Digital creative design dynamic optimization and inventory collaborative management method and device based on sales big data, and storage medium

Abstract

The application relates to the technical field of creative design, and particularly discloses a digital creative design dynamic optimization and inventory collaborative management method and device based on sales big data and a storage medium, which comprises the following steps: acquiring full-channel retail big data, preprocessing the full-channel retail big data based on user interaction to obtain preprocessed data, performing atomic-level design element decomposition on a digital creative design product to obtain corresponding design element features, determining the sales contribution of each design element feature based on the preprocessed data, determining the corresponding heat weight based on the sales contribution, performing a creative design dynamic optimization operation based on the heat weight to generate an optimized design, performing sales prediction on the optimized design to generate a predicted sales fluctuation, and generating a corresponding inventory management result based on the predicted sales fluctuation. The existing optimization method of the digital creative design product is improved based on sales data, the optimization design accuracy of the creative design product is improved, and the inventory of an enterprise is reduced.

Description

Technical Field

[0001] This invention relates to the field of cultural and creative design technology, specifically to a method, device, and storage medium for dynamic optimization of digital cultural and creative design and collaborative inventory management based on sales big data. Background Technology

[0002] With the deep integration of big data technology and the digital retail industry, digital cultural and creative products are playing an increasingly significant role in market expansion and brand building, serving as a core link between cultural soft power and commercial value. As the digital cultural and creative industry continues to develop, people are placing higher demands on sales management solutions for digital cultural and creative products.

[0003] The collaborative efficiency of the digital cultural and creative industry relies on the in-depth mining and real-time response to massive market interaction data. By accurately grasping consumer preferences and market trends, it is possible to effectively drive the rapid iteration and commercial transformation of cultural and creative products. However, in the current technology, cultural and creative design and sales management are often disconnected. Designers cannot obtain popular cultural and creative designs in a timely manner, while sales personnel cannot quickly obtain cultural and creative products that meet market demands.

[0004] Traditional sales systems focus on macro-level sales statistics, neglecting to incorporate feedback from outstanding cultural and creative designs in the market. This leads to blind optimization and difficulty in accurately measuring return on investment. Furthermore, visual strategy updates heavily rely on human experience, lacking automated triggering mechanisms based on data metrics. This results in lengthy design adjustment cycles and an inability to keep up with rapidly changing market trends. In addition, the supply chain system, lacking deep integration of design feedback logic, cannot predict sales fluctuations based on changes in solutions, leading to low omnichannel turnover efficiency and uncontrollable risks of inventory backlog or stockouts. Summary of the Invention

[0005] To overcome the aforementioned technical problems in the prior art, embodiments of the present invention provide a method, apparatus, and storage medium for dynamic optimization of digital cultural and creative design and collaborative inventory management based on sales big data. By improving existing optimization methods for digital cultural and creative design products based on sales data, the invention effectively enhances the accuracy of optimized design of cultural and creative design products and reduces enterprise inventory.

[0006] To achieve the above objectives, embodiments of the present invention provide a method for dynamic optimization of digital cultural and creative design and collaborative inventory management based on sales big data. The method includes: acquiring omnichannel retail big data; preprocessing the omnichannel retail big data based on user interaction to obtain preprocessed data; performing atomic-level design element decomposition on digital cultural and creative design products to obtain corresponding design element features; determining the sales contribution of each design element feature based on the preprocessed data, and determining the corresponding popularity weight based on the sales contribution; performing dynamic optimization of cultural and creative design based on the popularity weight to generate an optimized design; performing sales forecasting on the optimized design to generate predicted sales fluctuations; and generating corresponding inventory management results based on the predicted sales fluctuations.

[0007] Preferably, the preprocessing of the omnichannel retail big data based on user interaction to obtain preprocessed data includes: determining a standard parameter table; performing interactive analysis on user interaction parameters across different channel platforms and online / offline channels to generate a common set of interactive parameters; establishing a similar indicator mapping table based on the standard parameter table and the common set of interactive parameters; performing preliminary behavioral classification on the omnichannel retail big data based on the similar indicator mapping table to obtain behaviorally classified data; performing time-series analysis on the behaviorally classified data to generate user behavior analysis results; performing transaction contribution analysis on the user behavior analysis results based on a Hidden Markov Model to generate a transaction contribution probability corresponding to each user behavior; and generating preprocessed data based on the behaviorally classified data and the transaction contribution probabilities.

[0008] Preferably, the step of performing atomic-level design element decomposition on the digital cultural and creative design product to obtain corresponding design element features includes: performing image data processing on the digital cultural and creative design product based on a deep convolutional neural network to obtain a corresponding high-dimensional matrix; performing atomic-level visual feature extraction on the high-dimensional matrix to obtain corresponding atomic visual features; performing logical association analysis on the atomic visual features to obtain logical association feature groups; and generating design element features based on the atomic visual features and the logical association feature groups.

