Digital advertising method and system based on virtual community

Abstract

This invention relates to the technical field of digital advertising and discloses a digital advertising method and system based on virtual communities. The invention uses merchant virtual units to display digital models of target products, while customer virtual units can observe products from multiple dimensions. Based on the transaction records of the target products, offline observation information is obtained from the customer virtual units, and the product digital model is reconstructed accordingly, generating a set of measured reference models. By parallel matching of the product digital models of different merchants within the virtual community and the measured reference model set, a personalized temporary community scene is generated for the customer virtual units. Based on the customer's registration information, logistics delivery prediction is performed on the product information in the temporary community, generating a personalized product prediction model. This helps customers make more accurate shopping decisions. This method enhances the personalization and accuracy of advertising, improves the customer shopping experience and advertising effectiveness, and solves the problem of low advertising effectiveness in existing technologies.

Description

Technical Field

[0001] This invention relates to the technical field of digital advertising, and more particularly to a digital advertising method and system based on virtual communities. Background Technology

[0002] Among existing advertising technologies, traditional advertising models are mainly divided into display advertising, search engine advertising, social media advertising, and video advertising. Although existing advertising models have made some progress in accuracy and user reach, some obvious shortcomings still exist. Most traditional advertisements are still one-way, static displays. There is a lack of interaction between the advertisement and the user, and the advertisement content cannot be dynamically adjusted according to the user's real-time needs or environment, resulting in low relevance and effectiveness of the advertisement. At the same time, the single-target advertising model cannot fully and effectively present multifaceted information to the user. Summary of the Invention

[0003] The purpose of this invention is to provide a digital advertising method and system based on virtual communities, aiming to solve the problem of low advertising effectiveness in the prior art.

[0004] The present invention is implemented as follows: In a first aspect, the present invention provides a digital advertising method based on a virtual community, comprising:

[0005] The virtual community allows merchants to showcase digital models of target products, enabling customers to observe these models from multiple dimensions using their own virtual units.

[0006] Based on the transaction records of the target product, offline observation information is retrieved from the corresponding virtual customer unit, and the digital model of the product is reconstructed based on the offline observation information to generate a set of measured reference models for the target product.

[0007] By performing parallel matching of the digital models of the products displayed by each merchant's virtual unit in the virtual community and the actual test reference model set, a temporary community scene is generated for the customer's virtual unit in a personalized way.

[0008] Based on the registration information of customer virtual units in the virtual community, the product information of each merchant virtual unit in the temporary community scene after logistics transportation is predicted, so as to generate a product prediction model for customer virtual units to purchase target products in each merchant virtual unit.

[0009] Secondly, the present invention provides a digital advertising system based on a virtual community, used to implement the digital advertising method based on a virtual community as described in any one of the first aspects, comprising:

[0010] The product display module is used to display the digital model of the target product through the virtual units of merchants in the virtual community, so that customers can observe the digital model of the target product from multiple dimensions through their virtual units;

[0011] The reference modeling module is used to retrieve offline observation information from the corresponding virtual customer unit based on the transaction records of the target product, and reconstruct the digital model of the product based on the offline observation information to generate a set of measured reference models of the target product.

[0012] The personalized scene module is used to generate temporary community scenes for the customer's virtual unit by performing parallel matching of digital models of goods displayed by each merchant's virtual unit in the virtual community and a set of actual test reference models;

[0013] The information prediction module is used to predict the product information after logistics delivery based on the registered information of the customer's virtual unit in the virtual community and the measured reference model set of each merchant's virtual unit in the temporary community scenario, so as to generate a product prediction model for the customer's virtual unit to purchase target products from each merchant's virtual unit.

[0014] This invention provides a digital advertising method based on virtual communities, which has the following beneficial effects:

[0015] This invention utilizes digital modeling of target products displayed in merchant virtual units. Customer virtual units can observe products from multiple dimensions. Based on transaction records of the target products, offline observation information is obtained from the customer virtual units, and the product digital model is reconstructed accordingly, generating a set of measured reference models. By parallel matching of product digital models and the set of measured reference models from different merchants within the virtual community, a personalized temporary community scene is generated for the customer virtual units. Based on the customer's registration information, logistics delivery prediction is performed on product information in the temporary community, generating a personalized product prediction model. This helps customers make more accurate shopping decisions. This method enhances the personalization and precision of advertising, improves the customer shopping experience and advertising effectiveness, and solves the problem of low advertising effectiveness in existing technologies. Attached Figure Description

[0016] Figure 1 This is a schematic diagram illustrating the steps of a digital advertising method based on a virtual community provided in an embodiment of the present invention;

[0017] Figure 2 This is a schematic diagram of the structure of a digital advertising system based on a virtual community provided in an embodiment of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0019] The implementation of the present invention will be described in detail below with reference to specific embodiments.

[0020] Reference Figure 1 , Figure 2 The diagram shows a preferred embodiment of the present invention.

[0021] In a first aspect, the present invention provides a digital advertising method based on a virtual community, comprising:

[0022] S1: Display the digital model of the target product through the virtual units of merchants in the virtual community, so that customers can observe the digital model of the target product from multiple dimensions through their virtual units;

[0023] S2: Based on the transaction records of the target product, retrieve offline observation information from the corresponding customer virtual unit, and reconstruct the digital model of the product based on the offline observation information to generate a measured reference model set of the target product;

[0024] S3: By performing parallel matching of the digital product models displayed by each merchant's virtual unit in the virtual community and the actual test reference model set, a temporary community scene is generated for the customer's virtual unit in a personalized way;

[0025] S4: Based on the registration information of the customer's virtual unit in the virtual community, predict the product information after logistics transportation for the measured reference model set of each merchant's virtual unit in the temporary community scene, so as to generate a product prediction model for the customer's virtual unit to purchase target products from each merchant's virtual unit.

[0026] Specifically, in step S1 of the embodiment provided by the present invention, the merchant creates a virtual merchant unit in the virtual community and uses various digital modeling tools and technologies (such as 3D modeling, augmented reality (AR), virtual reality (VR), computer vision, etc.) to transform the target product into a digital model. The model includes not only the appearance of the product, but also the product's structure, material, color, size and other features.

[0027] More specifically, based on the attributes of the target product, the merchant's virtual unit designs a multi-dimensional display method in the virtual space to generate a 3D model of the product. Customers can rotate, zoom in, and zoom out to view the appearance of the product from different angles. The material characteristics of the product (such as metal, fabric, plastic, etc.) are displayed through detailed textures and lighting effects, allowing customers to intuitively perceive the quality and feel of the product. Merchants can design product versions with different colors, styles, or shapes and allow customers to switch freely in the digital model, thereby enhancing customers' interest and understanding of the product. Some products (such as electronic products and mechanical products) may display their internal structure, showing the internal components and their working principles through disassembled views.

[0028] More specifically, product displays are not limited to simple product models; products can also be placed in specific scenarios (such as a virtual store, home environment, or usage scenario). This allows customers to see how the product behaves in a real environment, thereby enhancing the immersive experience of the product display. Through digital modeling technology, various characteristics of the product (appearance, material, size, etc.) are accurately presented, allowing customers to understand the product more clearly. Customers can not only observe the product from different perspectives but also understand its internal structure, materials, and actual applications, improving the user experience. Customers can interact with the product model according to their needs (rotate, zoom, select different versions, etc.) to obtain personalized product information displays. Through AR and VR technologies, customers can place virtual product models in the real world for observation, further enhancing the shopping experience.

[0029] More specifically, customers virtually enter the virtual community platform, select the products they want to observe, and can access the product display interface through various devices (such as mobile phones, computers, VR headsets, etc.) to begin interacting with the digital model of the product. Customers can rotate the product, zoom in and out of the product model, and view different details of the product through the operation interface. If the product has multiple colors, styles, or versions, customers can switch in real time and observe the appearance and differences of each variant. Customers can view detailed information such as the material, weight, and size of the product, and can even virtually disassemble the product to understand its structure. Customers can switch between different display scenarios (such as placing the product in a virtual home scene to see how the product looks in different environments).

[0030] More specifically, based on customers' historical behavior data, interests, preferences, and real-time feedback, the virtual community system can personalize product display recommendations for customers. For example, if a customer is more interested in a certain type of product (such as electronic products), the system can automatically push disassembled views or technical details of the product. Through multi-dimensional interaction with product models, customers can obtain an immersive experience similar to offline shopping, gain a deeper understanding of the product. Customers' observation methods and preferences can be recorded in real time, and the platform optimizes product displays based on this information to better meet customer needs. Through this multi-dimensional interaction and display, customers can better understand the product, thereby increasing their confidence in making purchasing decisions and ultimately improving conversion rates.

[0031] Understandably, by digitally modeling target products through virtual units of merchants in virtual communities and providing customers with multi-dimensional product observation, customers can learn about products from multiple angles and dimensions. Customers are not just passively viewing products, but can interact with them. Based on customer interests and needs, product displays can be personalized, improving customer satisfaction. Through detailed product information displays and immersive experiences, customers are more likely to make purchasing decisions, thus increasing advertising conversion rates.

