Smart shop data processing method and device based on multi-dimensional data analysis

Abstract

The embodiment of the invention provides a smart store data processing method and device based on multi-dimensional data analysis, which innovatively realizes alignment and mapping of spatial-temporal feature sequences by collecting sensor, intelligent terminal and social media data. And performing multi-dimensional data fusion based on the feature weight coefficient, and constructing a commodity recommendation model by using a convolutional neural network and a recurrent neural network. The system continuously optimizes the model performance through incremental learning, and realizes personalized commodity recommendation and store operation trend analysis. And meanwhile, a marketing decision optimization model is introduced, intelligent evaluation and dynamic optimization are performed on a marketing strategy, and the model is continuously updated through feedback data. According to the method, the limitation of traditional single data analysis is broken through, and a comprehensive data processing solution is provided for precision marketing of the smart store.

Description

Technical Field

[0001] The present application relates to the field of data processing, and specifically to a smart store data processing method and device based on multidimensional data analysis. Background Art

[0002] Data analysis in traditional retail stores is usually limited to single-dimensional sales data statistics, making it difficult to fully grasp customer needs and market trends. Existing product recommendation systems often rely on simple purchase history analysis and lack deep integration and mining of multi-source data. Especially in terms of personalized recommendations and precision marketing, existing systems are unable to fully utilize multi-dimensional information such as sensor data, device data, and social media data.

[0003] At the same time, the existing data processing system has obvious deficiencies in data fusion and model training. Traditional methods often use simple data superposition, fail to effectively deal with the problem of spatiotemporal alignment of data from different sources, and lack a dynamic evaluation mechanism for the importance of data features. The system also lags behind in model updating and optimization, making it difficult to adapt to the rapidly changing market environment.

[0004] In addition, the existing system is also relatively mechanical in terms of marketing strategy formulation and execution. It lacks an intelligent decision-making optimization mechanism and cannot dynamically adjust marketing strategies based on real-time feedback data, which affects the continuous improvement of marketing effectiveness. Solving these problems is of great significance to improving the operational efficiency and marketing accuracy of smart stores. Summary of the invention

[0005] In response to the problems in the prior art, the present application provides a smart store data processing method and device based on multi-dimensional data analysis, which can break through the limitations of traditional single data analysis and provide a comprehensive data processing solution for the precise marketing of smart stores.

[0006] In order to solve at least one of the above problems, the present application provides the following technical solutions: In a first aspect, the present application provides a smart store data processing method based on multidimensional data analysis, comprising: Collecting multiple sensor data, smart terminal device data and social media data in the store, aligning the sensor data and the smart terminal device data with timestamps and mapping the spatial coordinates to obtain a spatiotemporal feature sequence, performing correlation analysis on the spatiotemporal feature sequence to obtain a feature weight coefficient, and performing multi-dimensional data fusion on the spatiotemporal feature sequence and the social media data based on the feature weight coefficient to generate a fused data set; The fused data set is cleaned and standardized to obtain sample data, the sample data is divided into a training data set and a validation data set according to a preset ratio, the training data set is used to train a product recommendation model based on a convolutional neural network and a recurrent neural network, the validation data set is used to perform performance evaluation on the trained model, the model parameters are automatically tuned based on the evaluation results and incremental learning updates are performed, the updated model is used for product preference prediction to generate a personalized product recommendation list, and the fused data set is subjected to time series analysis to obtain store operation trend data; An initial marketing strategy plan is generated based on the personalized product recommendation list and the store operation trend data, and the initial marketing strategy plan is input into a pre-trained marketing decision optimization model for strategy evaluation and optimization to obtain a personalized marketing strategy. The personalized marketing strategy is desensitized and encrypted before being stored in a cloud data center. The personalized marketing strategy is distributed to store terminals for execution through a message queue, and marketing strategy execution effect data is collected for model optimization and updating.

[0007] Furthermore, the sensor data and the smart terminal device data are timestamp aligned and spatial coordinate mapped to obtain a spatiotemporal feature sequence, and a correlation analysis is performed on the spatiotemporal feature sequence to obtain a feature weight coefficient, including: Establish a spatiotemporal reference coordinate system based on the collection time and device location information in the sensor data, map the sensor data to the time axis according to the timestamp marked with the collection time, convert the device number in the smart terminal device data into a corresponding spatial coordinate point and map it to the spatial axis, and resample the data with different sampling frequencies through bilinear interpolation to generate a spatiotemporal feature sequence; The Pearson correlation coefficient is used to calculate the correlation between the features in the spatiotemporal feature sequence, a feature correlation matrix is ​​constructed, the contribution of each feature to the target variable is calculated based on the feature correlation matrix, and the contribution is normalized to obtain a feature weight coefficient.

[0008] Furthermore, the performing multi-dimensional data fusion on the spatiotemporal feature sequence and the social media data based on the feature weight coefficient to generate a fused data set includes: The feature weight coefficient is weighted and calculated with the corresponding spatiotemporal feature sequence to obtain a weighted feature value, the social media data is subjected to topic classification and sentiment analysis to obtain a text feature vector, and the weighted feature value is aligned with the text feature vector in the time dimension according to the timestamp mark in the spatiotemporal feature sequence to construct a multidimensional feature tensor; A tensor decomposition algorithm is used to reduce the dimension of the multidimensional feature tensor to obtain a feature subspace, and a local linear embedding algorithm is used to map the heterogeneous data in the feature subspace to a unified feature space to generate a fused data set.

[0009] Furthermore, the fusion data set is cleaned and standardized to obtain sample data, the sample data is divided into a training data set and a validation data set according to a preset ratio, the training data set is used to train a product recommendation model based on a convolutional neural network and a recurrent neural network, the validation data set is used to perform a performance evaluation on the trained model, and the model parameters are automatically tuned based on the evaluation results and incremental learning updates are performed, including: The missing values ​​in the fused data set are supplemented by the K nearest neighbor interpolation method, the outlier detection algorithm is used to remove abnormal data points, the numerical features are normalized to the maximum and minimum values, and the categorical features are converted to one-hot encoding to obtain sample data, and the sample data are randomly divided into a training data set and a validation data set in a ratio of 8:2; A convolutional neural network is used to extract spatial features in the training data set, and the extracted spatial features are input into a recurrent neural network for time series modeling. The network parameters are optimized by a back propagation algorithm to obtain an initial product recommendation model. The accuracy, recall rate and F1 score of the model are calculated using the validation data set. The learning rate, batch size and number of network layers of the initial product recommendation model are tuned based on a grid search algorithm, and the sliding window method is used to perform incremental training on newly added data samples to update the model parameters.