[0009] Preferably, determining the sales contribution of each design element feature based on the preprocessed data, and determining the corresponding popularity weight based on the sales contribution, includes: performing normalization processing on the preprocessed data to obtain normalized sales data; determining the intensity value of each design element feature in the preprocessed data based on the normalized sales data; determining the conversion rate weighted value of each sales sample and the popularity duration of each design element feature in the market based on the normalized sales data; and analyzing the normalized sales data, the intensity value, the conversion rate weighted value, and the popularity duration based on the gradient boosting decision tree algorithm to generate a popularity weight corresponding to each design element feature.

[0010] Preferably, the heat weight is characterized as follows:

[0011]

[0012] in, For the first The heat weight of each design element feature Let be the intensity value of the i-th design element feature in the j-th sales sample. This represents the normalized sales volume corresponding to the sample. The conversion rate weighted by the sample. To prevent extremely small constants with a denominator of zero, The duration of the popularity of a design feature in the market.

[0013] Preferably, the step of performing dynamic optimization of cultural and creative design based on the popularity weight to generate an optimized design includes: determining a first trigger threshold corresponding to market hotspots and a second trigger threshold corresponding to aesthetic fatigue, wherein the first trigger threshold is less than the second trigger threshold; determining the weight increase trend and weight decrease trend of each design element feature based on the popularity weight; when any design element feature is in the weight increase trend and the corresponding popularity weight reaches the first trigger threshold, performing dynamic optimization of cultural and creative design based on the corresponding design element feature to generate an optimized design; when any design element feature is in the weight decrease trend and the corresponding popularity weight reaches the second trigger threshold, removing the corresponding design element feature from the dynamic optimization of cultural and creative design.

[0014] Preferably, the step of performing dynamic optimization of cultural and creative design based on the corresponding design element features to generate an optimized design includes: determining the pixels to be optimized and the parameters to be optimized based on the corresponding design element features; identifying the brand features of the digital cultural and creative design product to obtain the product brand features; adjusting the pixels to be optimized based on the product brand features to obtain the adjusted pixels; and performing dynamic optimization of cultural and creative design based on the adjusted pixels and the parameters to be optimized to generate the optimized design.

[0015] Preferably, the step of performing sales forecasting on the optimized design to generate predicted sales fluctuations includes: determining an initial sales forecast based on the optimized design; generating a design sensitivity adjustment coefficient and a weight change based on the popularity weight; obtaining the marginal contribution rate of design element features to sales and external influencing factors; generating predicted sales fluctuations based on the initial sales forecast, the design sensitivity adjustment coefficient, the weight change, the marginal contribution rate, and the external influencing factors, wherein the predicted sales fluctuations are characterized as follows:

[0016]

[0017] in, To predict sales fluctuations, For initial sales forecast, To design the sensitive adjustment coefficient, For design element characteristics The change in the weight of popularity The marginal contribution rate of this design element feature to sales. As an external influencing factor, This is a random error correction term.

[0018] Accordingly, the present invention also provides a device for dynamic optimization of digital cultural and creative design and collaborative inventory management based on sales big data feedback. The device includes: a data processing module for acquiring omnichannel retail big data and preprocessing the omnichannel retail big data based on user interaction to obtain preprocessed data; a decomposition module for performing atomic-level design element decomposition on digital cultural and creative design products to obtain corresponding design element features; a weight calculation module for determining the sales contribution of each design element feature based on the preprocessed data and determining the corresponding popularity weight based on the sales contribution; a design module for performing dynamic optimization of cultural and creative design based on the popularity weight to generate an optimized design; a prediction module for performing sales prediction on the optimized design to generate predicted sales fluctuations; and a management module for generating corresponding inventory management results based on the predicted sales fluctuations.

[0019] On the other hand, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method provided in the embodiments of the present invention.

[0020] The present invention has at least the following technical effects through the technical solution provided by the present invention:

[0021] This invention achieves a quantitative correlation between atomic-level design features and sales conversion. Through a design element popularity weight calculation model, it overcomes the limitations of traditional cultural and creative industries that rely on manual experience for design optimization. The system can accurately locate specific visual features affecting market conversion, breaking down macro sales volume into micro-design parameters. This atomic-level feature engineering process provides a scientific evaluation basis for cultural and creative products, making design iteration more targeted, significantly improving the return on investment in cultural and creative products, and greatly reducing the misjudgment rate.