[0032] Specifically, in step S2 of the embodiment provided by the present invention, the system accesses the transaction database in the virtual community platform to retrieve customer transaction records related to the target product. These records include information such as the customer's purchase history, shopping preferences, payment methods, purchase frequency, and product reviews.

[0033] More specifically, the customer virtual unit collects offline observation data related to the product through its connection with the offline environment. This data may include: customer feedback when actually using the product (such as product usage data obtained through smart home devices, mobile applications, etc.), perceptual information obtained by customers through AR / VR interaction in physical stores or certain places (such as the actual size, color, and user experience of the product), customer evaluation information of the product, such as product usage experiences posted in virtual communities, or opinions and suggestions about the product collected through other online / offline channels (such as social media, forums, etc.). This offline observation data from different channels is aggregated, cleaned, and preprocessed to ensure the consistency and accuracy of the information. In particular, for the collection of data such as the product's real physical attributes, user experience, and environmental adaptability, the system will perform multi-dimensional data fusion through multiple data sources (such as device sensors, user feedback, etc.).

[0034] More specifically, this approach effectively integrates offline observation information from different sources, including real-time customer feedback and actual usage, ensuring the comprehensiveness and accuracy of the data. Customer transaction behavior and usage experience can directly influence product modeling, making product displays more closely reflect actual usage.

[0035] More specifically, the system adjusts the digital model of the product based on offline observation data obtained from the customer's virtual unit. If the customer reports that the actual size or color of the product deviates from the displayed digital model, the system will correct the digital model of the product based on this information. Based on the customer's feedback on the texture of the product (such as touch, weight, material, etc.), the system optimizes the material properties of the product's digital model (such as re-presenting the texture of the product through augmented reality technology). Based on the user feedback of the product, the system can redesign the interactive functions of the product (for example, if the customer reports that the operation interface of an electronic product is not intuitive, the system will adjust the interactive design in the digital model).

[0036] More specifically, the reconstructed digital product models will become part of the real-world reference modeling. These models will not only reflect the product's appearance but will also incorporate customer feedback on actual use, post-purchase experiences, and the product's true performance to provide a more accurate and objective product display. These reconstructed models will be available to other customers on the virtual community platform for multi-dimensional observation. The reconstruction of product digital modeling is not a one-time process but a dynamic and continuous optimization process. As more customer feedback and transaction data accumulate, the system will continuously adjust and optimize the digital modeling of products, forming an adaptive product display system.

[0037] More specifically, through customer feedback and offline observation information, the digital modeling of products can more realistically reflect the actual effect of the products. The product model can be dynamically adjusted according to customer feedback to ensure that the digital model can adapt to the needs of different users and the performance of products in the real world. With the accumulation of data and the establishment of feedback mechanisms, the system can continuously optimize the display effect of products and achieve highly personalized and accurate recommendations.

[0038] More specifically, based on all customer transaction records, offline observation data, and digital modeling reconstruction results, the platform aggregates optimized versions of multiple product models into a "tested reference modeling set." These models include various versions of the product (such as different colors, sizes, materials, etc.) and multi-dimensional models reflecting the product's performance in different environments. To ensure the accuracy of the final tested reference modeling set, the system optimizes the models through cross-validation. For example, it uses transaction records and feedback from different customer groups to verify the model's accuracy and further adjusts the model's performance based on the validation results. Customers can choose the product version that best suits their needs from the tested reference modeling set (such as choosing the product size or color that best matches their expectations). Through tested reference modeling based on real data, customers can obtain a display effect that best reflects the actual product performance, greatly improving the product's purchase conversion rate.

[0039] Understandably, by retrieving transaction records and offline observation information from virtual customer units and feeding this information back into the digital modeling of products, the digital modeling of target products can be reconstructed, thereby generating a more accurate and realistic set of measured reference models. By continuously optimizing the product model to make it closer to the performance of real products, and dynamically adjusting the product display based on customer feedback and needs, the user experience can be improved. More accurate product displays can increase customer trust in products, thereby increasing the conversion rate of purchase decisions. As customer feedback accumulates, the system can continuously optimize the product model to achieve higher personalization and accuracy.

[0040] Specifically, in step S3 of the embodiment provided by this invention, digital models of the products displayed by each merchant virtual unit in the virtual community are extracted. These digital models include information such as the appearance, material, size, color, and function of the products, showcasing their virtual performance. Simultaneously, different versions of the products (including different colors, sizes, materials, etc.) are extracted from the actual reference model set of the target products. These versions are digital models optimized based on customer feedback and offline observation data, reflecting the products' performance in actual use. All product digital models and actual reference model sets are standardized and formatted to enable efficient comparison and analysis in subsequent matching and processing. This ensures that product digital models and actual reference model sets extracted from different merchants and platforms can be processed under a unified standard, improving data integration efficiency. Through high-quality product digital models and actual verified reference model sets, the accuracy and authenticity of the product display are ensured.

[0041] More specifically, after customers enter the virtual community platform, the platform collects their personal preference data, such as product categories, styles, brands, price ranges, historical purchase records, and hobbies. Through data such as customers' virtual unit behavior, interactions, comments, and ratings, the platform can create customer interest profiles. Based on these profiles and needs, the platform generates personalized scenario requirements. For example, a customer might need a product display in a home environment or a product display in a work setting. Customer needs can be expanded from multiple dimensions, from product attributes (such as color, material, and function) to scenario design (such as room layout and environmental atmosphere). Through in-depth analysis of customers' virtual unit behavior, the platform can accurately understand their shopping needs and design personalized scenarios based on this. Through continuous collection and analysis of customer behavior data, the platform can build more intelligent and accurate customer profiles, providing data support for personalized scenario generation.

[0042] More specifically, based on the needs of customers' virtual units, the system uses matching algorithms (such as content-based recommendation algorithms, collaborative filtering algorithms, deep learning models, etc.) to compare and match customer preferences with the digital models of goods displayed by merchants. This process considers the degree of matching between various characteristics of the goods (such as appearance, size, color, function, etc.) and customer needs (such as color preferences, style preferences, etc.). At the same time, it matches customer needs with a set of tested reference models to ensure that the performance of the goods displayed to customers in the actual use environment is consistent with customer expectations. For example, some customers may pay more attention to the durability and functionality of goods, and the system will prioritize matching and displaying those product models that have been tested and optimized.

[0043] More specifically, by combining digital product modeling with real-world reference modeling, a comprehensive product display set is generated. This set is optimized based on product attributes, performance, customer personalized needs, and actual product usage feedback, ensuring that customers can see product models that best meet their needs in the virtual community. By combining digital product modeling and real-world reference modeling, the platform can provide customers with product choices that better meet their needs, improving product relevance and accuracy. Based on the parallel matching of customer needs and actual usage data, each customer can receive product recommendations that best match their personal interests and actual needs.

[0044] More specifically, based on customers' needs for virtual units, the platform uses 3D modeling and virtual reality technology to generate personalized temporary community scenes. For example, if a customer is interested in a certain type of product (such as furniture, home appliances, etc.), the platform will automatically construct a virtual home scene containing these products. Customers can see how the products are matched with other items in the virtual environment. Customers can interact with the products in the personalized community scene, such as viewing the products from different angles, changing materials and colors, and adjusting the product layout. Through augmented reality (AR) technology, customers can also place these virtual product models into their own real-world environment for observation. The platform dynamically adjusts scenarios based on customer preferences. If a customer shows strong interest in certain products, the platform can adjust the scenario in real time, rearranging and displaying the products. The scenario is not just a static display, but changes in real time based on customer interaction and needs. By generating personalized temporary community scenarios, customers can experience the actual effects of products in an immersive way, thereby enhancing the shopping experience. Customers can adjust the scenario according to their needs and interests for interactive product displays, enhancing user engagement and confidence in purchasing decisions. As customer behavior changes, the scenario can be adjusted instantly to maintain efficiency and personalization, ensuring that customers receive the most relevant and expected display.

[0045] Understandably, the system will ultimately present a personalized temporary community scenario based on the customer's individual needs and matching results. This scenario will include multiple digital product models and actual test reference models. Customers can freely browse products within the scenario, view detailed product information, and make purchase decisions. Based on the customer's browsing behavior and interests, the system will further recommend related products or provide purchase suggestions to help customers make purchase decisions quickly. Through personalized scenarios and product displays, customers can more intuitively understand the performance of products, thereby improving the efficiency and accuracy of purchase decisions. The generation of personalized scenarios and product displays can attract more customer attention, increase their willingness to buy, and thus improve conversion rates.

[0046] Specifically, in step S4 of the embodiment provided by this invention, customer registration information in the virtual community is collected, including personal data (such as age, gender, and geographical location), behavioral data (such as browsing history, purchase history, and search keywords), and preference data (such as product category preference, price range, and brand preference). By analyzing the customer registration information, a personalized customer profile is generated, which includes the customer's shopping preferences, interests, and purchasing power. This process can be implemented using machine learning models (such as cluster analysis and collaborative filtering) to ensure that customer needs are fully identified. Through comprehensive collection and analysis of customer virtual unit information, a precise customer profile can be depicted, providing a valid basis for subsequent product recommendations and predictions. Intelligent analysis of customer behavior allows for a more accurate understanding of customer shopping needs, providing data support for product prediction and scenario generation.