[0010] Furthermore, the updated model is used for product preference prediction to generate a personalized product recommendation list, and the fused data set is subjected to time series analysis to obtain store operation trend data, including: Input the user's historical behavior data into the updated product recommendation model, extract the user's interest feature vector, calculate the similarity matrix between the user and the product through the collaborative filtering algorithm, sort the products according to the similarity score and select the products with the highest score to build a personalized recommendation list; The fused data set is divided into time series according to hours and dates, and the exponential smoothing method is used to perform trend decomposition on the customer flow and sales data. The cyclical and seasonal characteristics in the data are extracted through the autoregressive model, and the decomposed time series data is modeled and analyzed with the long short-term memory network to obtain the store operation trend data.

[0011] Furthermore, the generating of an initial marketing strategy plan based on the personalized product recommendation list and the store operation trend data, and inputting the initial marketing strategy plan into a pre-trained marketing decision optimization model for strategy evaluation and optimization to obtain a personalized marketing strategy includes: Based on the commodity categories and sales forecast values ​​in the personalized commodity recommendation list, combined with the customer flow fluctuation pattern and consumer portrait distribution in the store operation trend data, a decision tree algorithm is used to build a commodity promotion rule library, and an initial marketing strategy plan including promotion intensity, promotion period and target customer group is generated through rule matching; The initial marketing strategy plan is converted into a strategy feature vector and input into a marketing decision optimization model built based on reinforcement learning. The strategy parameters are explored through the Monte Carlo tree search method, the strategy benefit value is calculated based on historical marketing effect data, and the strategy gradient algorithm is used to maximize the benefit value to obtain a personalized marketing strategy.

[0012] Furthermore, the personalized marketing strategy is desensitized and encrypted and then stored in a cloud data center, the personalized marketing strategy is distributed to store terminals for execution through a message queue, and marketing strategy execution effect data is collected for model optimization and updating, including: Hash encryption is performed on the user identification information in the personalized marketing strategy, the policy parameters are reversibly encrypted, the sensitive product information is desensitized using data masking technology, the processed policy data is written to multiple storage nodes in the cloud data center in a distributed storage manner, a policy distribution channel is established based on the distributed message middleware, and the encrypted marketing policy data packet is pushed to the message queue server; Collect sales conversion rate, average order value change and member repurchase rate data of marketing strategies in different time periods from store terminals, parse the collected data into effect evaluation indicators in a preset format, and transmit the effect evaluation indicators back to the model optimization module through the data pipeline to establish sample data pairs containing strategy parameters and execution effects for model parameter updating.

[0013] In a second aspect, the present application provides a smart store data processing device based on multidimensional data analysis, comprising: A multidimensional data fusion module is used to collect multi-channel sensor data, smart terminal device data and social media data in the store, perform timestamp alignment and spatial coordinate mapping on the sensor data and the smart terminal device data to obtain a spatiotemporal feature sequence, perform correlation analysis on the spatiotemporal feature sequence to obtain a feature weight coefficient, and perform multidimensional data fusion on the spatiotemporal feature sequence and the social media data based on the feature weight coefficient to generate a fused data set; A multidimensional data processing module, for performing data cleaning and standardization processing on the fused data set to obtain sample data, dividing the sample data into a training data set and a validation data set according to a preset ratio, using the training data set to train a product recommendation model based on a convolutional neural network and a recurrent neural network, using the validation data set to perform performance evaluation on the trained model, automatically tuning model parameters based on the evaluation results and performing incremental learning updates, using the updated model to predict product preferences to generate a personalized product recommendation list, and performing time series analysis on the fused data set to obtain store operation trend data; An intelligent recommendation module is used to generate an initial marketing strategy plan based on the personalized product recommendation list and the store operation trend data, input the initial marketing strategy plan into a pre-trained marketing decision optimization model for strategy evaluation and optimization to obtain a personalized marketing strategy, perform data desensitization and encryption on the personalized marketing strategy and then store it in a cloud data center, distribute the personalized marketing strategy to store terminals for execution through a message queue, and collect marketing strategy execution effect data for model optimization and update.

[0014] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the smart store data processing method based on multidimensional data analysis are implemented.

[0015] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the smart store data processing method based on multidimensional data analysis.

[0016] In a fifth aspect, the present application provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the smart store data processing method based on multidimensional data analysis.

[0017] It can be seen from the above technical solution that the present application provides a smart store data processing method and device based on multidimensional data analysis, which innovatively realizes the alignment and mapping of spatiotemporal feature sequences by collecting sensor, smart terminal and social media data. Multidimensional data fusion is performed based on feature weight coefficients, and a product recommendation model is constructed using convolutional neural networks and recurrent neural networks. The system continuously optimizes model performance through incremental learning to achieve personalized product recommendations and store operation trend analysis. At the same time, a marketing decision optimization model is introduced to intelligently evaluate and dynamically optimize marketing strategies, and the model is continuously updated through feedback data. This method breaks through the limitations of traditional single data analysis and provides a comprehensive data processing solution for the precision marketing of smart stores. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 This is one of the flow charts of the smart store data processing method based on multidimensional data analysis in the embodiment of the present application; Figure 2 This is a second flow chart of the smart store data processing method based on multi-dimensional data analysis in an embodiment of the present application; Figure 3 This is a third flow chart of the smart store data processing method based on multi-dimensional data analysis in an embodiment of the present application; Figure 4 This is a fourth flow chart of the smart store data processing method based on multi-dimensional data analysis in an embodiment of the present application; Figure 5 This is a fifth flow chart of the smart store data processing method based on multi-dimensional data analysis in an embodiment of the present application; Figure 6 This is a sixth flow chart of the smart store data processing method based on multidimensional data analysis in an embodiment of the present application; Figure 7 FIG7 is a flow chart of a smart store data processing method based on multi-dimensional data analysis in an embodiment of the present application; Figure 8 This is a structural diagram of a smart store data processing device based on multidimensional data analysis in an embodiment of the present application; Fig. 9 It is a schematic diagram of the structure of an electronic device in an embodiment of the present application.

[0020] Reference numerals: Electronic device 9600, central processing unit 9100, memory 9140, communication module 9110, input unit 9120, audio processor 9130, display 9160, power supply 9170, buffer memory 9141, application / function storage unit 9142, data storage unit 9143, driver program storage unit 9144, antenna 9111, speaker 9131, microphone 9132. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0022] The acquisition, storage, use, and processing of data in the technical solution of this application comply with the relevant provisions of national laws and regulations.