[0022] At the same time, a data-driven digital creative automatic production closed loop has been built. By setting a preset popularity weight threshold, the system can trigger the parametric engine to correct the visual scheme as soon as market preferences fluctuate. This real-time response mode avoids the loss of the golden period of marketing caused by slow information feedback and long manual modification cycle in the traditional design process, and greatly enhances the core competitiveness of cultural and creative products in the market environment.

[0023] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0024] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings:

[0025] Figure 1 This is a flowchart illustrating the specific implementation of the method for dynamic optimization of digital cultural and creative design and collaborative inventory management based on sales big data provided in this embodiment of the invention.

[0026] Figure 2 This is a schematic diagram of the structure of the digital cultural and creative design dynamic optimization and inventory collaborative management device based on sales big data provided in an embodiment of the present invention. Detailed Implementation

[0027] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0028] In this invention, the terms "system" and "network" are used interchangeably. "Multiple" refers to two or more; therefore, in this invention, "multiple" can also be understood as "at least two." "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / ", unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship. Furthermore, it should be understood that in the description of this invention, words such as "first" and "second" are used only for descriptive purposes and should not be construed as indicating or implying relative importance or order.

[0029] To address the technical problems existing in the prior art, the method provided in this invention is applied to a cloud server cluster with high-performance computing capabilities, employing a highly decoupled microservice architecture. It combines an omnichannel retail monitoring module, a design element popularity weight calculation model, an automatic triggering mechanism for digital creative module parameters, and a supply chain inventory collaborative control engine to construct a closed-loop system encompassing sales data collection, design feature quantification, automatic creative iteration, and agile supply chain response. The modules communicate in real-time through standardized data interface protocols to ensure data timeliness and reliability.

[0030] Please see Figure 1 This invention provides a method for dynamic optimization of digital cultural and creative design and collaborative inventory management based on sales big data. The method includes:

[0031] S10, acquire omnichannel retail big data, preprocess the omnichannel retail big data based on user interaction, and obtain preprocessed data;

[0032] S20, perform atomic-level design element decomposition on the preprocessed data to obtain the corresponding design element features;

[0033] S30, determine the sales contribution of each design element feature, and determine the corresponding popularity weight based on the sales contribution;

[0034] S40, Perform dynamic optimization of cultural and creative design based on the popularity weight to generate the optimized design;

[0035] S50, Perform sales forecasting on the optimized design to generate predicted sales fluctuations;

[0036] S60, Generate corresponding inventory management results based on the predicted sales fluctuations.

[0037] In one possible implementation, the system first acquires omnichannel retail big data. This can be achieved by concurrently accessing raw data from multiple mainstream e-commerce platforms (such as Tmall, JD.com, and Douyin e-commerce) and multiple offline retail terminals (such as POS machines and smart store sensors) using a distributed web crawler engine and API integration program. During the data preprocessing stage, the system uses hash verification and logical association algorithms to identify and remove duplicate transaction records, malicious order-boosting interaction data with abnormal fluctuations, and invalid click traffic generated by web crawling or testing. This omnichannel retail big data should include at least click-through rate, conversion rate, and user behavior data to facilitate subsequent analysis such as sales forecasting and customer profiling.

[0038] Then, the omnichannel retail big data is preprocessed according to the principle of user interaction to obtain preprocessed data. Existing data preprocessing methods mainly deal with noise, errors, and duplicates in the original data. However, in practical applications, the results of the above processing cannot reflect the actual purchasing intentions and aesthetic preferences of customers. Therefore, they do not contribute much to the sales of digital cultural and creative design products and may even cause some interference, failing to meet actual needs.

[0039] For example, even if data on conversion rates, user behavior, and click-through rates are predetermined, different channels and platforms, as well as online and offline channels, may have different definitions for these parameters. This can lead to a chaotic collection of data with many duplicates, making it impossible to meet actual needs.

[0040] In this embodiment of the invention, the preprocessing of the omnichannel retail big data based on user interaction to obtain preprocessed data includes: determining a standard parameter table; performing interactive analysis on user interaction parameters across different channel platforms and online / offline channels to generate a common set of interactive parameters; establishing a similar indicator mapping table based on the standard parameter table and the common set of interactive parameters; performing preliminary behavioral classification on the omnichannel retail big data based on the similar indicator mapping table to obtain behaviorally classified data; performing time-series analysis based on the behaviorally classified data to generate user behavior analysis results; performing transaction contribution analysis based on the user behavior analysis results to generate a transaction contribution probability corresponding to each user behavior; and generating preprocessed data based on the behaviorally classified data and the transaction contribution probabilities.

