketing selling points with the highest weight.

[0011] Preferably, the framework orchestration center includes: The classification interaction unit is used to receive product classification information, and transmit the product classification information to the template matching unit after standardizing and encoding the information. The template matching unit, based on the encoded product classification information, uses a cosine similarity algorithm to calculate the matching degree between the classification and the template label. If the matching degree is... 85% are selected as candidate templates; if multiple candidate templates exist, their conversion rate coefficients from historical applications in the details page template library are considered. Select the optimal template and transmit the selected optimal template data to the structural analysis unit; The structure parsing unit is used to parse the structured framework of the template, extract the framework data in the template, including the arrangement order, content proportion and presentation logic of the title area, introduction area, parameter table area, selling point list area, usage scenario area and user review area, and transmit the framework data to the creative generation workshop.

[0012] Preferably, the intelligent knowledge base includes: The Marketing Experience Model Library stores standardized model clusters trained using machine learning methods. The model training data includes high-conversion e-commerce product detail page copy, popular social media influencer notes, and brand marketing materials. The Marketing Experience Model Library provides model invocation services to the Value Anchoring Engine through an API interface. The product detail page template library stores structured templates categorized by product type. Each template is associated with historical conversion data, applicable scenario tags, and update time. The product detail page template library provides template data to the framework orchestration center through a template call interface. The multimedia resource library stores product images, video clips, design elements, resource associations, product categories, and matching tags for selling points; the multimedia resource library provides resource data to the creative generation workshop through a resource matching interface; The compliance rule library stores a list of prohibited terms under the Advertising Law, platform guidelines, and industry compliance standards; it also provides compliance data to the Creative Generation Workshop through a compliance verification interface. The Influencer Knowledge Base stores data on influencers across the internet, including content tags and fan profile tags. This data includes basic influencer information, content style characteristics, and historical collaboration results. The Influencer Knowledge Base provides high-quality content feature data to the Creative Generation Workshop through an influencer feature interface.

[0013] Preferred creative generation workshops include: The data aggregation unit is used to receive the product's core attributes and emotional selling points combination from the value anchoring engine, the structured framework of the framework orchestration center, multimedia materials from the intelligent knowledge base, and marketing model parameters through an asynchronous data interface. It performs data format unification processing and transmits the processed data to the generative AI unit and the material matching unit. Generative AI Unit, used to integrate pre-trained generative AI models to generate the following content under the constraints of a marketing experience model: The title section uses a sentence structure that combines pain points and selling points; the lead section uses a logic that introduces scenarios and adds product value; the selling point list section uses a description that combines features and benefits; the usage scenario section generates multiple typical user scenario copy; and the generated typical user scenario copy data is transmitted to the material matching unit. The material matching unit is used to match materials to each content section based on the matching degree between selling points and material tags, and output layout suggestions, and transmit the copy, materials, and layout suggestions to the draft output unit; The initial draft output unit is used to integrate data according to a structured framework to form a complete initial draft of the product details page. The initial draft is transmitted to the data iteration engine and compliance verification module through the data interface, and the initial draft is stored in the system's historical database.

[0014] Preferably, the data iteration engine includes: The behavior collection unit is used to collect user behavior data after the details page goes live through page tracking technology, including dwell time, number of clicks in each section, conversion path data and user evaluation data, and transmit the collected data to the performance analysis unit in real time; The performance analysis unit uses statistical algorithms and machine learning models to quantify content effectiveness, first calculating the attention level of different sections. Conversion efficiency and evaluation of satisfaction The three indicators are then analyzed using the CART decision tree model to determine the relationship between the indicators and content elements, identify inefficient content elements, and then transmit the analysis results to the optimization feedback unit. The optimized feedback unit generates optimization suggestions by module, outputs suggestions for adjusting selling point weights to the value anchoring engine, and outputs suggestions for copywriting optimization and material replacement to the creative generation workshop.

[0015] Preferably, it also includes a compliance adaptation unit, which is integrated into the creative generation workshop. This unit is used to receive the initial draft of the product details page from the initial draft output unit, call the rule data of the intelligent knowledge base compliance rule library, and perform compliance verification on the initial draft of the product details page.