[0047] More specifically, in the temporary scenarios of the virtual community, product information from each merchant's virtual unit is extracted. This information includes digital product modeling, actual reference modeling, inventory status, price, sales history, etc. Actual reference modeling data provided by the merchants is also obtained. This data includes information on the product's performance in actual use environments, such as functionality, durability, customer reviews, and return / exchange rates. By processing and classifying this actual data, the accuracy of product display is ensured. Through the systematic collection of product information from merchant virtual units, a complete product database is formed, ensuring that the product recommendations and displays received by customers are based on comprehensive and accurate data. Actual reference modeling ensures that product displays better meet customers' actual needs and expectations, improving the realism of product displays.

[0048] More specifically, the predictive model incorporates information about potential changes in goods during the logistics process (such as transportation damage, inventory changes, and transportation delays). For example, some goods may be damaged or lost during transportation, affecting their final delivery status. This information needs to be combined with the merchant's inventory data and goods transportation information to generate post-logistics goods information. Based on customer demand profiles and merchant goods information, predictive algorithms (such as regression analysis, time series forecasting, and deep learning) are used to predict the types, quantities, and purchase times of goods that customers may buy. The model needs to make real-time predictions of goods information based on potential changes during the logistics process, such as delayed delivery and insufficient inventory. Common predictive algorithms include: collaborative filtering algorithms: predicting the goods that customers may buy based on the purchase history of similar customers; deep learning (such as RNN and LSTM): used to process time series data and predict customer purchasing behavior; and Bayesian network models: considering the correlation between goods, customer purchasing patterns, and the impact of logistics changes for comprehensive prediction.

[0049] More specifically, based on actual logistics information and real-time customer behavior, the predictive model will dynamically adjust to ensure the accuracy and timeliness of product information. For example, if a product cannot arrive on time due to logistics problems, the predictive model will update the customer's product recommendations in real time to avoid recommending inaccurate products. By modeling the logistics and transportation process of products, it can more accurately predict the products that customers may buy and provide customers with estimated information in advance. The predictive model can dynamically adapt to changes that occur during the logistics process, such as delays or damages, to ensure that customers can always see the latest and most accurate product information.

[0050] More specifically, based on the demand profiles of customers in virtual units, the product information of merchants in virtual units, and the prediction results of logistics delivery, a predictive model is generated to predict whether customers will purchase target products from various merchants' virtual units. This model not only covers product selection but also considers factors such as customer purchase preferences, time, and quantity. Based on the predicted purchase behavior of customers, a personalized product recommendation list is generated. The recommendation list optimizes the order and content of product display based on the customer's demand profile, product prediction model, and product information provided by merchants, thereby improving the customer's shopping experience. When customers interact with merchants' virtual units in the virtual community (such as browsing products, adding products to the shopping cart, etc.), the product prediction model will adjust in real time, providing personalized product recommendations and updated information. By integrating customer behavior data and product prediction model, it can accurately recommend products that customers may be interested in and that meet their needs, improving the relevance and effectiveness of recommendations. Based on the prediction model, customers can understand key information such as product inventory, price fluctuations, and logistics status in advance, thereby making more informed purchasing decisions. Personalized product prediction and recommendations can greatly improve customer satisfaction, thereby increasing the product conversion rate in the virtual community.

[0051] More specifically, in the temporary scenarios of the virtual community, based on customer needs and product prediction modeling, the system displays products that each customer may purchase and related information, including detailed product descriptions, prices, logistics status, and inventory status. The system updates product predictions in real time based on customer behavior, showing customers the most relevant products. The system also provides corresponding purchase suggestions and logistics information so that customers have sufficient information to support their purchasing decisions. The real-time nature of product prediction modeling ensures that the product display always meets the latest needs and actual situation of customers, improving the real-time nature and relevance of the display. Customers can see products that best meet their needs in the virtual community, while obtaining accurate logistics and inventory information, reducing uncertainty in purchasing decisions.

[0052] Understandably, by analyzing the registration information of customer virtual units and combining it with the actual measurement reference modeling data of merchant virtual units' products and the prediction of product information after logistics delivery, personalized product prediction models can be generated. This helps customers make purchasing decisions that better meet their needs in the virtual community. Through customer profiles, product data, and logistics predictions, personalized product recommendations can be provided to meet customers' actual needs. When logistics and inventory information change, the prediction model can be updated in real time to ensure the accuracy of the displayed products. Through accurate product recommendations and logistics information support, the efficiency and satisfaction of customers' purchasing decisions can be improved.

[0053] This invention provides a digital advertising method based on virtual communities, which has the following beneficial effects:

[0054] This invention utilizes digital modeling of target products displayed in merchant virtual units. Customer virtual units can observe products from multiple dimensions. Based on transaction records of the target products, offline observation information is obtained from the customer virtual units, and the product digital model is reconstructed accordingly, generating a set of measured reference models. By parallel matching of product digital models and the set of measured reference models from different merchants within the virtual community, a personalized temporary community scene is generated for the customer virtual units. Based on the customer's registration information, logistics delivery prediction is performed on product information in the temporary community, generating a personalized product prediction model. This helps customers make more accurate shopping decisions. This method enhances the personalization and precision of advertising, improves the customer shopping experience and advertising effectiveness, and solves the problem of low advertising effectiveness in existing technologies.

[0055] Preferably, the step of displaying digital product models of target goods through virtual merchant units in a virtual community, allowing customers to observe the digital product models of the target goods from multiple dimensions through virtual units, includes:

[0056] S11: Based on the merchant information registered by the merchant on the community platform, construct a virtual merchant unit for the merchant. At the same time, based on the advertising service contract opened by the merchant on the community platform, perform information scheduling on the virtual merchant unit to deploy the virtual merchant unit in several designated virtual communities on the community platform.

[0057] S12: Upload multi-angle observation information of the goods to be sold through the merchant virtual unit, and perform 3D digital modeling of the goods to be sold based on the multi-angle observation information using the NeRF deep learning model deployed on the community platform, so as to generate a basic version of the digital model of the goods.

[0058] S13: Upload several extended observation dimension information of the goods to be sold through the merchant virtual unit, so as to optimize the basic version of the digital model of the goods in multiple dimensions according to the extended observation dimension information, so as to generate the completed version of the digital model of the goods.

[0059] S14: Obtain the interaction behavior instructions of the customer virtual unit in the merchant virtual unit, select the specified goods to be sold in the merchant virtual unit as the target goods according to the interaction behavior instructions, and retrieve the digital model of the target goods and deploy it in the pre-built multi-dimensional observation and perception framework.

[0060] S15: Obtain the perception instructions of the customer's virtual unit through the multi-dimensional observation and perception framework, and display the digital model of the product in a corresponding form according to the perception instructions.

[0061] Specifically, after registering their information on the virtual community platform, merchants need to construct corresponding virtual merchant units. This information includes the merchant's basic information, shop display content, brand logo, and product types. In the platform backend, the merchant's virtual unit is mapped to a virtual shop in the virtual community scene. According to the advertising service agreement activated by the merchant on the platform, the merchant's virtual unit will be deployed to the designated virtual community. The platform uses an information scheduling system to adjust and optimize the advertising display content of the merchant's virtual unit according to the requirements of the agreement. Through the information scheduling system, the platform can accurately display the merchant's virtual unit in the appropriate virtual community, ensuring that the advertising content meets customer needs and increases the merchant's exposure. Merchants can customize the display content according to the advertising agreement to improve their marketing efficiency and customer conversion rate.

[0062] More specifically, merchants upload multi-angle observation information (such as product images, videos, 3D scans, etc.) of the products to be sold through the platform to ensure that all aspects of the products are fully captured. This information provides the basic data for subsequent digital modeling. Using the NeRF (Neural Radiation Field) deep learning model deployed on the community platform, a 3D digital model of the product is generated based on the multi-angle observation information. NeRF technology can reconstruct the real 3D form of the product from multiple perspectives according to different viewing angles and lighting conditions, thereby generating a basic version of the product digital model. The 3D digital model generated by the NeRF deep learning model can accurately restore the details of the product, ensuring the authenticity and detail of the product display. The product is modeled from multiple angles, allowing customers to view the product from all angles.

[0063] More specifically, merchants upload additional observational dimensions of product information in their virtual units, such as product size, color, material, usage effects, and functions. This information can be uploaded via images, videos, sensor data, etc. Based on this extended observational dimension information, the basic 3D digital model of the product is further optimized, enhancing its realism and interactivity. The optimized digital model will more closely match the various attributes of the actual product and provide a richer user interactive experience. By expanding the observational dimensions, the digital model of the product not only closely resembles the real product in appearance but also better reflects its other functions and characteristics. Customers can obtain richer product information through the optimized digital model, thereby improving the shopping experience.

[0064] More specifically, when a customer interacts with a merchant's virtual unit (such as browsing products, adding products to their cart, or clicking on product details), the system selects the target product from the merchant's virtual unit based on the customer's interaction. Following this interaction, the platform retrieves the digital model of the target product and prepares to deploy it to the customer's viewing interface. Through this operation, the customer can see a 3D digital model of the product they are interested in. Based on the customer's interaction, the system can accurately provide products that interest them, increasing customer engagement and satisfaction. Customers can easily select target products and view their detailed digital models through interaction, thereby improving the efficiency of their purchase decisions.