[0041] In one possible implementation, a standard parameter table is first established, defining the standard parameters to be acquired and their names, such as click-through rate (CTR) and conversion rate (CVR). Then, by analyzing the user interaction parameters defined across different channels and platforms, both online and offline, similar metrics are mapped between the same parameters from different channels and platforms and the standard parameters. This forms a standardized metric mapping, facilitating the subsequent development of a standardized database. This database includes data categorized into various behaviors. For example, "adding to cart" on e-commerce platforms and "time spent picking up items" in offline stores can be mapped to similar purchase intention weights, thereby establishing unified dimensions for CTR, CVR, and user behavior trajectory evaluation.

[0042] After establishing a standardized database, time-series analysis is performed. For example, for online platforms, in-depth analysis is conducted on user page dwell time, scroll depth, and click sequences to generate user behavior analysis results for browsing and purchasing. Then, Hidden Markov Models or Recurrent Neural Networks are used to calculate the probability of each interaction node's contribution to the final transaction (such as viewing the details page, clicking on color options, or viewing the comments section). Specifically, clicks, favorites, adding to cart, and placing orders are set as observation sequences, and the strength of users' potential purchase motives is set as hidden states. The Baum-Welch algorithm is used to estimate parameters of omnichannel retail big data and calculate the state transition probability of each interaction node in the hidden path to generate the transaction contribution probability corresponding to each user behavior. This allows sales results to be accurately traced back to specific visual interaction links, enabling targeted analysis of the original sales data and ensuring that the data in the database objectively and purely reflects consumers' aesthetic preferences and purchasing intentions for cultural and creative products.

[0043] Furthermore, for preprocessed data, anomaly detection algorithms can be used to remove noisy data caused by system latency, network congestion, or user misoperation, thereby improving data accuracy, avoiding data interference, and enhancing the accuracy of subsequent optimization design and inventory forecasting.

[0044] In this embodiment of the invention, by further focusing on user interactions during the sales process of digital cultural and creative design products on the basis of traditional data cleaning, the front end and back end are connected, providing more accurate and reliable guidance for the optimization design of digital cultural and creative design products and improving accuracy.

[0045] After preprocessing the raw sales big data, it is decomposed using feature engineering. Existing design decomposition schemes often only analyze the design scheme from the module and organizational levels. However, in actual application, simply referring to excellent design modules on the market to guide the optimization design of digital cultural and creative products (such as incorporating a certain excellent design into the product) not only fails to achieve the expected sales growth effect, but also achieves poor sales results in some optimized digital cultural and creative products.

[0046] In this embodiment of the invention, the step of performing atomic-level design element decomposition on digital cultural and creative design products to obtain corresponding design element features includes: performing image data processing on the digital cultural and creative design products based on a deep convolutional neural network to obtain a corresponding high-dimensional matrix; performing atomic-level visual feature extraction on the high-dimensional matrix to obtain corresponding atomic visual features; acquiring historical sales data within a preset time period; analyzing the atomic visual features and the historical sales data based on a gradient boosting decision tree to generate a popularity weight coefficient corresponding to each atomic visual feature; and generating design element features based on the atomic visual features and the popularity weight coefficient.

[0047] In one possible implementation, machine learning algorithms are used to decompose the cultural and creative scheme into feature engineering components, and regression analysis is used to calculate the weight of each atomic-level design element's contribution to sales conversion. Specifically, firstly, a deep convolutional neural network (CNN) is used to extract the visual features of the digital cultural and creative design products, transforming the image data of the digital cultural and creative design products into a high-dimensional matrix. Then, multiple convolutional layers are used to perform atomic-level visual feature extraction on the high-dimensional matrix to extract features such as the hue distribution of the main color tone (based on HSV space component statistics), color contrast (based on spatial frequency contrast calculation), composition center position (based on salient object detection algorithm), line features of the IP image (such as curvature distribution and edge density), and complexity of auxiliary border elements (based on fractal dimension calculation), thereby forming atomic visual features.

[0048] In practical applications, directly using atomic visual features for optimization design, especially when using artificial intelligence for optimization design, may produce a large number of digital cultural and creative design products that are theoretically good but actually rather "strange", which will greatly reduce the user experience.

[0049] Therefore, further logical association analysis is performed on the aforementioned atomic visual features to identify and extract popular feature groups from individual atomic visual features. This allows for the extraction of "popular elements" from a holistic perspective. Specifically, by calculating the Euclidean distance between atomic visual features in spatial coordinates and combining it with the consistency of semantic labels, when the distance is less than a preset spatial threshold and the semantic similarity is higher than 0.8, the corresponding atomic visual features are determined to have logical associations and are merged to generate logically associated feature groups. Simultaneously, ungrouped individual atomic visual features are retained to generate the final design element features. In subsequent optimization design of digital cultural and creative products using artificial intelligence (e.g., AIGC), this significantly reduces the proportion of "weird" products generated by the "random" combination of original atomic visual features, while greatly improving the accuracy of identifying current market hotspots and popular elements, thereby increasing the popularity of digital cultural and creative design products and meeting the actual needs of enterprises.