[0016] Preferably, the specific process for compliance verification of the initial draft of the product details page is as follows: After removing words that violate advertising laws and correcting expressions that do not conform to platform guidelines, if the proportion of non-compliant content is... If the percentage is 10%, return to the Creative Creation Workshop to regenerate the product details page; The product detail page format was adjusted based on the characteristics of the target platform, including vertical images and dual-format layouts. The product detail page was then transmitted to an external publishing system, and the compliance verification results were fed back to the data iteration engine for optimization and analysis.

[0017] Therefore, the product detail page content generation system based on the AI ​​experience model described above has the following beneficial effects: (1) This invention automatically identifies the core attributes and emotional selling points of a product through a value anchoring engine, and combines the template matching of the framework arrangement center to significantly shorten the production cycle of the details page, meet the batch generation needs of e-commerce promotions and other scenarios, and improve production efficiency.

[0018] (2) Based on the marketing experience model and historical data in the intelligent knowledge base, the present invention systematically and deeply explores the relationship between user needs and product selling points, generates personalized and scenario-based content, avoids homogenization, and significantly improves user dwell time and conversion rate.

[0019] (3) This invention collects user behavior data in real time based on a data iteration engine, analyzes the content effect through a machine learning model, and optimizes the selling point weight and copywriting materials in reverse, so that the details page always adapts to market changes and forms a closed-loop optimization; the intelligent knowledge base continuously accumulates high-conversion templates, influencer content and multimedia materials, realizes the standardized reuse of marketing experience, reduces the dependence on human experience, and significantly reduces the overall marketing cost.

[0020] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0021] Figure 1 This is an overall flowchart of an embodiment of the product details page content generation system based on an AI experience model of the present invention; Figure 2 This is a schematic diagram illustrating the value anchoring engine weight calculation logic of an embodiment of the product details page content generation system based on an AI experience model of the present invention. Detailed Implementation

[0022] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0023] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0024] like Figure 1 As shown, the product detail page content generation system based on AI experience models includes a value anchoring engine, a framework orchestration hub, an intelligent knowledge base, a creative generation workshop, and a data iteration engine.

[0025] The value anchoring engine receives raw product data, identifies and outputs core product attributes and emotional marketing selling points based on a built-in marketing experience model, and includes the following components: The raw data receiving unit is used to receive raw product data, which includes objective product attribute data and marketing value-related data. The objective product attribute data includes product specifications, designer notes, and technical white papers, while the marketing value-related data includes an initial list of selling points, draft user feedback, and notes from past collaborators.

[0026] The feature recognition unit receives data from the raw data receiving unit and extracts the core attributes and emotional marketing selling points of the product. When extracting the core attributes of the product, natural language processing (NLP) technology is used to process the raw text data, and entity recognition is performed through a BERT pre-trained model to extract the physical attributes, technical attributes, and performance attributes of the product. Computer vision technology is used to process the raw image data, and morphological attributes and material texture attributes are extracted through a ResNet50 model.

[0027] When extracting emotional marketing selling points, the feature recognition unit is based on the core attributes of the product and combines the historical case mapping rules of the marketing experience model library in the intelligent knowledge base to construct the relationship between core attributes, user needs and emotional selling points. Then, the TF-IDF algorithm is used to supplement emotional expressions from the marketing value association data, and the positive scenario-based expressions are filtered by the TextCNN model.

[0028] like Figure 2As shown, the weight calculation unit calculates the weights of the product's core attributes and emotional marketing selling points based on multiple predefined influencing factors. The weight calculation of emotional marketing selling points is affected by the weights of their associated product core attributes.

[0029] Product core attribute weight The calculation formula is as follows: ; in, The proportion of attributes in user decision-making. To highlight the differentiation from competing products based on attributes, For attribute perceptibility. , and for , and Their respective weighting coefficients, among which, , , ,and .

[0030] The formula for calculating the weight of emotional marketing selling points is as follows: ; in, The frequency of appearance of selling points on high-conversion product detail pages, To boost user engagement as a selling point To adapt to a wide range of selling points and scenarios, , , They are respectively , and The weighting coefficients, , , ,and Simultaneously, verify the correlation between the product's core attributes and emotional marketing selling points. If the product's core attributes have a higher weight... A score of 0.3 corresponds to a 30% reduction in the weight of emotional marketing selling points.