[0023] Taking into account the problems existing in the prior art, the present application provides a smart store data processing method and device based on multidimensional data analysis, which innovatively realizes the alignment and mapping of spatiotemporal feature sequences by collecting sensor, smart terminal and social media data. Multidimensional data fusion is performed based on feature weight coefficients, and a product recommendation model is constructed using convolutional neural networks and recurrent neural networks. The system continuously optimizes model performance through incremental learning to achieve personalized product recommendations and store operation trend analysis. At the same time, a marketing decision optimization model is introduced to intelligently evaluate and dynamically optimize marketing strategies, and the model is continuously updated through feedback data. This method breaks through the limitations of traditional single data analysis and provides a comprehensive data processing solution for the precision marketing of smart stores.

[0024] In order to break through the limitations of traditional single data analysis and provide a comprehensive data processing solution for the precision marketing of smart stores, this application provides an embodiment of a smart store data processing method based on multidimensional data analysis, see Figure 1 The smart store data processing method based on multidimensional data analysis specifically includes the following contents: Step S101: Collect multiple sensor data, smart terminal device data and social media data in the store, perform timestamp alignment and spatial coordinate mapping on the sensor data and the smart terminal device data to obtain a spatiotemporal feature sequence, perform correlation analysis on the spatiotemporal feature sequence to obtain a feature weight coefficient, and perform multi-dimensional data fusion on the spatiotemporal feature sequence and the social media data based on the feature weight coefficient to generate a fused data set; Optionally, this embodiment designs a comprehensive data collection system for smart store application scenarios. High-precision infrared sensor arrays are deployed in different areas of the store, with each sensor covering an area of ​​3 square meters to collect crowd density data in real time; temperature and humidity sensors are arranged in key locations such as product display areas and rest areas to record environmental parameters every 15 minutes; high-definition cameras are installed at imports and exports, main channels and shelf areas to collect customer behavior images; RFID readers cover all product display areas to track changes in product locations in real time. Smart terminal devices include new POS machines with biometric functions, smart electronic price tags and self-service cash registers to collect transaction data, price changes and inventory information. At the same time, user comments, picture sharing and interaction data related to the store on social platforms such as Weibo and Xiaohongshu are obtained in real time through the API interface.

[0025] This embodiment innovatively proposes a unified spatiotemporal mapping method for multi-source heterogeneous data. First, a standard timeline based on the store opening time is established, and all sensor data are preliminarily aligned according to millisecond timestamps. Since the sampling frequencies of different sensors vary significantly, for example, the crowd flow sensor samples once every 1 minute, while the temperature and humidity sensor samples once every 15 minutes, an improved bilinear interpolation algorithm is used for data resampling. Specifically, for any time t, the data value of the interpolation point is calculated through a two-dimensional weight matrix based on the data value and time interval of two adjacent sampling points to ensure that the interpolation result has a smooth transition and conforms to the physical meaning.

[0026] In the spatial dimension, this embodiment establishes a spatial coordinate system based on the three-dimensional model of the store. The installation position of the fixed sensor is determined by a surveying instrument, and mobile devices such as handheld POS machines update the location information in real time through the indoor positioning system. Through the coordinate conversion matrix, all position data are uniformly mapped to a standardized grid coordinate system, and each grid unit is 0.5 meters × 0.5 meters in size, so as to achieve precise alignment of spatial data. For example, the RFID identification data of a shelf area can be accurately associated with the passenger flow density, residence time and other data of the area.

[0027] For the correlation analysis of spatiotemporal feature sequences, this embodiment designs a hierarchical and progressive analysis method. First, the local correlation based on the time window is calculated, and the improved Pearson correlation coefficient method is used to consider the time series characteristics and periodic changes of the data. Then a multidimensional correlation matrix is ​​constructed to analyze not only the linear correlation between the features, but also the nonlinear correlation through mutual information and conditional entropy. For example, the analysis found that there is a significant time-lagged correlation between the residence time in the rest area and the purchase rate of surrounding products, and this correlation is more obvious during holidays.

[0028] This embodiment introduces a dynamic feature weight calculation mechanism. Based on the correlation analysis results, combined with the information gain rate and feature importance score, a feature weight evaluation model is constructed. The model considers the timeliness, stability and business relevance of the feature and dynamically adjusts the weight coefficient. For example, when predicting the sales of high-value goods, the weight of customer portrait features will be increased; when predicting the sales of fast-moving consumer goods, more emphasis will be placed on the weight of historical sales data.

[0029] In the multi-dimensional data fusion stage, this embodiment adopts a hierarchical fusion strategy. First, deep semantic analysis is performed on social media data, and text features are extracted using the pre-trained BERT model. Hot topics are identified through the topic model, and user emotional tendencies are quantified through the sentiment analysis model. Then, the text feature vector is aligned with the spatiotemporal feature sequence in the time dimension to construct a feature tensor containing multiple views.

[0030] This embodiment innovatively uses an improved tensor decomposition algorithm to process high-dimensional feature data. The original feature tensor is decomposed into a core tensor and multiple factor matrices by the Tucker decomposition method, and the optimal decomposition dimension is determined in combination with singular value analysis. On this basis, an improved local linear embedding algorithm is used to map heterogeneous data to a unified low-dimensional feature space, maintaining the local structural characteristics of the data.

[0031] This multi-dimensional data processing solution not only solves the data inconsistency problem in traditional methods, but also can adaptively handle data quality fluctuations. For example, when some sensor data is briefly interrupted, the system can repair the data through the complementarity of related features. At the same time, the update of the fused data set adopts an incremental calculation method, and the feature weight is adjusted every hour to ensure the timeliness and accuracy of the data analysis results.

[0032] The innovative data processing solution of this embodiment significantly improves the operational efficiency of smart stores. Through the deep integration of multi-dimensional data, it can accurately depict customer behavior patterns, predict commodity demand trends, provide data support for personalized marketing decisions, and effectively improve the operational efficiency and user experience of stores.

[0033] Step S102: performing data cleaning and standardization processing on the fused data set to obtain sample data, dividing the sample data into a training data set and a validation data set according to a preset ratio, using the training data set to train a product recommendation model based on a convolutional neural network and a recurrent neural network, using the validation data set to perform performance evaluation on the trained model, automatically tuning the model parameters based on the evaluation results and performing incremental learning updates, using the updated model to predict product preferences and generate a personalized product recommendation list, and performing time series analysis on the fused data set to obtain store operation trend data; Optionally, this embodiment first designs a multi-dimensional anomaly detection mechanism for data cleaning of fused data sets. In the time dimension, a sliding window with a period of 24 hours is used to calculate the mean, standard deviation, skewness, kurtosis and other statistics of each feature. For continuous data such as passenger flow and transaction amount, the improved 3-sigma criterion and exponentially weighted moving average method are combined to identify mutation points. At the same time, the correlation of the spatial dimension is considered. When the sensor data in a certain area is abnormal, cross-validation is performed through the data of the adjacent areas. For text data such as user comments, semantic vectors are calculated based on the pre-trained BERT model, and abnormal text is detected using the local anomaly factor algorithm.