[0065] More specifically, a multi-dimensional observation and perception framework is built on the platform for customers to display digital models of goods in virtual units. The framework supports customers to observe goods from multiple dimensions (such as color, size, material, function, etc.) and can provide a 360-degree panoramic view. The digital model of the target product is deployed into this multi-dimensional observation and perception framework. The framework will display the product's 3D model, interactive features, and any other additional information related to the product. Through the multi-dimensional observation and perception framework, customers can observe the product from multiple dimensions, including perspective, color, details, etc. Customers can control the observation angle in real time and view all the details of the product, enhancing the immersive shopping experience.

[0066] More specifically, when customers observe products within a multi-dimensional observation and perception framework, the system provides feedback based on their perceptual commands. For example, customers may adjust the way products are displayed by rotating, zooming, or selecting colors. Based on these commands, the platform will adjust the digital model of the product accordingly, such as changing the display angle, adding or removing details, or providing more product-related information. Customers can control various dimensions of the product display through commands, obtaining a customized product viewing experience. By precisely controlling the product display content through multi-dimensional perceptual commands, customers can enhance their confidence and satisfaction in making purchasing decisions.

[0067] Understandably, through the above steps, merchants' virtual units can effectively showcase the digital modeling of their products in the virtual community, and customers can interact and observe within a multi-dimensional framework. This ensures that customers receive more accurate and richer product information. The NeRF deep learning model and information optimization through expanded observation dimensions ensure that the product display is realistic and detailed. Customers can select products, adjust the display angle, and observe details according to their needs and behavioral instructions, enhancing the personalization and interactivity of shopping. Based on customer perception instructions, the product display can be adjusted in real time, improving customer participation and the accuracy of purchasing decisions. This approach not only improves the product display effect in the virtual community but also enhances the interaction and immersion between customers and products, increasing customer shopping satisfaction and platform conversion rates.

[0068] Preferably, the step of uploading several extended observation dimensions of the goods to be sold through the merchant's virtual unit, and optimizing the basic version of the digital model of the goods in multiple dimensions based on the extended observation dimensions to generate the final version of the digital model of the goods includes:

[0069] S131: The extended observation dimensions of the goods to be sold uploaded by the merchant's virtual unit include morphological change dimension information, scene effect dimension information, internal disassembly dimension information, and time influence dimension information. By extending the observation mechanism of the corresponding morphological change dimension information, scene effect dimension information, internal disassembly dimension information, and time influence dimension information through the digital modeling of the goods, a complete version of the digital model of the goods is generated.

[0070] S132: Based on the morphological change dimension information, assign variable parameters and material elastic coefficients to the basic version of the digital model of the commodity, so that the digital model of the commodity has a morphological change observation mechanism;

[0071] S133: Based on the scene effect dimension information, perform several types of scene modeling arrangements on the basic version of the product digital modeling to generate product digital models arranged in various scene models;

[0072] S134: Perform ray tracing simulation and physical collision simulation on the digital product model arranged in various scene modeling to generate scene effect information of various types of digital product model, and perform pre-retrieval processing on the basic version of the digital product model according to the scene effect information of various types of scene model, so that the digital product model has a scene effect observation mechanism.

[0073] S135: Based on the internal disassembly dimension information, the basic version of the digital modeling of the product is disassembled into modeling units, and internal observation information is assigned to the disassembled modeling units so that the digital modeling of the product has an internal disassembly observation mechanism.

[0074] S136: Perform time decay simulation on the basic version of the digital model of the product based on the time influence dimension information to generate a dynamic mask based on time change parameters and usage frequency parameters. By superimposing the dynamic mask on the digital model of the product, the digital model of the product has a time influence observation mechanism that displays the corresponding time decay simulation effect according to the input time change parameters and usage frequency parameters.

[0075] Specifically, the merchant's virtual unit uploads extended observational dimension information of the goods to be sold to the platform. This dimension information includes: morphological change dimension information: describing the possible morphological changes of the goods under different conditions, such as material changes, expansion, contraction, deformation, etc.; scene effect dimension information: describing the display effect of the goods in different virtual environments or scenes, such as lighting, shadows, reflection, transparency, etc.; internal disassembly dimension information: involving the internal structure of the goods, including the layout, function, and working principle of internal parts; and time impact dimension information: reflecting the changes that the goods may undergo over time, such as aging, wear and tear, and changes caused by usage frequency. This information provides an important basis for the subsequent optimization of the digital modeling of the goods.

[0076] More specifically, based on the dimensional information of product form changes, developers add variable parameters (such as stretching, folding, deformation, etc.) and material elasticity coefficients to the basic version of the product's digital model. These parameters can adjust the product's form changes under different usage scenarios, realizing a form change observation mechanism in digital modeling. This allows the product to display the expected form change effect under different conditions. By assigning material elasticity coefficients and variable parameters, the digital modeling of the product can achieve more realistic physical responses, such as the deformation of soft objects or the size changes of hard objects. Customers can interact based on the product's form change mechanism, such as viewing the changes of the product under different conditions, thus providing a richer experience.

[0077] More specifically, based on scene effect information, digital product models are deployed across multiple scenarios, including indoor and outdoor displays under varying lighting conditions. Each scenario features different backgrounds, light sources, colors, and material reflections. Within these deployed scenarios, ray tracing algorithms are used to simulate lighting and reflection, creating realistic light and shadow effects. Simultaneously, physical collision simulation technology is employed to ensure the product's display exhibits reasonable physical behavior within the scenarios (e.g., positional changes after collisions, bounces, etc.). Based on scene effect information, the product's display effects in different scenarios are generated, including physical reactions, lighting changes, and reflections. These effects are then retrieved and optimized to ensure a more realistic product display. The digital product model can adapt to different scenarios and display different effects in various environments, such as lighting changes and shadow reflections. Customers can view the product in different scenarios, perceiving different display effects and enhancing the immersive experience.

[0078] More specifically, based on the internal disassembly dimensions of the product, the digital model of the product is disassembled into units. Each disassembled unit represents a part of the product's interior, such as components or modules. Each disassembled modeling unit is assigned internal observation information, including material properties, functional display, and working principle. Customers can interactively view the product's internal structure and its working method. Through visualization, customers can interactively view the product's internal structure, understand its components and operating principles. Customers can view the product's internal components from a disassembled perspective, understanding the product's structure and function. This function not only helps customers understand the actual composition of the product but also helps increase their trust in the product and their desire to purchase.

[0079] More specifically, based on the time-related dimension information, the digital model of the product simulates time decay, mimicking changes in the product over time, such as wear, aging, and failure. The frequency of use and changes over time affect the product's appearance and function. A dynamic mask is created to simulate the product's state changes after prolonged use. By overlaying the dynamic mask with the digital model, the product displays effects related to usage time and frequency, such as scratches, fading, and functional degradation. A time-related observation mechanism is implemented in the digital model, allowing customers to see the product's decay process over time and experience the changes during use. The digital model realistically simulates the effects of long-term use, such as wear, aging, and component damage, increasing the product's realism. Customers can make more rational purchasing decisions based on time-related information, such as understanding the impact of product durability and usage frequency on product lifespan.

[0080] More specifically, by optimizing the overall form changes, scene effects, internal disassembly, and the impact of time, a complete version of the product digital model is generated. This version not only showcases the product's appearance but also includes its multi-dimensional characteristics, such as display effects, form changes, and internal structure in different scenarios. The completed product digital model is deployed to a virtual community platform, where customers can view and experience the product's various characteristics through an interactive interface, increasing the product's attractiveness. Through multi-dimensional optimization, customers can gain a comprehensive understanding of the product, enhancing their purchase confidence. Customers can also customize their viewing experience according to their needs, improving the interactive experience and shopping enjoyment.

[0081] Understandably, by processing information across four dimensions—form changes, scene effects, internal disassembly, and the impact of time—merchant virtual units can generate a highly detailed digital model of a product. In the virtual environment, customers can not only observe the appearance of the product but also gain a deeper understanding of its structure, applicable scenarios, lifespan, and other aspects, greatly enhancing the virtual shopping experience and interactivity. These optimizations will effectively enhance customer immersion and help merchants provide more attractive and trustworthy product displays.

[0082] Preferably, the step of retrieving offline observation information from the corresponding virtual customer unit based on the transaction records of the target product, and reconstructing the digital model of the product based on the offline observation information to generate a measured reference model set for the target product includes:

[0083] S21: Based on the offline data reference service activated by the customer's virtual unit on the community platform, the transaction records of the target product are located and searched to determine several offline observation information that conforms to the customer's virtual unit's offline data reference service; wherein, the offline observation information is the information collected by the customer's virtual unit after purchasing the target product by observing the target product from multiple angles and in multiple ways;

[0084] S22: The digital model of the target product is used as the verification object. Data verification processing is performed on the verification object based on the offline observation information. Based on the results of the data verification processing, the confidence level of the digital model of the product is marked at each specific location.

[0085] S23: Based on the confidence level of the markings at various specific locations in the digital model of the product, perform a representative analysis on each of the offline observation information to construct several representative measured reference models, which together constitute the measured reference model set of the target product.