[0031] The Selling Point Output Unit is used to transmit the combination of the product's core attributes and emotional marketing selling points with the highest weight to the Creative Generation Workshop, and simultaneously output the priority marks of the top 3 emotional marketing selling points with the highest weight.

[0032] The framework orchestration center of this invention is used to match and call the corresponding details page template from the intelligent knowledge base based on the product's classification information, and parse the structured framework of the output content; the framework orchestration center includes: The classification interaction unit is used to receive product classification information, and transmit the product classification information to the template matching unit after standardizing and encoding the product classification information.

[0033] The template matching unit is used to match templates from the details page template library of the intelligent knowledge base: based on the encoded product category information, it uses a cosine similarity algorithm to calculate the matching degree between the category and the template tag. If the matching degree is... 85% are selected as candidate templates; if multiple candidate templates exist, their conversion rate coefficients from historical applications in the details page template library are considered. Filter the best template. The optimal template data is transmitted to the structural analysis unit.

[0034] The structure parsing unit is used to parse the structured framework of the template, extract the framework data in the template, including the arrangement order, content proportion and presentation logic of the title area, introduction area, parameter table area, selling point list area, usage scenario area and user review area, and transmit the framework data to the creative generation workshop.

[0035] The intelligent knowledge base stores marketing experience models, product detail page templates, multimedia materials, compliance rules, and expert knowledge. The intelligent knowledge base includes: The Marketing Experience Model Library stores standardized model clusters trained using machine learning classification, regression, clustering, and deep learning algorithms. Model training data includes high-conversion e-commerce product detail page copy, popular social media influencer notes, and brand marketing materials. The Marketing Experience Model Library provides model invocation services to the Value Anchoring Engine via API interfaces.

[0036] The product detail page template library stores structured templates categorized by product. Each template is associated with historical conversion data, applicable scenario tags, and update time. The product detail page template library provides template data to the framework orchestration center through a template call interface.

[0037] The multimedia resource library stores product images, video clips, design elements, resource associations, product categories, and matching tags for selling points; the multimedia resource library provides resource data to the creative generation workshop through a resource matching interface.

[0038] The compliance rule library stores a list of prohibited terms under the Advertising Law, platform guidelines, and industry compliance standards. It provides compliance data to the Creative Generation Workshop through a compliance verification interface.

[0039] The Influencer Knowledge Base stores data on influencers across the internet, including content tags and fan profile tags. This data includes basic influencer information, content style characteristics, and historical collaboration results. The Influencer Knowledge Base provides high-quality content feature data to the Creative Generation Workshop through an influencer feature interface.

[0040] The Creative Generation Workshop is used to gather marketing resources from the value anchoring engine, the content framework of the framework orchestration center, and the intelligent knowledge base. By integrating generative AI models, it generates creative copy and material matching solutions and outputs the first draft of the product detail page.

[0041] The Creative Generation Workshop includes: The data aggregation unit receives the product's core attributes and emotional selling points from the value anchoring engine, the structured framework of the framework orchestration center, multimedia materials from the intelligent knowledge base, and marketing model parameters through an asynchronous data interface. It performs unified format processing on the data, converting text data to JSON format and image data to WebP format, and then transmits the processed data to the generative AI unit and the material matching unit.

[0042] Generative AI Unit, used to integrate pre-trained generative AI models to generate the following content under the constraints of a marketing experience model: The title section uses a sentence structure that combines pain points and selling points; the lead section uses a logic that introduces scenarios and adds product value; the selling point list section uses descriptions that combine features and benefits; the usage scenario section generates multiple typical user scenario copy; the generated typical user scenario copy data is transmitted to the material matching unit; the material matching unit is used to match materials for each content section based on the tag matching degree between selling points and materials, and outputs layout suggestions; the copy, materials, and layout suggestions are transmitted to the draft output unit.