[0050] To enhance the promotional effect on the actual sales market, a quantitative correlation logic is established between design elements and market conversion indicators. In this embodiment of the invention, determining the sales contribution of each design element feature based on the preprocessed data, and determining the corresponding popularity weight based on the sales contribution, includes: performing normalization processing on the preprocessed data to obtain normalized sales data; determining the intensity value of each design element feature in the preprocessed data based on the normalized sales data; determining the conversion rate weighted value of each sales sample and the popularity duration of each design element feature in the market based on the normalized sales data; and analyzing the normalized sales data, the intensity value, the conversion rate weighted value, and the popularity duration based on the gradient boosting decision tree algorithm to generate a popularity weight corresponding to each design element feature.

[0051] Furthermore, in this embodiment of the invention, the heat weight is characterized as follows:

[0052]

[0053] in, For the first The popularity weight of each design element Let be the intensity value of the i-th feature in the j-th sales sample. This represents the normalized sales volume corresponding to the sample. The conversion rate weighted by the sample. To prevent extremely small constants with a denominator of zero, The duration of the popularity of a design feature in the market.

[0054] In one possible implementation, the preprocessed data is first normalized to facilitate subsequent sales contribution analysis on a uniform scale. Then, from the normalized sales data, the intensity value of each design element feature is determined based on its proportion. Further, the conversion data for each sales sample is statistically analyzed from the normalized sales data, and a conversion rate weighted value for each sales sample is determined according to set weights, as well as the duration of popularity of each design element feature in the market.

[0055] At this point, the design element features and historical sales data mentioned above are transformed into multi-dimensional feature vectors, which are used as input variables for the regression analysis algorithm. Then, based on the gradient boosting decision tree algorithm, the normalized sales data, intensity value, conversion rate weighted value, and popularity duration are analyzed to calculate the heat weight of each feature vector in the contribution to sales conversion, so as to characterize the degree of influence of specific design elements (such as the application ratio of "vermilion" and / or "rounded" outline, etc.) on improving click-through rate and conversion rate.

[0056] Preferably, in this embodiment of the invention, the heat weight is characterized as follows:

[0057]

[0058] in, The heat weight of the i-th design element feature. Let be the intensity value of the i-th design element feature in the j-th sales sample. This represents the normalized sales volume corresponding to the sample. The conversion rate weighted by the sample. To prevent extremely small constants with a denominator of zero, The duration factor of the popularity of design features in the market.

[0059] In subsequent applications, the system updates the popularity weights based on real-time sales big data and retrains the model at preset time intervals (such as hourly) to capture subtle changes in market trends. Through this dynamic update mechanism, the system can identify subtle shifts in current audience aesthetics, such as the shift from "minimalist style" to "intricate traditional Chinese style," thereby providing rigorously scientific quantitative evaluation indicators for the automated iteration of subsequent cultural and creative designs.

[0060] In this embodiment of the invention, by establishing a mapping relationship between design element features and sales data, each feature that contributes to sales can be extracted and its corresponding popularity weight can be determined. This facilitates subsequent guidance on optimizing the design focus, allows for real-time tracking of changes in current market aesthetic styles, promotes the sales of digital cultural and creative design products, and meets the actual interests of enterprises.

[0061] After determining the popularity weight of each design element feature, the optimization design of digital cultural and creative products begins. In existing technologies, design elements features with a certain popularity are often directly used for optimization. However, in actual application, some design elements features may be newly emerging and have great market expansion potential. If optimization is not carried out in time, market opportunities may be missed. On the other hand, some design elements features may have already flooded the market, leading to customer aesthetic fatigue or not conforming to current market preferences, which may result in a sharp increase in inventory and losses for the company.

[0062] In this embodiment of the invention, the step of performing dynamic optimization of cultural and creative design based on the popularity weight to generate an optimized design includes: determining a first trigger threshold corresponding to market hotspots and a second trigger threshold corresponding to aesthetic fatigue, wherein the first trigger threshold is less than the second trigger threshold; determining the weight increase trend and weight decrease trend of each design element feature based on the popularity weight; when any design element feature is in the weight increase trend and the corresponding popularity weight reaches the first trigger threshold, determining the pixel to be optimized and the parameter to be optimized based on the corresponding design element feature, performing dynamic optimization of cultural and creative design based on the pixel to be optimized and the parameter to be optimized to generate an optimized design; and removing the corresponding design element feature from the dynamic optimization of cultural and creative design when any design element feature is in the weight decrease trend and the corresponding popularity weight reaches the second trigger threshold.