[0086] Specifically, customer virtual units can activate offline data reference services through the platform to provide offline observation data after purchasing and using target products. The system will search for relevant transactions that match the offline data reference service of customer virtual units based on the transaction records of the target products. These transaction records will be used to locate the offline observation information provided by customers after purchasing products. Through precise retrieval of transaction records, it can ensure that offline observation information related to the target products can be obtained from customer virtual units, ensuring the authenticity of the offline data reference service and further improving the reliability and practicality of product modeling.

[0087] More specifically, offline observation information comes from customers' multi-angle and multi-method observations of the products after purchase, such as product wear and tear, functional changes, appearance damage, and durability. This information can be provided in various forms, such as photos, videos, and text descriptions uploaded by customers' virtual units. Integrating this offline observation information into a unified database or system ensures that each piece of observation information matches the specific transaction record of the target product. The offline observation information provided by customers comes from different angles and methods, which can comprehensively reflect the performance of the product in actual use. This multi-dimensional information helps to enrich the accuracy of product digital modeling. By collecting actual usage information from customer feedback, the digital modeling of the product can be closer to the real usage state, enhancing the realism of the virtual product display.

[0088] More specifically, the original digital product model is used as the verification object for subsequent verification operations. Based on the offline observation information provided by the customer, the system performs data verification on each part of the product model. According to each piece of offline observation information, each part of the digital product model is verified. The purpose of verification is to confirm whether the digital model matches the actual performance of the customer. For example, whether the scratches on the product surface, color fading, and component wear are consistent with customer feedback. Based on the data verification results, a confidence level mark is marked for each part of the digital product model. The confidence level mark indicates the degree of consistency between the digital model data of that part and the offline observation information, representing the accuracy of that part in actual testing. Through data verification processing, the difference between the digital model and the actual situation can be detected, and corrections can be made based on feedback, thereby improving the accuracy of the product model. The confidence level mark provides customers with a direct display of the model's accuracy, enabling customers to have greater trust in the performance and reality of the digital product model.

[0089] More specifically, based on the confidence level markers of each part in the product digital modeling, a representative analysis is conducted on the offline observation information. The performance of each offline observation information in different scenarios and its impact on digital modeling are analyzed to determine which observation information is highly representative. Based on the results of the representativeness analysis, several representative measured reference models are generated. Each reference model is based on the comprehensive performance of the customer's offline observation information and the product digital modeling, thereby optimizing the product modeling effect. All generated representative measured reference models are integrated to form a complete set of measured reference models. These reference models can serve as the basis for optimizing product digital modeling and can be used for further display, verification, or customer experience optimization. Through representativeness analysis, reference models under various different conditions can be accurately constructed. These models reflect the state and performance of the product in actual use. Through the collection of multiple measured reference models, it is possible to better address different customer needs and product display in different scenarios, increasing the applicability and flexibility of product digital modeling.

[0090] More specifically, the generated set of measured reference models for the target product is applied to a virtual display or simulation environment for customers to view and interact with. When customers view the target product through virtual reality, augmented reality, or other means, they can obtain a more realistic product display effect based on the measured reference model set, enhancing the accuracy and confidence of their purchase decisions. Through the reference model set based on actual usage feedback, the product display is more realistic, and customers can obtain a feeling closer to actual use on the virtual platform. By displaying accurate measured reference models, customers can more clearly understand the performance of the product in actual use, thereby improving the market acceptance and customer trust of the product.

[0091] Understandably, through the above steps, merchants can use offline observation information provided by customers to verify and reconstruct digital models of products, thereby generating a more accurate set of measured reference models. This process can improve the realism, transparency, and interactivity of product displays, while enhancing customers' trust in products and confidence in their purchasing decisions. This combination of digital modeling and measured data helps merchants improve sales performance, while also enhancing customer satisfaction and loyalty.

[0092] Preferably, the step of performing a representative analysis on each of the offline observation information based on the confidence level of the markings at specific locations in the digital model of the product, in order to construct several representative measured reference models to jointly constitute a set of measured reference models for the target product, includes:

[0093] S231: Based on the confidence level of the markings at various specific locations in the digital modeling of the product, analyze the overall feedback value of each offline observation information, and divide each offline observation information into several information sets based on the analysis results;

[0094] S232: Encode the information features of the offline observation information in each information set, and perform information mapping and coding representativeness analysis on the information feature codes of each offline observation information in the same information set relative to the digital modeling of the commodity, so as to generate the coding representativeness index of each offline observation information in the same information set.

[0095] S233: Based on the coding representativeness index of each offline observation information in the same information set, extract and combine representative information of each offline observation information in the same information set to generate initial representative observation information of each information set. At the same time, perform representative difference analysis on the initial representative observation information of each information set as a supervision condition to reconstruct the initial representative observation information of each information set to obtain the final representative observation information of each information set.

[0096] S234: Reconstruct the digital model of the commodity based on each of the final representative observation information to generate a measured reference model corresponding to each of the final representative observation information, so as to jointly constitute the measured reference model set of the target commodity.

[0097] Specifically, the system uses confidence markers at various locations in the digital product modeling to understand which parts of the modeling data better match the actual usage of customers. Based on this confidence data, the system analyzes the overall feedback value of each offline observation piece of information. This means evaluating the impact and importance of each offline observation piece of information on improving product modeling, thus preparing for subsequent classification and optimization. Through confidence-based feedback value analysis, the system can identify which offline observation pieces of information are most valuable for reconstructing the model, thereby optimizing resource allocation, improving the accuracy of product modeling, ensuring that offline observation information is scientifically and reasonably evaluated, and making subsequent steps more accurate and efficient.

[0098] More specifically, based on the results of the overall feedback value analysis, the offline observation information is divided into several information sets. These information sets can be classified according to different characteristics (such as observation angle, product part, usage scenario, etc.). Dividing the offline observation information into multiple sets helps to clearly distinguish different categories of observation data in subsequent analysis, thereby improving data processing efficiency and accuracy.

[0099] More specifically, information feature encoding is performed on the offline observation information in each information set. The encoding process converts the observation information into digital form to facilitate subsequent processing and analysis. For the offline observation information in each information set, a mapping is performed relative to the digital modeling of the product. That is, the offline observation information is compared and mapped with the data in the digital modeling of the product to analyze the representativeness of the information. Information encoding ensures the standardization of data processing, so that offline observation information from different sources and in different formats can be processed uniformly to facilitate subsequent analysis. Through information mapping and representativeness analysis, the relationship between each piece of offline observation information and the product modeling can be quantified to ensure that subsequent steps can make reasonable judgments based on scientific data.

[0100] More specifically, based on the results of the coding representativeness analysis, a coding representativeness index is generated for each offline observation. This index is used to measure the representativeness of the information in its set for the digital modeling of commodities. By generating the coding representativeness index, a quantitative indicator can be provided for each piece of information, which is convenient for subsequent decision-making and screening. The coding representativeness index helps to screen out the observation information that has the greatest impact on modeling from massive amounts of data, thereby optimizing the modeling process.

[0101] More specifically, based on the representativeness index of the encoded offline observation information in the same information set, the most representative information is extracted and combined to form the initial representative observation information of the information set. A difference analysis is then performed on the initial representative observation information to analyze the differences between different information sets, ensuring that the representative observation information in each information set fully represents its characteristics. Then, using these differences as supervisory conditions, the representative observation information of each information set is reconstructed to obtain the final representative observation information for each information set. Through the extraction and combination of representative information, the most representative parts for product modeling can be selected from complex data, thereby improving the modeling quality. The difference analysis helps to discover the differences between different information sets, further optimizing the representative observation information of each set and ensuring the diversity and adaptability of product modeling.

[0102] More specifically, the digital model of the product is reconstructed based on the final representative observation information in each information set. This step ensures that the digital model of the product better reflects the actual feedback and usage of customers. Through the reconstructed digital model of the product, a measured reference model corresponding to each final representative observation information is generated. These measured reference models will together constitute the measured reference model set of the target product. By combining actual customer usage data, the generated measured reference model can more realistically reflect the usage status of the product and improve the accuracy of product display. Different final representative observation information can cover multiple features of the product, thereby ensuring that the digital model of the product is more comprehensive and can meet the display needs of different customers and scenarios.

[0103] More specifically, through the above steps, multiple test reference models are ultimately generated. These models together constitute the test reference model set for the target product. Multiple test reference models can provide diverse product display methods, enabling customers to choose according to different scenarios and needs, thereby improving the user experience. The accuracy and richness of the test reference models enhance the transparency and realism of the product, which helps to increase customers' trust in the product and thus improve the purchase conversion rate.

[0104] Understandably, through the above steps, product digital modeling can be reconstructed based on actual customer usage data and generate multiple reference models based on measured data. These reference models can comprehensively reflect the performance of the product in actual use, improve the authenticity and accuracy of product display, and enhance customers' confidence in purchasing decisions. This modeling method based on customer feedback provides merchants with a more accurate and reliable product display tool, which helps to improve sales performance.

[0105] Preferably, the step of generating a temporary community scene for the customer's virtual unit by performing parallel matching of the digital models of the goods displayed by each merchant's virtual unit in the virtual community and the actual test reference model set includes:

[0106] S31: Take the digital model of the goods displayed by the merchant virtual unit that the customer virtual unit is viewing as the benchmark object, and simultaneously obtain the product reference service opened by the user on the community platform and the advertising recommendation service opened by each merchant virtual unit.