[0043] The initial draft output unit integrates data according to a structured framework to form a complete initial draft of the product details page. This initial draft is then transmitted to the data iteration engine and compliance verification module via a data interface, while also being stored in the system's historical database. The data iteration engine collects user behavior data after the details page goes live, analyzes and quantifies the content's effectiveness to generate optimization suggestions, and then feeds these suggestions back to the value anchoring engine and creative generation workshop.

[0044] The data iteration engine of this invention includes: The behavior collection unit is used to collect user behavior data after the details page goes live through page tracking technology, including dwell time, number of clicks in each section, conversion path data and user review data, and transmits the collected data to the performance analysis unit in real time.

[0045] The performance analysis unit uses statistical algorithms and machine learning models to quantify content effectiveness, first calculating the attention level of different sections. Conversion efficiency and evaluation of satisfaction The three indicators are then analyzed using the CART decision tree model to determine the relationship between the indicators and content elements, identify inefficient content elements, and transmit the analysis results to the optimization feedback unit.

[0046] The optimized feedback unit generates optimization suggestions by module, outputs suggestions for adjusting selling point weights to the value anchoring engine, and outputs suggestions for copywriting optimization and material replacement to the creative generation workshop.

[0047] This invention also includes a compliance adaptation unit, integrated into the creative generation workshop. This unit receives the initial draft of the product details page from the initial draft output unit, calls the rule data from the intelligent knowledge base's compliance rule library, and performs compliance verification on the initial draft of the product details page. The specific process for performing compliance verification on the initial draft of the product details page is as follows: After removing words that violate advertising laws and correcting expressions that do not conform to platform guidelines, if the proportion of non-compliant content is... If 10% fail, the product details page is regenerated in the Creative Generation Workshop. Meanwhile, the Compliance Adaptation Unit adjusts the product details page format based on the target platform characteristics, including vertical images and dual-format layouts, and transmits the product details page to the external publishing system. At the same time, the compliance verification results are fed back to the data iteration engine for optimization and analysis.

[0048] Example 1: To make the technical solution of this invention clearer and easier to understand, the following description is based on a certain brand of smart drum washing machine.

[0049] This embodiment focuses on the scenario of generating a product details page for a brand A smart drum washing machine on a popular e-commerce platform, Red Book. The specific execution flow of each module in the system is as follows: ① The raw data receiving unit in the value anchoring engine receives two types of data: Objective product attribute data: The product specification sheet includes AI direct drive motor, 3D transparent drying, light plasma sterilization, and 10kg capacity; the designer notes include minimalist white body, embedded design, touch panel, and technical white paper, which includes noise ≤48dB and energy efficiency level 1.

[0050] Marketing value related data: The initial selling point list included quiet washing and drying, 99.9% sterilization rate, and the initial draft of user feedback included small apartments that cannot fit a large washing machine, clothes that are wrinkle-free after drying, and notes from past collaborating influencers, including notes from a certain influencer on a certain social media platform, home organization expert XX, and notes on transforming a mini washing and drying area on a balcony, with 12,000 interactions.

[0051] The feature recognition unit extracts core attributes by combining natural language processing (NLP) technology with a BERT pre-trained model: technical attributes include AI direct drive motor and 3D transparent drying; performance attributes include 10kg capacity and noise ≤48dB; form attributes include minimalist white body and embedded design; and material texture attributes are extracted by analyzing product images using a ResNet50 model, including ABS antibacterial panel and tempered glass door.

[0052] When extracting emotional selling points, the system combines high-conversion case models of home appliances in the intelligent knowledge base to construct correlations, such as AI direct-drive motor → quiet operation → safe washing for mothers and babies. The TF-IDF algorithm is used to supplement emotional expressions from influencer notes, such as "saving space on the balcony" and "wearing clothes directly after drying." The TextCNN model filters positive scenario-based expressions, such as "a washer-dryer combo that can be installed even on a small balcony" and "sterilization without residue for baby clothes." The weight calculation unit, with specific factor values ​​as shown in Table 1, yields the weight ranking. Table 1. Weights of Core Attributes of a Certain Brand's Smart Front-Loading Washing Machine

[0053] Table 2 Weighting of Emotional Marketing Selling Points

[0054] The selling point output unit transmits the top 3 selling points to the creative generation workshop, including sterilization of maternity and baby clothing (0.89), space saving for small apartments (0.81), and wrinkle-free and ready-to-wear after drying (0.71), and marks them with priority.