[0034] This embodiment adopts a hierarchical adaptive normalization strategy in the data standardization stage. For strong periodic features such as pedestrian density, the periodic decomposition is first performed through wavelet transform, and the trend component and the periodic component are standardized respectively. For long-tail distribution features such as commodity sales, Min-Max normalization after logarithmic transformation is adopted. For environmental parameters such as temperature and humidity, reasonable normalization intervals are set based on domain knowledge. Category-type features such as commodity categories are encoded using feature hashing technology, and collision detection is used to ensure the uniqueness of the encoding.

[0035] When dividing the data set, this embodiment innovatively proposes a time-series-aware stratified sampling method. First, the data is divided into multiple time segments in chronological order, and each segment contains 24 hours of data. Then, within each time segment, stratification is performed according to dimensions such as user activity and product category to ensure that the training set and the validation set have similar data distribution. The specific ratio is 80% for the training set and 20% for the validation set, and it is ensured that the samples for each time segment and each data layer are allocated according to this ratio.

[0036] This embodiment designs a deep neural network model based on a dual-stream architecture. The spatial feature stream uses an improved ResNet structure, which contains multiple residual blocks and dilated convolutional layers, and can extract spatial features of different scales. The temporal feature stream uses a BiLSTM structure, which captures long-term dependencies through two forward and backward LSTM layers. The two feature streams perform feature fusion through the attention mechanism, and dynamically adjust the importance weights of different features.

[0037] The input features of the model include: user features (historical browsing sequence, purchase record, membership level, etc.), product features (category, price, inventory, sales, etc.), time features (time period, week, holidays, etc.), spatial features (product location, regional customer flow, etc.) and environmental features (temperature, humidity, weather, etc.). The interaction between features is modeled through a multi-head self-attention mechanism, and each attention head is responsible for capturing different types of feature associations.

[0038] During the training process, this embodiment adopts a multi-stage training strategy. The first stage uses pre-training tasks to learn the basic representation of features through mask prediction and sequence reconstruction tasks. The second stage conducts multi-task joint training to simultaneously optimize multiple objectives such as click prediction, purchase conversion, and dwell time prediction. A dynamic weight adjustment method based on task importance is used to adaptively balance the training losses of each task.

[0039] In order to improve the robustness of the model, this embodiment introduces a variety of regularization techniques. Spatial dropout is used in the convolutional layer to randomly discard feature maps; zoneout regularization is used in the recurrent layer to randomly keep the hidden state unchanged; and label smoothing technology is used to reduce overfitting. During the training process, a cosine annealing learning rate scheduler is used to periodically adjust the learning rate to help the model escape the local optimum.

[0040] Model evaluation uses a multi-dimensional evaluation system. In addition to basic indicators such as accuracy and recall, it also includes: coverage (to evaluate the breadth of recommended products), diversity (using a metric based on information entropy), novelty (based on the inverse calculation of product popularity), and relevance (evaluated through A / B testing). Based on these indicators, the Bayesian optimization algorithm is used to automatically adjust model hyperparameters, such as network depth, number of attention heads, dropout rate, etc.

[0041] The incremental learning mechanism of this embodiment includes three levels: at the feature level, feature distribution statistics are updated through incremental feature engineering; at the model level, elastic weight integration is adopted, and new and old model parameters are weighted and fused through importance; at the sample level, sample importance weighting based on time decay is used to reduce the influence of historical samples.

[0042] When generating the recommendation list, this embodiment adopts a multi-constrained sorting optimization method. First, multiple scores (click probability, conversion rate, profit margin, etc.) of candidate products are obtained through model prediction, and then the optimal recommendation sequence is solved by combining the multi-objective planning method, considering inventory constraints, exposure constraints and diversity constraints. At the same time, the exploration-exploitation mechanism is introduced to reserve a certain proportion of positions for recommending new products.

[0043] For store operation trend analysis, this embodiment combines empirical mode decomposition and spectral analysis methods to extract trend features at multiple time scales. Through multi-scale analysis at hourly, daily, and weekly levels, the changing patterns of key indicators such as customer flow and sales are identified, providing decision support for dynamic adjustment of marketing strategies and inventory management.

[0044] The innovative solution of this embodiment significantly improves the performance of the recommendation system. The quality of training data is guaranteed through multi-dimensional data cleaning and standardization; the dual-stream architecture of the deep neural network effectively integrates spatiotemporal features; multi-task learning and incremental update mechanisms enable the model to continuously adapt to business changes; and the multi-constrained recommendation generation strategy ensures the practicality of the recommendation results. This solution has shown strong adaptability and stability in practical applications, can quickly respond to changes in user interests, and improve recommendation results through continuous optimization.

[0045] Step S103: Generate an initial marketing strategy plan based on the personalized product recommendation list and the store operation trend data, input the initial marketing strategy plan into a pre-trained marketing decision optimization model for strategy evaluation and optimization to obtain a personalized marketing strategy, perform data desensitization and encryption on the personalized marketing strategy and then store it in a cloud data center, distribute the personalized marketing strategy to store terminals for execution through a message queue, and collect marketing strategy execution effect data for model optimization and update.

[0046] Optionally, this embodiment first constructs a multi-dimensional marketing strategy generation framework based on the personalized product recommendation list and store operation trend data. The framework comprehensively considers factors such as product characteristics (such as inventory turnover rate, gross profit margin, shelf life), user portraits (such as consumption ability, shopping frequency, category preference) and market environment (such as competitor product prices, seasonal characteristics, promotional activities), and determines the weight coefficient of each dimension through hierarchical analysis to generate an initial marketing strategy plan.

[0047] In the process of strategy generation, this embodiment adopts a combination of rule-based and heuristic algorithms. For high-liquidity products, the discount intensity is dynamically adjusted according to inventory levels and sales forecasts; for seasonal products, a step-by-step promotion plan is formulated in combination with historical sales curves; for related products, a bundling sales strategy is designed. At the same time, the characteristics of customer flow in different time periods are considered, and promotion efforts are increased during peak customer flow periods, while brand marketing is emphasized during off-seasons.

[0048] This embodiment innovatively designs a marketing decision optimization model based on reinforcement learning. The model adopts a dual Q network architecture. The state space includes characteristics such as product attributes, market environment and user behavior, and the action space includes decision variables such as price adjustment, promotion methods, and marketing channels. The reward function comprehensively considers multiple goals such as sales revenue, gross profit margin, and user satisfaction, and balances the importance of each goal through a weighted adaptive method.