[0107] S32: Customize personalized services for the customer's virtual unit based on the product reference service to generate the basic framework of a temporary community scene;

[0108] S33: Based on the basic framework of the temporary community scenario, perform merchant qualification analysis on each merchant virtual unit to obtain the first priority of each merchant virtual unit corresponding to the basic framework of the temporary community scenario;

[0109] S34: Based on the first priority of each merchant's virtual unit, user profiles are inferred for customer virtual units. At the same time, based on the pre-built knowledge graph, interactive verification is performed on the user profiles inferred from each merchant's virtual units based on the first priority of each merchant's virtual units to generate the core profile of the user.

[0110] S35: Based on the advertising recommendation service of each merchant virtual unit, conduct an adaptive assessment of the advertising value of the core profile to generate an advertising bidding strategy for each merchant virtual unit in the temporary community scenario.

[0111] S36: The advertising bidding strategy of each merchant's virtual unit is simulated and analyzed by a multi-agent bidding simulation algorithm. The product digital model of the most prioritized merchant virtual units is selected as the extended object. The benchmark object and its actual reference model set, as well as each extended object and its actual reference model set, are substituted into the basic framework of the temporary community scene to construct a personalized temporary community scene for customer virtual units.

[0112] Specifically, the digital model of the goods displayed by the merchant virtual unit that the customer's virtual unit is viewing is used as the benchmark object. This benchmark object is the core reference for the interaction between the customer's virtual unit and the merchant virtual unit. At the same time, the system will synchronously obtain the product reference service opened by the customer's virtual unit on the community platform and the advertising recommendation service opened by each merchant virtual unit. The product reference service is used to recommend relevant products based on the customer's interests and needs, while the advertising recommendation service displays advertisements based on the needs of the merchants. By using the digital model of the goods that the customer is viewing as the benchmark, it is ensured that the temporary community scene generated later is highly matched with the customer's current interests. By synchronously obtaining the product reference service and the advertising recommendation service, the needs of customers and merchants can be understood more comprehensively, providing rich data support for the construction of personalized scenes.

[0113] More specifically, personalized services are customized for virtual units of customers based on product reference services. By using data such as customers' historical behavior and preferences, a basic framework for temporary community scenarios is generated. This framework defines the virtual community environment in which customers will participate, including the layout of product recommendations and the location of merchant advertisements. Customizing the scenario framework according to the specific needs of customers can improve the customer's user experience, increase the level of interaction and participation, and provide a clear scenario foundation based on customer needs for subsequent steps such as merchant qualification analysis and advertising.

[0114] More specifically, a qualification analysis is conducted on each merchant's virtual unit, analyzing factors such as the merchant's reputation, historical performance, product quality, and advertising effectiveness. This step aims to understand the overall strength of each merchant. Based on the results of the merchant qualification analysis, each merchant's virtual unit is assigned a corresponding priority, generating the first priority of the basic framework for each merchant's virtual unit in the temporary community scene. Merchant qualification analysis ensures that the merchants displayed on the platform have high reputation and product quality, avoiding low-quality merchants from affecting the customer experience. Determining the priority helps the platform decide which merchants can participate in advertising display first, which helps improve advertising effectiveness and enhance customer satisfaction.

[0115] More specifically, based on the first priority of merchant virtual units, user profiles are inferred for customer virtual units. By using customers' historical behavioral data (such as purchase records and browsing history), the potential interests, needs, and consumption preferences of customers are inferred. Based on a pre-built knowledge graph, the inferred user profiles are interactively verified. Through data interaction with merchant virtual units, the accuracy of the user profiles is verified, ensuring that the inferred results are consistent with actual needs. Through inference and interactive verification, accurate customer profiles are generated, making subsequent advertising recommendations and product displays more personalized and targeted. The application of knowledge graphs helps improve the accuracy of profile inference, and through comprehensive knowledge links, it ensures that the profiles better match the real needs of customers.

[0116] More specifically, based on the advertising recommendation services of each merchant's virtual unit, an adaptive assessment of the advertising value of user profiles is conducted. This step analyzes the matching degree between each merchant's advertising content and the customer's virtual unit profile, and evaluates the potential effect of advertising. By assessing the suitability between advertising and customer profiles, it ensures that advertising is not only accurate but also highly valuable, improving the ROI (Return on Investment) of advertising. Accurate advertising enhances the customer experience and avoids interference or churn caused by mismatch between advertising content and customer needs.

[0117] More specifically, based on the results of the adaptive assessment, an advertising bidding strategy for each merchant's virtual unit is generated. The bidding strategy determines the price, placement, and priority of the merchant's advertisements. By generating the bidding strategy, merchants can dynamically adjust their placement strategy according to the matching degree between their advertising content and customer profiles, thereby maximizing advertising effectiveness. The bidding mechanism ensures the fairness of advertising placement, allowing different merchants to adjust their placement strategies based on actual results, rather than solely relying on their budget.

[0118] More specifically, through a multi-agent bidding simulation algorithm, the platform simulates and analyzes the advertising bidding strategies of each merchant's virtual units. During the simulation, the platform uses a multi-agent model to simulate the bidding process between merchants and selects the optimal advertising plan for each merchant's virtual units. Through the multi-agent bidding simulation algorithm, the platform can identify the most valuable advertising strategies, helping to make the interaction between merchants and customers more precise. The simulation results help optimize advertising placement, ensuring the maximization of advertising effectiveness and the improvement of merchant revenue.

[0119] More specifically, based on the auction simulation results, the digital product models of several top-priority merchant virtual units are selected as extension objects. The selected product digital models and their actual test reference model sets are then substituted into the basic framework of the temporary community scene along with the benchmark object and its reference model set to construct a personalized temporary community scene for customer virtual units. By integrating customer profiles, merchant advertising strategies, and product modeling, the generated temporary community scene can highly match the individual needs of customers, enhance customer participation and satisfaction, and improve the interactivity of the virtual community, enabling customers to enjoy a more realistic and immersive shopping experience in the virtual environment.

[0120] Understandably, through this series of steps, the platform can build personalized virtual community scenarios based on customer needs and merchant advertising bidding. This process integrates multiple factors such as customer profiles, merchant qualifications, and advertising placement, ultimately achieving a highly customized virtual community shopping experience, which not only improves customer satisfaction but also optimizes the advertising effectiveness for merchants.

[0121] Preferably, the step of predicting the product information after logistics delivery for each merchant virtual unit in the temporary community scenario based on the registered information of the customer virtual unit in the virtual community, and generating a product prediction model for the customer virtual unit to purchase target products from each merchant virtual unit, includes:

[0122] S41: Generate corresponding delivery addresses based on the registration information of customer virtual units in the virtual community, generate shipping addresses based on the delivery addresses of each merchant virtual unit, and construct the product delivery route based on the delivery addresses;

[0123] S42: Retrieve the delivery path information corresponding to each measured reference model in the measured reference model set, and perform the closest delivery path matching analysis on the product delivery path according to the measured reference model set, so as to retrieve the corresponding measured reference model based on the matching structure as the standard version of the product prediction model for purchasing the target product in the virtual unit of the merchant.

[0124] S43: Based on the transportation path information corresponding to each measured reference model in the measured reference model set, analyze the influence factors of each element of the transportation path for each measured reference model in the measured reference model set, so as to generate an influence factor map of the target commodity in the transportation path.

[0125] S44: Based on the factor influence map, assign influence factors to each element of the commodity transportation path to obtain the path influence characteristics of the commodity transportation path, and perform a rationality assessment and adaptive correction on the commodity prediction model according to the path influence characteristics to generate an optimized version of the commodity prediction model.

[0126] Specifically, based on the registration information of customer virtual units in the virtual community, a corresponding delivery address is generated. This address is the final location where the customer receives the goods and may include the customer's personal information, delivery area, etc. According to the business location of each merchant virtual unit, a merchant's shipping address is generated. The shipping address and delivery address are combined to determine the logistics route of the goods. Combining the customer's delivery address and the merchant's shipping address, a delivery route for the goods is constructed. This route takes into account the distance, route, time, and possible delivery methods. Through accurate delivery and shipping address generation, personalized delivery routes can be provided to customers, ensuring that the logistics process meets customer needs. Constructing the delivery route provides a basic framework for subsequent route matching and predictive analysis.

[0127] More specifically, the measured reference model sets of each merchant's virtual unit are retrieved to obtain the path information of each product from the merchant's shipment to the customer's receipt. The path information includes the specific transportation route, delivery time, and possible traffic conditions. Matching analysis is performed between the product delivery path and the path information in each measured reference model set. By calculating the similarity of the paths, the closest delivery path is found. By analyzing the path matching results, the path that best matches the actual delivery situation of the customer and the merchant is determined. The measured reference model of this path is used as the standard version to generate a product prediction model for the customer to purchase goods in the merchant's virtual unit. The path matching analysis ensures the consistency between the product delivery path and the actual logistics environment, making subsequent product predictions more accurate. By determining the standard version of the product prediction model through the matching structure, more accurate product information and delivery expectations are provided to customers.