[0055] The classification interaction unit of the framework arrangement center of this invention receives product classification: home appliances - washing machine - intelligent drum washer-dryer, with standardized code APPL-WM-SMART-DRY-001; the template matching unit calculates cosine similarity from the details page template library of the intelligent knowledge base.

[0056] Candidate template 1 is a home appliance scenario template, with tags including small apartment, maternal and infant, and washer and dryer, with a matching degree of 92%.

[0057] Candidate template 2 is a parameterized template for home appliances, with tags including performance, energy consumption, and capacity, and a matching degree of 86%. Based on historical conversion effect coefficients, candidate template 1 has a conversion coefficient of 1.2, and candidate template 2 has a conversion coefficient of 0.9. Therefore, candidate template 1, i.e., the home appliance scenario template, is selected as the optimal template.

[0058] The template framework extracted from the structural analysis unit is shown in Table 3: Table 3. Parameter Table of Frame Structure for a Product Details Page Template on a Certain Red Book

[0059] In this embodiment, the marketing experience model library in the intelligent knowledge base calls a high-conversion copywriting model for home appliances from a certain popular book, and the training data includes more than 100,000 notes on best-selling washing machines.

[0060] The influencer knowledge base utilizes content characteristics from influencers on platforms like Douyin (TikTok), home organization expert XX, and parenting blogger Xiaotao, such as scene-based images and short, concise text. The multimedia resource library matches images tagged with "small apartment balcony + washing machine" with a 90% match rate, and short videos comparing sterilization effects with an 85% match rate. The compliance rules library uses filtering rules from the Advertising Law that prohibit the words "best" and "first," and adheres to Douyin's requirement of at least three vertical images.

[0061] In this embodiment's Creative Generation Workshop, the Data Aggregation Unit processes the selling point data (JSON format), template framework (JSON format), and materials (WebP format) in a unified manner and then transmits them to the Generative AI Unit. An example of the generated content is as follows: The title section includes the headline "Small Apartment Balconies Feeling Crowded? A Certain Brand's AI Direct Drive Washer Dryer Saves Space and Achieves 99.9% Sterilization." The scene image section features three vertical images: one showing a built-in installation in a 10-inch... Balcony, comparison of hand washing and machine washing for mothers and babies, clothes unfolded without wrinkles after drying.

[0062] The selling points are as follows: (1) Sterilization for mothers and babies: photo-plasma technology, leaving no residue on baby clothes; (2) Small apartment friendly: built-in design, saving 30% space compared to traditional washing and drying; (3) Wrinkle-free drying: 3D transparent drying, so you can wear it directly after taking it out.

[0063] Here's a review from popular parenting blogger Xiaotao in the user review section: My baby has allergies, so I feel very safe using this antibacterial function to wash drool bibs. They're so soft after drying.

[0064] During the compliance adaptation unit verification, the word "most" in the name of the quietest washing machine in the initial draft was removed, and the images were adjusted to three vertical images (in accordance with the guidelines of a certain red book). The proportion of non-compliant content was 0.5% (<10%), and the compliant initial draft was output.

[0065] The optimization process of the data iteration engine is as follows: 7 days after the details page goes live, the behavior collection unit collects data through event tracking, as shown in Table 4: Table 4 Comparison of User Behavior Data and Effects Before and After the Details Page Goes Live

[0066] The performance analysis unit, using the CART decision tree model, identified that the "wrinkle-free after drying" selling point only accounted for 15% of clicks, classifying it as an inefficient element due to a lack of "clothing wrinkle comparison" materials. The optimization feedback unit output suggestions: The weight of sterilization for maternity and baby clothing was increased from 0.89 to 0.95 in the Value Anchoring Engine, while the weight of wrinkle-free after drying was decreased from 0.71 to 0.65. The "wrinkle-free after drying" material was replaced with "comparison of wrinkles before and after drying of jeans" in the Creative Generation Workshop, and the text was added: "Say goodbye to ironing! Wear them directly out after drying."