[0049] The model training uses an experience replay mechanism to store historical decisions and their effects, and selects representative samples for learning through a priority sampling method. In order to improve the generalization ability of the model, an adversarial training mechanism is introduced to enhance the diversity of training samples by adding perturbation data. At the same time, an exploration strategy based on Thompson sampling is designed to ensure the stability of the strategy while maintaining moderate innovation.

[0050] In the marketing strategy evaluation stage, this embodiment builds a multi-level evaluation system. First, the expected effect of the strategy is evaluated through offline simulation, including indicators such as sales increase, gross profit change, and inventory turnover. Then, the strategy effect is verified in a small sample through A / B testing, and the strategy parameters are dynamically adjusted according to actual feedback. The evaluation results are fed back to the optimization model through the attention mechanism to guide the improvement direction of the strategy.

[0051] In terms of data security, this embodiment adopts a multiple encryption scheme. First, differential privacy processing is performed on sensitive data to prevent individual information leakage while maintaining data availability. Then, homomorphic encryption technology is used to encrypt marketing strategies to ensure the security of strategies during transmission and storage. A hierarchical encryption strategy is adopted for data of different security levels to balance security and system performance.

[0052] This embodiment designs a policy distribution mechanism based on message queues. Using the publish-subscribe model, marketing policies are grouped and pushed according to store attributes and geographic locations. The message queue supports message persistence and retry upon failure to ensure that policies can be delivered reliably. At the same time, policy version control is implemented to support rapid rollback and real-time update of policies.

[0053] In terms of feedback on the effect of strategy execution, this embodiment establishes a real-time monitoring system. Through the tracking technology, user interaction data, sales conversion data, inventory change data, etc. are collected. The stream processing framework is used to calculate the core indicators in real time. When the indicators fluctuate abnormally, the dynamic adjustment mechanism of the strategy is triggered. At the same time, the decision-making model is continuously optimized through incremental learning to improve the model's ability to adapt to market changes.

[0054] Through this solution combining deep learning and operational optimization, this embodiment realizes the intelligent generation and dynamic optimization of marketing strategies. The solution can quickly adjust strategies according to changes in the market environment, while improving sales results and ensuring the balance between inventory and profits. The scalability and security design of the solution enable it to adapt to retail scenarios of different sizes and provide merchants with accurate, safe and efficient marketing decision support.

[0055] From the above description, it can be seen that the smart store data processing method based on multidimensional data analysis provided by the embodiment of the present application can innovatively realize the alignment and mapping of spatiotemporal feature sequences by collecting sensor, smart terminal and social media data. Multidimensional data fusion is performed based on feature weight coefficients, and a product recommendation model is constructed using convolutional neural networks and recurrent neural networks. The system continuously optimizes model performance through incremental learning to achieve personalized product recommendations and store operation trend analysis. At the same time, a marketing decision optimization model is introduced to intelligently evaluate and dynamically optimize marketing strategies, and the model is continuously updated through feedback data. This method breaks through the limitations of traditional single data analysis and provides a comprehensive data processing solution for the precision marketing of smart stores.

[0056] In one embodiment of the smart store data processing method based on multidimensional data analysis of the present application, see Figure 2 , and can also include the following: Step S201: establishing a spatiotemporal reference coordinate system based on the acquisition time and device location information in the sensor data, mapping the sensor data to the time axis according to the timestamp marked with the acquisition time, converting the device number in the smart terminal device data into a corresponding spatial coordinate point and mapping it to the spatial axis, and resampling data with different sampling frequencies by bilinear interpolation to generate a spatiotemporal feature sequence; Step S202: using the Pearson correlation coefficient to calculate the correlation between the features in the spatiotemporal feature sequence, constructing a feature correlation matrix, calculating the contribution of each feature to the target variable based on the feature correlation matrix, and normalizing the contribution to obtain a feature weight coefficient.

[0057] Optionally, this embodiment first constructs a unified spatiotemporal reference coordinate system based on real-time collected sensor data and smart terminal device data. In the time dimension, the timestamp of each data point is standardized, and the times in different time zones and date formats are uniformly converted to UTC timestamps. In the spatial dimension, based on the store's floor plan, a two-dimensional rectangular coordinate system with the store entrance as the origin is established, and the physical location of each device is mapped to a specific coordinate point.

[0058] In view of the sampling frequency differences of different types of sensors such as temperature sensors and humidity sensors, this embodiment designs an adaptive resampling strategy. First, the sampling characteristics of various sensors are analyzed, and a higher sampling frequency is used for rapidly changing parameters (such as human flow), and a lower sampling frequency is used for slowly changing parameters (such as temperature). Then, an improved bilinear interpolation algorithm is used for data resampling. This algorithm introduces a time weight attenuation factor based on traditional bilinear interpolation, so that the interpolation result is more biased towards the measured data at the near moment.

[0059] In order to deal with the problem of dynamic changes in device location, this embodiment innovatively proposes a dynamic space mapping mechanism. The location changes of mobile devices are tracked in real time through Bluetooth beacons and WiFi positioning technology, and the location data is smoothed in combination with the Kalman filter algorithm to reduce the impact of positioning noise. At the same time, a device movement trajectory model is established to model the movement patterns of the device and improve the accuracy of location prediction.

[0060] In the process of generating the spatiotemporal feature sequence, this embodiment adopts a multi-dimensional feature extraction method. For the time dimension, in addition to the basic timestamp, time features such as time period, week, and holidays are also extracted; for the space dimension, in addition to the rectangular coordinates, spatial features such as the relative distance between devices and density distribution are also calculated. Through this multi-dimensional feature extraction, the spatiotemporal distribution characteristics of the data are fully captured.

[0061] In the feature correlation analysis phase, this embodiment first calculates the linear correlation between features by using the improved Pearson correlation coefficient. Taking into account the nonlinear characteristics of the data, a nonlinear correlation measure based on mutual information is introduced. By combining these two methods, a more comprehensive feature correlation matrix is ​​constructed.

[0062] In the feature contribution calculation process, this embodiment designs a feature importance evaluation method based on random forest. First, a random forest model is constructed, and the influence of each feature on the model performance is calculated by the feature random permutation method. At the same time, considering the interaction between features, the SHAP value calculation method is introduced to evaluate the marginal contribution of the feature.

[0063] For the normalization of feature weights, this embodiment uses a soft maximization function for optimization. This method can maintain the relative proportional relationship of feature weights while ensuring that the sum of weights is 1. In order to improve the stability of weight allocation, an L1 regularization term is introduced to suppress excessively small weight values ​​and highlight the role of important features.