[0128] More specifically, the influencing factors of the transportation path information corresponding to each measured reference model in the measured reference model set are analyzed. These factors may include transportation mode, traffic conditions, logistics company efficiency, weather factors, etc. By analyzing the impact of various elements in the path, an influencing factor map of the factors influencing the goods in the transportation path is constructed. This map shows how each logistics link affects the efficiency and quality of goods transportation. By deeply analyzing the influencing factors of each logistics link, the potential problems and challenges in the goods distribution process can be understood more accurately. The generation of the influencing factor map provides a visual understanding of the logistics process, helping merchants and customers to anticipate and optimize the logistics process.

[0129] More specifically, based on the generated factor map, influencing factors are assigned to various elements in the goods delivery path. For example, some paths may require more time due to traffic congestion, or delivery efficiency may decrease due to weather conditions. These influencing factors are assigned to the delivery path to obtain the path's influence characteristics. These characteristics reflect the actual changes in the goods delivery process and are used to predict the arrival time and quality of goods. Based on the path influence characteristics, the rationality assessment and adaptive correction of the goods prediction model are performed. For example, if the influence factor prediction of a certain path indicates that it may lead to delays, the system will adjust the goods arrival time prediction accordingly to ensure that it is more in line with the actual situation. Through the assignment and correction of influencing factors, the goods prediction model can dynamically adapt to changes in actual logistics conditions, improve the accuracy of prediction, and the assessment and correction ensure that the final goods prediction model is more in line with the logistics characteristics of the real world, and can provide customers with a more reliable shopping experience.

[0130] More specifically, based on the rationality assessment and adaptive correction, an optimized version of the product prediction model is finally generated. This optimized version takes into account all influencing factors and actual conditions, and is the most accurate expectation model for customers after purchasing goods in the merchant's virtual unit. The optimized product prediction model can more accurately reflect the delivery time, quality and other related factors of the goods, helping customers make more informed shopping decisions. Customers can understand various information in the logistics and delivery process based on the optimized product prediction model, thereby enhancing the transparency and predictability of the shopping process and improving customer satisfaction.

[0131] Understandably, through the above steps, the platform can accurately predict product information based on the customer's virtual unit's delivery address, the merchant's virtual unit's shipping address, and the delivery path information in the actual test reference model set. This process uses technical means such as path matching, influencing factor analysis, and prediction model correction to ultimately generate an optimized version of the product prediction model, ensuring that customers can obtain accurate logistics information and improve their shopping experience and decision-making quality.

[0132] Preferably, the step of retrieving the transmission path information corresponding to each measured reference model in the measured reference model set includes:

[0133] S401: Retrieve the modeling information source for each measured reference model in the measured reference modeling set to obtain reference source information for constructing each measured reference model; wherein, the reference source information is the offline observation information corresponding to each measured reference model;

[0134] S402: Perform depth information positioning on each offline observation information in the reference source information to obtain the transport path information corresponding to each offline observation information, and perform information conversion on the transport path information corresponding to each offline observation information to generate the corresponding transport path map.

[0135] S403: Overlay the various transport path maps and analyze the overlaid transport path maps using a pre-trained graph neural network model to generate the most representative transport path information for the measured reference model.

[0136] Specifically, for each measured reference model, the modeling information sources are retrieved. These sources include historical logistics data, offline observation records, and real-time data of the delivery process. This information will serve as the basis for constructing the measured reference model, ensuring the accuracy of the model. Relevant data is extracted from various offline observation information, including the time, route, mode of transport, and temperature and humidity changes of goods. This data reflects the operational status of the actual logistics path. By retrieving and integrating real offline observation information, the constructed measured reference model is ensured to be closer to the actual logistics process. Through the retrieval of multi-source data, the transportation path of goods from merchants to customers can be fully reflected.

[0137] More specifically, deep information positioning is performed on each offline observation data in the reference source information to determine the specific logistics path of each piece of information, including the origin, waypoints, and destination. Through data association, geographic information system (GIS) technology, and other methods, detailed information related to the path is further extracted. Through deep positioning, the delivery path information corresponding to each offline observation information is obtained. This information includes the actual delivery route, delivery method, and transportation time of each link. Through deep information positioning, the logistics path can be accurately determined, improving the accuracy of path prediction. Multi-dimensional path features, such as path time, traffic conditions, and weather conditions, are extracted, which helps in the subsequent further analysis and optimization of the path.

[0138] More specifically, the obtained transportation route information undergoes information transformation, converting unstructured data (such as text and images) into analyzable structured data. The transformation process may include data cleaning, standardization, and uniform formatting to ensure the consistency and usability of the route data. The transformed route information is then used to construct a transportation route map according to dimensions such as time sequence and spatial location. This map displays the route distribution, changing trends, and interrelationships between various links in the entire logistics process. Through information transformation, complex route information can be simplified and standardized, facilitating subsequent model analysis. The route map provides an effective tool for data visualization, contributing to an intuitive understanding and analysis of logistics routes.

[0139] More specifically, multiple generated transport path maps are overlaid. During the overlay process, data from different path maps are weighted and merged based on factors such as time, space, and importance, ultimately resulting in a comprehensive transport path map. A pre-trained graph neural network (GNN) model is then used to analyze the overlaid transport path map. The GNN can capture nodes, edges, and their relationships in the path map, extract hidden patterns and features through graph convolution operations, and extract the most representative path information through graph neural network analysis. This includes core path features, optimized transport routes, and potential bottlenecks. Finally, the most realistic measured reference model of transport path information is generated. Graph neural networks can handle complex path data, optimize path prediction through deep learning algorithms, and enhance the model's expressive power. Through GNN analysis, more accurate path predictions can be generated, providing optimized path solutions for merchants, logistics companies, and customers, thereby improving logistics efficiency and service quality.

[0140] More specifically, based on the analysis results of graph neural networks, the path information is optimized to generate the final version of the delivery path information. This version of information comprehensively considers various influencing factors, such as traffic flow, weather, and delivery time, to ensure optimal path. The final generated path information will be the most representative and optimal logistics route, which can significantly improve the timeliness, accuracy, and cost-effectiveness of goods delivery. The optimized path information provides more accurate data support for goods prediction modeling and enhances the customer's shopping experience.

[0141] Understandably, through the above steps, based on the measured reference modeling set and its corresponding transportation path information, logistics path data can be accurately retrieved, transformed, analyzed, and optimized. By performing in-depth analysis of the superimposed path map using graph neural networks, the most representative path information can be generated, providing accurate and optimized path data for commodity prediction modeling. This process will greatly improve the predictive capability of the logistics process and the efficiency of actual operations.

[0142] Reference Figure 2 As shown, in a second aspect, the present invention provides a digital advertising system based on a virtual community, used to implement the digital advertising method based on a virtual community as described in any one of the first aspects, comprising:

[0143] The product display module is used to display the digital model of the target product through the virtual units of merchants in the virtual community, so that customers can observe the digital model of the target product from multiple dimensions through their virtual units;

[0144] The reference modeling module is used to retrieve offline observation information from the corresponding virtual customer unit based on the transaction records of the target product, and reconstruct the digital model of the product based on the offline observation information to generate a set of measured reference models of the target product.

[0145] The personalized scene module is used to generate temporary community scenes for the customer's virtual unit by performing parallel matching of digital models of goods displayed by each merchant's virtual unit in the virtual community and a set of actual test reference models;

[0146] The information prediction module is used to predict the product information after logistics delivery based on the registered information of the customer's virtual unit in the virtual community and the measured reference model set of each merchant's virtual unit in the temporary community scenario, so as to generate a product prediction model for the customer's virtual unit to purchase target products from each merchant's virtual unit.

[0147] In this embodiment, the specific implementation of each module in the above system embodiment is described in the above method embodiment, and will not be repeated here.

[0148] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A digital advertising method based on virtual communities, characterized in that, include: The virtual community allows merchants to showcase digital models of target products, enabling customers to observe these models from multiple dimensions using their own virtual units. Based on the transaction records of the target product, offline observation information is retrieved from the corresponding virtual customer unit, and the digital model of the product is reconstructed based on the offline observation information to generate a set of measured reference models for the target product. By performing parallel matching of the digital models of the products displayed by each merchant's virtual unit in the virtual community and the actual test reference model set, a temporary community scene is generated for the customer's virtual unit in a personalized way. Based on the registration information of the customer's virtual unit in the virtual community, the product information of each merchant's virtual unit in the temporary community scene after logistics transportation is predicted, so as to generate a product prediction model for the customer's virtual unit to purchase target products in each merchant's virtual unit. The steps of retrieving offline observation information from the corresponding virtual customer unit based on the transaction records of the target product, and reconstructing the digital model of the product based on the offline observation information to generate a measured reference model set for the target product include: Based on the offline data reference service activated by the customer's virtual unit on the community platform, the transaction records of the target product are located and searched to determine several offline observation information that conforms to the customer's virtual unit's offline data reference service; wherein, the offline observation information is the information collected by the customer's virtual unit after purchasing the target product by observing the target product from multiple angles and in multiple ways; The digital model of the target product is used as the verification object. Data verification processing is performed on the verification object based on the offline observation information. Based on the results of the data verification processing, the confidence level of the digital model of the product is marked at each specific location. Based on the confidence level of the markings at various specific locations in the digital model of the product, a representative analysis is performed on each of the offline observation information to construct several representative measured reference models, which together constitute the measured reference model set of the target product.