[0067] After optimization and relaunch, the "Wrinkle-Free After Baking" section saw a 30% increase in clicks, and the overall conversion rate further improved to 28%.

[0068] The above embodiments are only typical applications of the present invention. The present invention can be adapted to multiple categories such as home appliances, cosmetics, and home furnishings. By adjusting the weight factors of the value anchoring engine, the template tags of the framework arrangement center, and the industry model of the intelligent knowledge base, the intelligent generation and optimization of cross-category detail pages can be achieved.

[0069] Therefore, the present invention adopts the above-mentioned product detail page content generation system based on AI experience model. Through multi-module collaboration, it realizes the automation, intelligence and continuous optimization of product detail pages, and is applicable to e-commerce, new media and other multi-platform scenarios.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A product detail page content generation system based on an AI experience model, characterized in that, This includes a value anchoring engine, a framework orchestration hub, an intelligent knowledge base, a creative generation workshop, and a data iteration engine; The value anchoring engine receives raw product data and identifies and outputs core product attributes and emotional marketing selling points based on a built-in marketing experience model. The framework orchestration hub is used to match and call the corresponding details page template from the intelligent knowledge base based on the product's classification information, and parse the structured framework of the output content; An intelligent knowledge base is used to store marketing experience models, product detail page templates, multimedia materials, compliance rules, and expert knowledge. The Creative Generation Workshop is used to gather marketing resources from the value anchoring engine, the content framework of the framework arrangement center, and the intelligent knowledge base. It generates creative copy and material matching solutions by integrating generative AI models and outputs the first draft of the product detail page. The data iteration engine is used to collect user behavior data after the details page goes live, analyze and quantify the content effect to generate optimization suggestions, and then feed these optimization suggestions back to the value anchoring engine and the creative generation workshop.

2. The product detail page content generation system based on an AI experience model according to claim 1, characterized in that, Value anchoring engines include: The raw data receiving unit is used to receive raw product data, which includes objective product attribute data and marketing value-related data. The feature recognition unit receives data from the raw data receiving unit and extracts the core attributes and emotional marketing selling points of the product. When extracting the core attributes of the product, natural language processing (NLP) technology is used to process the raw text data, and entity recognition is performed through a BERT pre-trained model to extract the physical attributes, technical attributes, and performance attributes of the product. Computer vision technology is used to process the raw image data, and morphological attributes and material texture attributes are extracted through a ResNet50 model. When extracting emotional marketing selling points, the feature recognition unit is based on the core attributes of the product and combines the historical case mapping rules of the marketing experience model library in the intelligent knowledge base to construct the relationship between core attributes, user needs and emotional selling points. Then, the TF-IDF algorithm is used to supplement emotional expressions from the marketing value association data, and the positive scenario-based expressions are filtered by the TextCNN model. The weight calculation unit calculates the weights of the product's core attributes and emotional marketing selling points based on multiple predefined influencing factors. The weight calculation of emotional marketing selling points is affected by the weights of their associated product core attributes. The Selling Point Output Unit is used to transmit the combination of the product's core attributes and emotional marketing selling points with the highest weight to the Creative Generation Workshop, and simultaneously output the priority marks of the top 3 emotional marketing selling points with the highest weight.

3. The product detail page content generation system based on an AI experience model according to claim 1, characterized in that, The framework orchestration hub includes: The classification interaction unit is used to receive product classification information, and transmit the product classification information to the template matching unit after standardizing and encoding the information. The template matching unit, based on the encoded product classification information, uses a cosine similarity algorithm to calculate the matching degree between the classification and the template label. If the matching degree is... 85% are selected as candidate templates; if multiple candidate templates exist, their conversion rate coefficients from historical applications in the details page template library are considered. Select the optimal template and transmit the selected optimal template data to the structural analysis unit; The structure parsing unit is used to parse the structured framework of the template, extract the framework data in the template, including the arrangement order, content proportion and presentation logic of the title area, introduction area, parameter table area, selling point list area, usage scenario area and user review area, and transmit the framework data to the creative generation workshop.