[0064] The innovative solution of this embodiment effectively solves the problem of spatiotemporal alignment of multi-source heterogeneous data, and achieves seamless integration of data from different sources and different sampling frequencies through a unified spatiotemporal reference coordinate system and intelligent resampling strategy. At the same time, through multi-dimensional feature extraction and in-depth correlation analysis, key features that have a significant impact on the target variable are accurately identified, providing a reliable foundation for subsequent data modeling and analysis.

[0065] In practical applications, the solution has shown strong adaptability and scalability. Through the dynamic spatial mapping mechanism, it can accurately track the location changes of equipment; through multi-dimensional feature extraction, it can fully explore the spatiotemporal characteristics of data; through comprehensive feature importance evaluation, it can accurately identify key influencing factors. These technological innovations have significantly improved the efficiency and accuracy of data processing, providing high-quality data support for subsequent intelligent decision-making.

[0066] In one embodiment of the smart store data processing method based on multidimensional data analysis of the present application, see Figure 3 , and can also include the following: Step S301: performing weighted calculation on the feature weight coefficient and the corresponding spatiotemporal feature sequence to obtain a weighted feature value, performing topic classification and sentiment analysis on the social media data to obtain a text feature vector, aligning the weighted feature value with the text feature vector in the time dimension according to the timestamp mark in the spatiotemporal feature sequence, and constructing a multidimensional feature tensor; Step S302: using a tensor decomposition algorithm to perform dimensionality reduction processing on the multidimensional feature tensor to obtain a feature subspace, and mapping the heterogeneous data in the feature subspace to a unified feature space based on a local linear embedding algorithm to generate a fused data set.

[0067] Optionally, this embodiment first intelligently weights the spatiotemporal feature sequence according to the feature weight coefficients obtained in the previous step. An adaptive weighting algorithm is used to dynamically adjust the weight value according to the importance of features in different time periods and different regions. For example, during peak passenger flow periods, the weight of the passenger flow density feature will be increased accordingly; during promotional activities, the weight of the sales conversion rate feature will be strengthened. This dynamic weight adjustment mechanism can better reflect the actual importance of features in different scenarios.

[0068] In order to process social media data, this embodiment designs a multi-level text analysis framework. First, an improved LDA topic model is used for topic classification. By introducing a time decay factor, the model pays more attention to the changing trend of recent topics. In terms of sentiment analysis, a sentiment classifier for retail scenarios is constructed by combining the BERT pre-training model and domain adaptability fine-tuning. This classifier can not only identify basic sentiment polarity, but also capture consumers' fine-grained evaluation of specific products and services.

[0069] In the time alignment phase, this embodiment innovatively proposes a flexible time window mechanism. Considering the release delay of social media data, a time window of variable length is designed, and the window size is optimized through a dynamic programming algorithm to ensure that the temporal dependencies between data are captured. At the same time, the time decay function is introduced to make the model pay more attention to the changing characteristics of recent data.

[0070] In the process of constructing the multidimensional feature tensor, this embodiment adopts a hierarchical tensor structure. The first layer represents the time dimension, including different time scale features; the second layer represents the space dimension, including location and region features; the third layer represents the attribute dimension, including sensor data and text features. Through this multi-level tensor structure, the complex correlation between data is effectively retained.

[0071] In order to solve the computational efficiency problem of high-dimensional feature tensors, this embodiment uses an improved Tucker decomposition algorithm for dimensionality reduction. By introducing sparse constraints and non-negative constraints, the interpretability of the decomposition results is improved. At the same time, a regularization term based on the nuclear norm is designed to effectively suppress the influence of noise and outliers and extract a more stable feature subspace.

[0072] In the feature space mapping process of heterogeneous data, this embodiment improves the traditional local linear embedding algorithm. First, the number of neighboring points is dynamically determined according to the local distribution characteristics of the data through the adaptive neighbor graph construction method. Then, the manifold regularization term is introduced to maintain the local structural characteristics of the data during the dimensionality reduction process. Finally, the feature representations from different sources are fused through the multi-view learning framework.

[0073] This embodiment innovatively introduces a feature consistency check mechanism. The quality of feature mapping is evaluated by calculating the mutual information and conditional entropy between different feature spaces. When inconsistent features are found, the mapping parameters are optimized through a feedback adjustment mechanism to ensure that the fused features have good representation capabilities.

[0074] In the process of generating the fusion data set, this embodiment adopts a stratified sampling strategy to ensure the representativeness and balance of the data. Different normalization methods are used for different types of features to make the feature value distribution more reasonable. At the same time, the sample space is expanded through data enhancement technology to improve the generalization ability of the model.

[0075] Through the above innovative technical solutions, this embodiment achieves efficient fusion of multi-source heterogeneous data. The solution can effectively handle the heterogeneity of spatiotemporal features and text features, and generate a fused data set with strong expression capabilities through intelligent feature extraction and mapping mechanisms. In practical applications, the solution significantly improves the quality and efficiency of data fusion, and provides a reliable data foundation for subsequent analysis and decision-making. At the same time, the scalable design of the solution enables it to adapt to data fusion needs of different scales and types, and has a wide range of application value.

[0076] In one embodiment of the smart store data processing method based on multidimensional data analysis of the present application, see Figure 4 , and can also include the following: Step S401: supplement the missing values ​​in the fused data set by the K nearest neighbor interpolation method, remove abnormal data points by using the outlier detection algorithm, perform maximum and minimum value normalization on the numerical features, perform one-hot encoding conversion on the categorical features to obtain sample data, and randomly divide the sample data into a training data set and a validation data set in a ratio of 8:2; Step S402: Use a convolutional neural network to extract spatial features in the training data set, input the extracted spatial features into a recurrent neural network for time series modeling, optimize network parameters through a back propagation algorithm to obtain an initial product recommendation model, use the verification data set to calculate the accuracy, recall rate and F1 score of the model, tune the learning rate, batch size and number of network layers of the initial product recommendation model based on a grid search algorithm, and use a sliding window method to perform incremental training on newly added data samples to update model parameters.

[0077] Optionally, this embodiment first comprehensively deals with the data quality issues in the fusion data set. In the missing value filling stage, an improved K nearest neighbor interpolation algorithm is used, and the most similar K neighbors are determined according to the distance relationship of the samples in the feature space through a dynamic weight calculation method. Different similarity metrics are used for different types of features, such as using Euclidean distance for numerical features and editing distance for categorical features to ensure the rationality of the filling value.