2. The digital advertising method based on virtual communities as described in claim 1, characterized in that, The steps involved in showcasing digital product models of target goods through virtual merchant units within a virtual community, allowing customers to observe these models from multiple dimensions using their virtual units, include: Based on the merchant information registered by the merchant on the community platform, a virtual merchant unit is constructed for the merchant. At the same time, based on the advertising service contract opened by the merchant on the community platform, information scheduling is performed on the virtual merchant unit to deploy the virtual merchant unit in several designated virtual communities on the community platform. The merchant's virtual unit uploads multi-angle observation information of the goods to be sold, and the NeRF deep learning model deployed on the community platform performs 3D digital modeling of the goods to be sold based on the multi-angle observation information to generate a basic version of the goods digital model. By uploading several extended observation dimensions of the goods to be sold through the merchant's virtual unit, the basic version of the digital model of the goods is optimized in multiple dimensions based on the extended observation dimensions to generate the completed version of the digital model of the goods. The system obtains the interaction behavior instructions of the customer virtual unit in the merchant virtual unit, selects the specified goods to be sold in the merchant virtual unit as the target goods according to the interaction behavior instructions, and retrieves the digital model of the target goods and deploys it on the pre-built multi-dimensional observation and perception framework. The multi-dimensional observation and perception framework is used to obtain the perception instructions of the customer's virtual unit, and the digital model of the product is displayed in a corresponding form according to the perception instructions.

3. The digital advertising method based on virtual communities as described in claim 2, characterized in that, The steps of uploading several extended observation dimensions of the goods to be sold through the merchant's virtual unit, and optimizing the basic version of the digital model of the goods in multiple dimensions based on the extended observation dimensions to generate the final version of the digital model of the goods include: The extended observation dimensions of the goods uploaded by merchants' virtual units include morphological change dimension information, scene effect dimension information, internal disassembly dimension information, and time influence dimension information. By extending the observation mechanism of corresponding morphological change dimension information, scene effect dimension information, internal disassembly dimension information, and time influence dimension information through the digital modeling of goods, a complete version of the digital model of goods is generated. Based on the dimensional information of morphological changes, the basic version of the digital model of the product is assigned variable parameters and material elastic coefficients so that the digital model of the product has a morphological change observation mechanism. Based on the scene effect dimension information, the basic version of the product digital model is arranged in several types of scene modeling to generate product digital models arranged in various scene modeling. Ray tracing and physics collision simulations are performed on the digital product models deployed in various scene modeling to generate scene effect information of various types of digital product models. Based on the scene effect information of each type, the basic version of the digital product model is pre-retrieved to enable the digital product model to have a scene effect observation mechanism. The basic version of the digital modeling of the product is decomposed into modeling units based on the internal decomposition dimension information, and internal observation information is assigned to the decomposed modeling units so that the digital modeling of the product has an internal decomposition observation mechanism. Based on the time influence dimension information, a time decay simulation is performed on the basic version of the digital product model to generate a dynamic mask based on time change parameters and usage frequency parameters. By superimposing the dynamic mask on the digital product model, the digital product model has a time influence observation mechanism that displays the corresponding time decay simulation effect according to the input time change parameters and usage frequency parameters.

4. The digital advertising method based on virtual communities as described in claim 3, characterized in that, Based on the confidence level of the markings at specific locations in the digital model of the product, the representativeness analysis of each offline observation information is performed to construct several representative measured reference models, which together constitute the measured reference model set of the target product. The steps include: Based on the confidence level of the markings at various specific locations in the digital modeling of the product, an overall feedback value analysis is performed on each of the offline observation information, and the offline observation information is divided into several information sets based on the analysis results. Information feature encoding is performed on the offline observation information in each information set, and information mapping and encoding representativeness analysis are performed on the information feature encoding of each offline observation information in the same information set relative to the digital modeling of the commodity, so as to generate the encoding representativeness index of each offline observation information in the same information set. Based on the representativeness index of the coding of each offline observation information in the same information set, representative information of each offline observation information in the same information set is extracted and combined to generate initial representative observation information of each information set. At the same time, the representativeness difference of the initial representative observation information of each information set is analyzed, which is used as a supervision condition to reconstruct the initial representative observation information of each information set to obtain the final representative observation information of each information set. The digital model of the commodity is reconstructed based on each of the final representative observation information to generate a measured reference model corresponding to each of the final representative observation information, so as to jointly constitute the measured reference model set of the target commodity.

5. The digital advertising method based on virtual communities as described in claim 1, characterized in that, The steps for generating a personalized temporary community scene for the customer's virtual unit by performing parallel matching of the digital models of the goods displayed by each merchant's virtual unit in the virtual community and the actual test reference model set include: The digital model of the goods displayed by the merchant virtual unit that the customer virtual unit is viewing is used as the benchmark object, and the product reference service opened by the user on the community platform and the advertising recommendation service opened by each merchant virtual unit are obtained simultaneously. Personalized service customization is performed for virtual units of customers based on the product reference service to generate the basic framework of temporary community scenarios; Based on the basic framework of the temporary community scenario, the merchant qualifications of each virtual merchant unit are analyzed to obtain the first priority of each virtual merchant unit corresponding to the basic framework of the temporary community scenario. Based on the first priority of each merchant's virtual unit, user profiles are inferred for customer virtual units. At the same time, based on the pre-built knowledge graph, interactive verification is performed on the user profiles inferred from each merchant's virtual units based on the first priority of each merchant's virtual units, so as to generate the core profile of the user. Based on the advertising recommendation services of each merchant's virtual unit, an adaptive assessment of the advertising value of the core profile is conducted to generate an advertising bidding strategy for each merchant's virtual unit in the temporary community scenario. The advertising bidding strategies of various merchants' virtual units are simulated and analyzed using a multi-agent bidding simulation algorithm. The product digital modeling of several merchants' virtual units with the highest priority is selected as the extended objects. The benchmark objects and their actual test reference modeling sets, as well as each extended object and its actual test reference modeling set, are then substituted into the basic framework of the temporary community scene to construct a personalized temporary community scene for customer virtual units.

6. The digital advertising method based on virtual communities as described in claim 1, characterized in that, Based on the registration information of customer virtual units in the virtual community, the steps of predicting the product information after logistics delivery for the measured reference model set of each merchant virtual unit in the temporary community scenario, in order to generate a product prediction model for customer virtual units to purchase target products from each merchant virtual unit, include: The system generates corresponding delivery addresses based on the registration information of customers' virtual units in the virtual community, generates shipping addresses based on the delivery addresses of each merchant's virtual units, and constructs a goods delivery route based on the delivery addresses. The transport path information corresponding to each measured reference model is retrieved from the measured reference model set, and the closest transport path is matched and analyzed according to the measured reference model set. The measured reference model retrieved based on the matching structure is used as the standard version of the product prediction model for purchasing target products in the virtual unit of the merchant. Based on the transportation path information corresponding to each measured reference model in the measured reference modeling set, the influence factors of each element of the transportation path of each measured reference model in the measured reference modeling set are analyzed to generate the element influence factor map of the target commodity in the transportation path. Based on the factor influence map, the influence factors of each element are assigned to the commodity transportation path to obtain the path influence characteristics of the commodity transportation path. Based on the path influence characteristics, the rationality evaluation and adaptive correction of the commodity prediction model are carried out to generate an optimized version of the commodity prediction model.

7. The digital advertising method based on virtual communities as described in claim 6, characterized in that, The steps for retrieving the transport path information corresponding to each measured reference model in the measured reference model set include: The modeling information source of each measured reference model in the measured reference modeling set is retrieved to obtain the reference source information used to construct each measured reference model; wherein, the reference source information is the offline observation information corresponding to each measured reference model; The depth information of each offline observation information in the reference source information is located to obtain the transport path information corresponding to each offline observation information, and the transport path information corresponding to each offline observation information is transformed to generate the corresponding transport path map. The various transport path maps are overlaid, and the overlaid transport path maps are analyzed by a pre-trained graph neural network model to generate the most representative transport path information for the measured reference model.

8. A digital advertising system based on a virtual community, characterized in that, A digital advertising method based on a virtual community as described in any one of claims 1-7 includes: The product display module is used to display the digital model of the target product through the virtual units of merchants in the virtual community, so that customers can observe the digital model of the target product from multiple dimensions through their virtual units; The reference modeling module is used to retrieve offline observation information from the corresponding virtual customer unit based on the transaction records of the target product, and reconstruct the digital model of the product based on the offline observation information to generate a set of measured reference models of the target product. The personalized scene module is used to generate temporary community scenes for the customer's virtual unit by performing parallel matching of digital models of goods displayed by each merchant's virtual unit in the virtual community and a set of actual test reference models; The information prediction module is used to predict the product information after logistics delivery based on the registered information of the customer's virtual unit in the virtual community and the measured reference model set of each merchant's virtual unit in the temporary community scenario, so as to generate a product prediction model for the customer's virtual unit to purchase target products from each merchant's virtual unit.

Images