[0063] In one possible implementation, the parameters of the cultural and creative visual scheme are adjusted in real time by invoking Artificial Intelligence Generated Content (AIGC) or a parametric design engine to optimize the design of digital cultural and creative products. In the specific implementation process, the accuracy of utilization is improved by setting trigger range constraints for design element features. First, a first trigger threshold corresponding to market trends is determined. For example, this first trigger threshold is pre-set by technicians based on data on the popularity changes of market trends or blockbuster products during their actual emergence. For instance, if technicians find that the popularity weight of a certain design element feature is continuously increasing and reaches 0.4, it may gradually become a market trend or a blockbuster product, at which point it can be monitored and optimized.

[0064] On the other hand, a second trigger threshold corresponding to aesthetic fatigue is also determined. For example, if technicians find that the popularity weight of a certain design element feature continues to decrease and once it reaches 0.7, it indicates that the product has already caused aesthetic fatigue in the market, and sales will continue to decline and be replaced by other popular products. Therefore, it is necessary to remove it from the product to sell the products in the inventory as soon as possible, use the remaining popularity of the product to reduce inventory, and improve the actual operating efficiency of the enterprise.

[0065] In this embodiment of the invention, on the one hand, by adopting a dual-threshold constraint optimization design scheme, the optimization direction and scope of digital cultural and creative design products can be effectively constrained, avoiding ineffective or low-quality optimization; on the other hand, by adopting a "counterintuitive" design method where the first threshold is lower than the second threshold to guide the optimization design action, market hotspots can be discovered more quickly while avoiding aesthetic fatigue design, increasing sales while effectively avoiding inventory increases, improving the company's product turnover efficiency, reducing inventory, and improving the company's operating efficiency.

[0066] If popular design elements are directly used to optimize a company's products during the design optimization process, the company's products may lose their unique features or recognizability, which is detrimental to the company's brand building and, in the long run, will result in financial losses for the company.

[0067] In this embodiment of the invention, the step of performing dynamic optimization of cultural and creative design based on corresponding design element features to generate an optimized design includes: determining the pixels to be optimized and the parameters to be optimized based on the corresponding design element features; identifying the brand features of the digital cultural and creative design product to obtain product brand features; adjusting the pixels to be optimized based on the product brand features to obtain adjusted pixels; and performing dynamic optimization of cultural and creative design based on the adjusted pixels and the parameters to be optimized to generate the optimized design.

[0068] In one possible implementation, during the dynamic optimization process, the pixels and parameters to be optimized are first determined. This can be achieved by using color space conversion algorithms (such as converting from RGB to CIELAB color space) or extracting feature contour curvature, thus converting the aesthetic preferences suggested by the heat weighting into specific color values. The pixels and parameters to be optimized include, but are not limited to, the hexadecimal value (HEX value) mapping relationship of the main color tone, the proportion of the IP character's movement (joint angle parameters based on skeletal animation), the dynamic stretching of the composition proportions (layout adjustment based on the golden ratio algorithm), and the distribution density control of auxiliary elements. Further, brand characteristic identification is performed on the digital cultural and creative design product to obtain the product's brand characteristics. Then, in the subsequent dynamic optimization process, conventional color difference calculation methods are used to automatically correct the color components in the product design draft, ensuring that the adjusted visual scheme, while maintaining the original brand recognition (i.e., maintaining a certain Euclidean distance in the color space), maximizes its compatibility with the current audience's aesthetic threshold. For example, the system defines the core color gamut range of the brand visual identity system within the color space. After calculating the color scheme based on market preferences, it calculates the distance between the scheme and the core color gamut range. If the adjusted color exceeds the allowable deviation range of brand identity, the centripetal projection algorithm is used to pull the color back to the edge of the range, ensuring that the automated design does not damage the brand's long-term visual assets while meeting the audience's aesthetic threshold.

[0069] In the specific dynamic optimization process, an automated pipeline completes the entire workflow from design modification and rendering (using a cloud-based GPU rendering farm) to updating materials across all channels. This system can control the visual adjustment cycle within a preset, extremely short time (e.g., within 15 minutes), enabling banner images and product detail page materials on online e-commerce platforms to be updated instantly based on real-time sales feedback.

[0070] In this embodiment of the invention, by further considering the company's own brand characteristics during the optimization design process, the company can launch products that meet market preferences while ensuring brand recognition, which is beneficial to the company's actual interests in the long run.

[0071] At this point, establishing collaborative control between design schemes and inventory management is crucial to optimizing the deep collaboration between production, sales, and research of digital cultural and creative design products, thereby increasing product popularity, reducing inventory backlog, and improving business efficiency.