[0078] In terms of outlier detection, this embodiment designs a multi-level detection mechanism. First, the interquartile range of the feature is calculated based on statistical methods to identify global outliers. Then, local outliers are detected by the isolation forest algorithm, which calculates the anomaly score of the sample by constructing a random decision tree. Finally, a reasonable threshold is set based on domain knowledge to eliminate abnormal samples.

[0079] For feature normalization, this embodiment adopts an adaptive normalization strategy. For numerical variables with different distribution characteristics, select the corresponding normalization method: for long-tail distribution characteristics, use logarithmic transformation and then normalize, and for normal distribution characteristics, directly perform maximum and minimum value normalization. For categorical features, considering the correlation between features, an improved one-hot encoding method is used to reduce the encoding dimension by combining features.

[0080] In the data set division phase, this embodiment innovatively proposes a stratified sampling strategy. First, the data set is divided into multiple time windows in chronological order, and then random sampling is performed in each window at a ratio of 8:2 to ensure that the training set and the validation set have similar time distribution characteristics. At the same time, considering the uneven distribution of commodity categories, oversampling technology is used to balance the number of samples in each category.

[0081] This embodiment designs an end-to-end deep learning framework for product recommendation. In the spatial feature extraction module, an improved convolutional neural network structure is used to capture spatial dependencies of different ranges through multi-scale convolution kernels. In order to deal with the sparsity problem of spatial features, an attention mechanism is introduced to highlight the feature representation of important areas.

[0082] In the time series modeling phase, this embodiment uses a bidirectional LSTM network structure to fully capture the dynamic characteristics of user behavior sequences through forward and backward time series information transmission. Considering the difficulty of long sequence modeling, a time gating mechanism is introduced to adaptively adjust the influence weight of historical information according to the time interval.

[0083] For the model training process, this embodiment adopts a multi-task learning framework to optimize both click prediction and purchase conversion. By designing a loss function with dynamic weights, the importance of each task is adaptively adjusted during the training process. During back propagation optimization, gradient clipping technology is used to prevent gradient explosion, and the Warmup strategy is used to stabilize the change of learning rate.

[0084] In terms of model tuning, this embodiment designs a two-stage optimization strategy. First, a rough range of key hyperparameters is determined through grid search, and then the Bayesian optimization method is used to find the optimal configuration in the fine-grained space. The optimization goal comprehensively considers accuracy, recall rate, and F1 score, and balances the importance of each indicator in a weighted manner.

[0085] In order to adapt to the dynamic changes in data distribution, this embodiment implements an online update mechanism for the model. The training samples are managed in a sliding window manner, and incremental training is triggered when the number of new samples reaches a preset threshold. Through transfer learning technology, the parameters of the trained model are used to initialize the new model to accelerate the model convergence process.

[0086] The innovative solution of this embodiment effectively solves the data quality and model optimization problems in product recommendations. Through systematic data preprocessing and deep learning models, accurate modeling of user shopping behavior is achieved. In practical applications, this solution can accurately capture the dynamic changes of user interests, provide personalized product recommendations, and significantly improve user shopping experience and platform operation efficiency.

[0087] In one embodiment of the smart store data processing method based on multidimensional data analysis of the present application, see Figure 5 , and can also include the following: Step S501: Input the user's historical behavior data into the updated product recommendation model, extract the user's interest feature vector, calculate the similarity matrix between the user and the product through the collaborative filtering algorithm, sort the products according to the similarity scores and select the products with the highest scores to build a personalized recommendation list; Step S502: Divide the fused data set into time series according to hours and dates, use exponential smoothing to perform trend decomposition on customer flow and sales data, extract the cyclical and seasonal characteristics in the data through an autoregressive model, and use a long short-term memory network to model and analyze the decomposed time series data to obtain store operation trend data.

[0088] Optionally, in the personalized recommendation stage, this embodiment first conducts an in-depth analysis of the user's historical behavior data. By constructing a user behavior sequence, including multi-dimensional behavior data such as browsing, collecting, adding to cart, and purchasing, the attention mechanism is used to assign different weights to different behavior types. For example, the weight of purchase behavior is higher than that of browsing behavior, and the weight of recent behavior is higher than that of long-term behavior. This differentiated weight distribution can more accurately portray the user's true interest preferences.

[0089] In the process of extracting user interest features, this embodiment designs a multi-level feature learning framework. First, the underlying feature representation is constructed through the attribute features of the product (such as category, brand, price range, etc.), and then the high-level semantic features are extracted using the deep neural network in the updated recommendation model. Through the residual connection mechanism, the features at different levels are fused to obtain a comprehensive user interest representation.

[0090] For the calculation of user-product similarity, this embodiment proposes an improved collaborative filtering algorithm. First, a bipartite graph structure is constructed based on the user's historical behavior, and the high-order association between the user and the product is captured through a graph neural network. At the same time, a time decay factor is introduced to make the model pay more attention to the user's recent interest changes. When calculating the similarity, the similarity of the product content characteristics and the user behavior sequence is comprehensively considered, and the final similarity score is obtained through weighted fusion.

[0091] In the product sorting stage, this embodiment adopts a multi-objective sorting strategy. In addition to considering the similarity score, multiple factors such as the real-time popularity, inventory status, and promotional information of the product are also introduced. By designing a dynamic weight sorting function, the importance of each factor is adaptively adjusted in different scenarios to ensure the diversity and timeliness of the recommendation results.

[0092] For the analysis of store operation trends, this embodiment innovatively proposes a multi-scale time series decomposition method. First, the fused data set is stratified according to the two granularities of hour and date, and the customer flow and sales data are trend-decomposed using an improved exponential smoothing algorithm. The algorithm can better handle noise and abnormal fluctuations in the data by adaptively adjusting the smoothing parameters.

[0093] In terms of periodicity and seasonality feature extraction, this embodiment designs an autoregressive model based on Fourier transform. The main periodic components in the data are identified through frequency domain analysis, such as weekday-weekend cycle, holiday effect, etc. At the same time, external factors (such as weather and promotional activities) are introduced as auxiliary features to improve the prediction accuracy of the model for special periods.

[0094] This embodiment adopts an improved long short-term memory network structure in the time series modeling stage. By designing multiple memory units, the dependencies of different time scales are captured respectively. The short-term memory unit focuses on the fluctuation characteristics at the hour level, and the long-term memory unit learns the trend characteristics at the date level. The weights of each time scale feature are dynamically adjusted through the gating mechanism to achieve effective modeling of multi-scale time series patterns.