4. The product detail page content generation system based on an AI experience model according to claim 3, characterized in that, The intelligent knowledge base includes: The Marketing Experience Model Library stores standardized model clusters trained using machine learning methods. The model training data includes high-conversion e-commerce product detail page copy, popular social media influencer notes, and brand marketing materials. The Marketing Experience Model Library provides model invocation services to the Value Anchoring Engine through an API interface. The product detail page template library stores structured templates categorized by product type. Each template is associated with historical conversion data, applicable scenario tags, and update time. The product detail page template library provides template data to the framework orchestration center through a template call interface. The multimedia resource library stores product images, video clips, design elements, resource associations, product categories, and matching tags for selling points; the multimedia resource library provides resource data to the creative generation workshop through a resource matching interface; The compliance rule library stores a list of prohibited terms under the Advertising Law, platform guidelines, and industry compliance standards; it also provides compliance data to the Creative Generation Workshop through a compliance verification interface. The Influencer Knowledge Base stores data on influencers across the internet, including content tags and fan profile tags. This data includes basic influencer information, content style characteristics, and historical collaboration results. The Influencer Knowledge Base provides high-quality content feature data to the Creative Generation Workshop through an influencer feature interface.

5. The product detail page content generation system based on an AI experience model according to claim 1, characterized in that, The Creative Generation Workshop includes: The data aggregation unit is used to receive the product's core attributes and emotional selling points combination from the value anchoring engine, the structured framework of the framework orchestration center, multimedia materials from the intelligent knowledge base, and marketing model parameters through an asynchronous data interface. It performs data format unification processing and transmits the processed data to the generative AI unit and the material matching unit. Generative AI Unit, used to integrate pre-trained generative AI models to generate the following content under the constraints of a marketing experience model: The title section uses a sentence structure that combines pain points and selling points; the lead section uses a logic that introduces scenarios and adds product value; the selling point list section uses a description that combines features and benefits; the usage scenario section generates multiple typical user scenario copy; and the generated typical user scenario copy data is transmitted to the material matching unit. The material matching unit is used to match materials to each content section based on the matching degree between selling points and material tags, and output layout suggestions, and transmit the copy, materials, and layout suggestions to the draft output unit; The initial draft output unit is used to integrate data according to a structured framework to form a complete initial draft of the product details page. The initial draft is transmitted to the data iteration engine and compliance verification module through the data interface, and the initial draft is stored in the system's historical database.

6. The product detail page content generation system based on an AI experience model according to claim 1, characterized in that, The data iteration engine includes: The behavior collection unit is used to collect user behavior data after the details page goes live through page tracking technology, including dwell time, number of clicks in each section, conversion path data and user evaluation data, and transmit the collected data to the performance analysis unit in real time; The performance analysis unit uses statistical algorithms and machine learning models to quantify content effectiveness, first calculating the attention level of different sections. Conversion efficiency and evaluation of satisfaction The three indicators are then analyzed using the CART decision tree model to determine the relationship between the indicators and content elements, identify inefficient content elements, and then transmit the analysis results to the optimization feedback unit. The optimized feedback unit generates optimization suggestions by module, outputs suggestions for adjusting selling point weights to the value anchoring engine, and outputs suggestions for copywriting optimization and material replacement to the creative generation workshop.

7. The product detail page content generation system based on an AI experience model according to claim 1, characterized in that, It also includes a compliance adaptation unit, which is integrated into the creative generation workshop. This unit receives the initial draft of the product details page from the initial draft output unit, calls the rule data of the intelligent knowledge base compliance rule library, and performs compliance verification on the initial draft of the product details page.

8. The product detail page content generation system based on an AI experience model according to claim 7, characterized in that, The specific process for compliance verification of the initial draft of the product details page is as follows: After removing words that violate advertising laws and correcting expressions that do not conform to platform guidelines, if the proportion of non-compliant content is... If the percentage is 10%, return to the Creative Creation Workshop to regenerate the product details page; The product detail page format was adjusted based on the characteristics of the target platform, including vertical images and dual-format layouts. The product detail page was then transmitted to an external publishing system, and the compliance verification results were fed back to the data iteration engine for optimization and analysis.

Images