[0072] In this embodiment of the invention, the step of performing sales forecasting on the optimized design to generate predicted sales fluctuations includes: determining an initial sales forecast based on the optimized design; generating a design sensitivity adjustment coefficient and a weight change based on the popularity weight; obtaining the marginal contribution rate of design element features to sales and external influencing factors; and generating predicted sales fluctuations based on the initial sales forecast, the design sensitivity adjustment coefficient, the weight change, the marginal contribution rate, and the external influencing factors. The predicted sales fluctuations are characterized as follows:

[0073]

[0074] in, To predict sales fluctuations, For initial sales forecast, To design the sensitive adjustment coefficient, For design element characteristics The change in the weight of popularity The marginal contribution rate of this design element feature to sales. As an external influencing factor, This is a random error correction item (pre-set by technicians).

[0075] In one possible implementation, an initial sales forecast is first determined based on the optimized design. For example, a rough forecast can be generated by using historical sales time-series data to predict the optimized design based on historical sales patterns. Then, the concept of design change sensitivity is introduced. A design sensitivity adjustment coefficient can be generated based on the ratio of the weight change rate of design element features to the sales change rate in the previous forecast period, thereby achieving a more accurate sales forecast. Furthermore, the marginal contribution rate of design element features to sales (e.g., pre-set by technical personnel) and external influencing factors (e.g., determined by weighted summation based on platform activity discounts, holidays, etc.) are obtained. Finally, a corresponding prediction model is constructed to calculate a more accurate prediction of sales fluctuations.

[0076] In practical applications, by establishing the above prediction model, the expected click-through rate increment and conversion rate increment after the design scheme optimization are used as core input variables. Combined with the current inventory turnover performance of the omnichannel, the Long Short-Term Memory Neural Network (LSTM) algorithm is used to model and predict the future turnover efficiency, and generate corresponding sales forecasts. The replenishment instructions or production scheduling adjustment tasks are automatically issued to the supply chain system to achieve high accuracy and timeliness of inventory management.

[0077] Finally, based on predicted sales fluctuations, corresponding inventory management results are generated. For example, the system automatically issues replenishment orders through a real-time synchronization protocol between the Enterprise Resource Planning (ERP) system and the Warehouse Management System (WMS). When the predicted sales indicate an inventory shortage risk (i.e., the number of days after dividing the current inventory by the predicted daily sales) reaches a preset threshold (e.g., below 50% of the safety stock turnover days), the engine automatically calculates the optimal replenishment batch size and lead time. Subsequently, the system automatically sends a production order correction request to the production end, adjusting the priority of the production schedule, or directly triggering pre-set procurement agreements with upstream raw material suppliers. This mechanism effectively solves the sales fluctuations caused by design changes in cultural and creative products and reduces the risk of redundant inventory accumulation due to outdated designs.

[0078] In this embodiment of the invention, by improving the existing sales forecasting method, design sensitivity and external promotional factors are further introduced on the basis of traditional historical sales forecasting, thereby further improving the accuracy of sales forecasting and improving the accuracy of inventory management.

[0079] Please see Figure 2 Based on the same inventive concept, embodiments of the present invention provide an apparatus, the apparatus comprising:

[0080] Furthermore, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the methods described in the embodiments of the present invention.

[0081] The optional embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above embodiments. Within the scope of the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention.

[0082] It should also be noted that the various specific technical features described in the above embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not describe the various possible combinations separately.

[0083] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a microcontroller, chip, or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0084] Furthermore, various different implementations of the present invention can be combined arbitrarily, as long as they do not violate the spirit of the present invention, they should also be regarded as the content disclosed in the present invention.

Claims

1. A method for dynamic optimization of digital cultural and creative design and collaborative inventory management based on sales big data, characterized in that, The method includes: Acquire omnichannel retail big data, preprocess the omnichannel retail big data based on user interaction, and obtain preprocessed data; Perform atomic-level design element decomposition on digital cultural and creative design products to obtain the corresponding design element characteristics; The sales contribution of each design element feature is determined based on the preprocessed data, and the corresponding popularity weight is determined based on the sales contribution. Based on the aforementioned popularity weights, a dynamic optimization operation for cultural and creative design is performed to generate an optimized design. The optimized design is used to perform sales forecasting, generating predicted sales fluctuations. Based on the predicted sales fluctuations, corresponding inventory management results are generated.

2. The method according to claim 1, characterized in that, The preprocessing of the omnichannel retail big data based on user interaction to obtain preprocessed data includes: Establish a standard parameter table, conduct interactive analysis on user interaction parameters across different channels and platforms, both online and offline, and generate a unified set of interactive parameters. A similar index mapping table is established based on the standard parameter table and the same set of interactive parameters; Based on the aforementioned similar indicator mapping table, preliminary behavioral classification is performed on the omnichannel retail big data to obtain behaviorally classified data. Based on the behavioral classification data, perform time-series analysis to generate user behavior analysis results; Based on the hidden Markov model, the transaction contribution analysis is performed on the user behavior analysis results to generate the transaction contribution probability corresponding to each user behavior. Preprocessed data is generated based on the behavior classification data and the transaction contribution probability.

3. The method according to claim 1, characterized in that, The process of performing atomic-level design element decomposition on digital cultural and creative design products to obtain corresponding design element features includes: Image data processing is performed on the digital cultural and creative design products based on a deep convolutional neural network to obtain the corresponding high-dimensional matrix; Atomic-level visual feature extraction is performed on the high-dimensional matrix to obtain the corresponding atomic visual features; Perform logical association analysis on the atomic visual features to obtain logically associated feature groups; Design element features are generated based on the atomic visual features and the logical association feature group.

4. The method according to claim 3, characterized in that, The process of determining the sales contribution of each design element feature based on the preprocessed data, and determining the corresponding popularity weight based on the sales contribution, includes: Normalization is performed on the preprocessed data to obtain normalized sales data; The intensity value of each design element feature in the preprocessed data is determined based on the normalized sales data. Based on the normalized sales data, the conversion rate weighting value for each sales sample and the duration of popularity of each design element feature in the market are determined. The gradient boosting decision tree algorithm is used to analyze the normalized sales data, the intensity value, the conversion rate weighted value, and the popularity duration to generate popularity weights corresponding to each design element feature.

5. The method according to claim 4, characterized in that, The heat weight is characterized as follows: in, For the first The heat weight of each design element feature Let be the intensity value of the i-th design element feature in the j-th sales sample. This represents the normalized sales volume corresponding to the sample. The conversion rate weighted value corresponding to the sample. To prevent extremely small constants with a denominator of zero, The duration of the popularity of a design feature in the market.

6. The method according to claim 1, characterized in that, The step of performing dynamic optimization of cultural and creative design based on the popularity weight to generate optimized design includes: Determine a first trigger threshold corresponding to market hotspots and a second trigger threshold corresponding to aesthetic fatigue, wherein the first trigger threshold is less than the second trigger threshold; The upward and downward trends of the weights for each design element feature are determined based on the aforementioned heat weights. When any design element feature is in the upward trend of the weight and the corresponding popularity weight reaches the first trigger threshold, a dynamic optimization operation of cultural and creative design is performed based on the corresponding design element feature to generate an optimized design. If any design element feature is in a decreasing weight trend and the corresponding popularity weight reaches the second trigger threshold, the corresponding design element feature will be removed from the dynamic optimization operation of cultural and creative design.

7. The method according to claim 6, characterized in that, The step of performing dynamic optimization of cultural and creative design based on the corresponding design element features to generate an optimized design includes: The pixels to be optimized and the parameters to be optimized are determined based on the corresponding design element features; The brand characteristics of the digital cultural and creative design products are identified to obtain the product brand characteristics. The pixel to be optimized is adjusted based on the product brand characteristics to obtain the adjusted pixel; Based on the adjusted pixels and the parameters to be optimized, a dynamic optimization operation for cultural and creative design is performed to generate an optimized design.

8. The method according to claim 1, characterized in that, The step of performing sales forecasting on the optimized design and generating predicted sales fluctuations includes: Determine the initial sales forecast based on the optimized design; Based on the aforementioned heat weights, a design sensitivity adjustment coefficient and weight change amount are generated; Obtain the marginal contribution rate of design element features to sales and external influencing factors; Based on the initial sales forecast, the design sensitivity adjustment coefficient, the weight change, and the external influencing factors, a predicted sales fluctuation is generated, which is characterized as follows: in, To predict sales fluctuations, For initial sales forecast, To design the sensitive adjustment coefficient, For design element characteristics The change in popularity weight, The marginal contribution rate of this design element feature to sales. As an external influencing factor, This is a random error correction term.

9. A device for dynamic optimization of digital cultural and creative design and collaborative inventory management based on sales big data feedback, characterized in that, The device includes: The data processing module is used to acquire omnichannel retail big data, preprocess the omnichannel retail big data based on user interaction, and obtain preprocessed data. The decomposition module is used to perform atomic-level design element decomposition on digital cultural and creative design products to obtain the corresponding design element features. The weight calculation module is used to determine the sales contribution of each design element feature based on the preprocessed data, and to determine the corresponding popularity weight based on the sales contribution. The design module is used to perform dynamic optimization operations on cultural and creative designs based on the popularity weights, and generate optimized designs. The prediction module is used to perform sales forecasting on the optimized design and generate predicted sales fluctuations. The management module is used to generate corresponding inventory management results based on the predicted sales fluctuations.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the method described in any one of claims 1-8.

Images