[0095] Through the above innovative solution, this embodiment achieves accurate personalized product recommendations and reliable store operation trend prediction. In practical applications, the solution can accurately grasp the dynamic changes of user interests, provide product recommendations that meet user needs, and accurately predict store operation trends, providing data support for merchants to formulate operation strategies. The adaptive characteristics of the solution enable it to flexibly respond to recommendation needs in different scenarios, and it has strong practical value.

[0096] During the model application process, this embodiment continuously optimizes the model performance through an online learning mechanism. The model parameters are updated according to the real-time feedback data of users, ensuring that the recommendation system can timely adapt to the changes in user interests and the dynamic characteristics of the market environment, and continuously provide high-quality recommendation services and accurate trend forecasts.

[0097] In one embodiment of the smart store data processing method based on multidimensional data analysis of the present application, see Figure 6 , and can also include the following: Step S601: Based on the commodity categories and sales forecast values ​​in the personalized commodity recommendation list, combined with the customer flow fluctuation pattern and consumer portrait distribution in the store operation trend data, a commodity promotion rule library is constructed using a decision tree algorithm, and an initial marketing strategy plan including promotion intensity, promotion period and target customer group is generated through rule matching; Step S602: Convert the initial marketing strategy plan into a strategy feature vector, input it into a marketing decision optimization model built based on reinforcement learning, explore the strategy parameters through the Monte Carlo tree search method, calculate the strategy benefit value based on historical marketing effect data, and use the policy gradient algorithm to maximize the benefit value to obtain a personalized marketing strategy.

[0098] Optionally, this embodiment first adopts a multi-level decision analysis framework for the construction of product promotion rules. By analyzing the product characteristics in the personalized product recommendation list, including dimensions such as product category, price range, brand positioning, etc., a basic product portrait is constructed in combination with the sales forecast value. On this basis, the customer flow fluctuation law in the store operation trend data, such as the difference in customer flow between weekdays and weekends, holiday effects and other time series characteristics, as well as the consumer's age group, consumption ability, shopping preferences and other portrait characteristics are integrated to form a complete decision input.

[0099] In the design of the decision tree algorithm, this embodiment adopts an improved CART decision tree structure. Each node of the decision tree represents a judgment condition of a promotion rule, such as commodity category matching, customer flow threshold, target customer group characteristics, etc. The optimal split feature is selected through the information gain ratio criterion to ensure that the generated rules have a strong degree of discrimination. At the same time, a rule confidence evaluation mechanism is introduced to analyze the historical promotion effect data and screen out a rule combination with high reliability.

[0100] In view of the specific content of the promotion rules, this embodiment designs a multi-dimensional strategy combination scheme. The promotion intensity is set by considering the gross profit margin and inventory turnover rate of the product, and a step-by-step discount scheme is adopted. The selection of the promotion period is based on the customer flow forecast results, and the peak customer flow period is given priority. The determination of the target customer group is combined with the matching degree of the product characteristics and the consumer portrait to achieve accurate customer group targeting.

[0101] In the optimization phase of marketing strategy, this embodiment innovatively applies reinforcement learning method. First, the initial marketing strategy plan is converted into a multi-dimensional feature vector, including promotion parameters, time window, customer group characteristics and other elements. By designing a reasonable feature encoding method, the integrity and learnability of the strategy representation are ensured.

[0102] This embodiment adopts a dual network architecture in the construction of the reinforcement learning model. The value network is responsible for evaluating the expected benefits of the strategy, and the strategy network is responsible for generating optimized strategy parameters. Through the Monte Carlo tree search method, multi-step exploration is performed in the strategy space to simulate the execution effects of different strategy combinations. During the exploration process, multiple evaluation indicators are considered, such as sales increase, inventory digestion rate, customer satisfaction, etc., and the strategy effect is comprehensively evaluated in a weighted manner.

[0103] In the process of policy optimization, this embodiment adopts an improved policy gradient algorithm. By designing a reward function based on temporal difference, short-term sales targets and long-term customer value are included in the optimization target. At the same time, an experience replay mechanism is introduced to store historical excellent policy samples to improve learning efficiency. In order to balance the relationship between exploration and utilization, the ε-greedy strategy is used to adaptively adjust the exploration probability.

[0104] The strategy optimization of this embodiment also includes a constraint processing mechanism. By setting a safety threshold, it is ensured that the optimized promotion intensity will not excessively affect the profit of the product. By limiting the time window, it is avoided that the promotion activities are too frequent. By controlling the overlap of customer groups, it is prevented that different promotion activities interfere with each other.

[0105] Through the above innovative solution, this embodiment realizes intelligent marketing strategy optimization. In practical applications, the solution can adaptively generate personalized promotion plans based on product characteristics, market environment and consumer behavior characteristics. Through continuous optimization of reinforcement learning, promotion strategies can continuously adapt to market changes and improve marketing effectiveness. At the same time, the interpretable design of the solution enables merchants to understand and adjust the optimization results, which has strong practicality and controllability.

[0106] This solution has shown significant advantages in actual operation, and can effectively increase product sales, optimize inventory structure, improve customer satisfaction, and create greater business value for merchants. Through continuous data accumulation and model optimization, the accuracy and effectiveness of marketing strategies will continue to improve, forming a virtuous optimization cycle.

[0107] In one embodiment of the smart store data processing method based on multidimensional data analysis of the present application, see Figure 7 , and can also include the following: Step S701: Hash-encrypt the user identification information in the personalized marketing strategy, reversibly encrypt the strategy parameters, desensitize the sensitive product information using data masking technology, write the processed strategy data into multiple storage nodes in the cloud data center in a distributed storage manner, establish a strategy distribution channel based on the distributed message middleware, and push the encrypted marketing strategy data packet to the message queue server; Step S702: Collect sales conversion rate, average order value change and member repurchase rate data of marketing strategies in different time periods from store terminals, parse the collected data into effect evaluation indicators in a preset format, and transmit the effect evaluation indicators back to the model optimization module through the data pipeline to establish sample data pairs containing strategy parameters and execution effects for model parameter updating.

[0108] Optionally, when constructing a personalized marketing strategy, this embodiment first conducts a multi-dimensional analysis based on the product recommendation list and store operation trend data. The rule base constructed by the decision tree algorithm contains key features such as product category, sales forecast, customer flow fluctuation, and consumer portrait. For example, for the high-end cosmetics category, a corresponding promotion plan is formulated in combination with the peak customer flow period on weekends and the consumption capacity of target users; for the daily necessities category, a differentiated marketing strategy is designed based on the customer flow characteristics on weekdays and the characteristics of user groups with high promotion sensitivity.

[0109] In the strategy optimization phase, this embodiment innovatively uses reinforcement learning technology to build a marketing decision optimization model. First, the dimensions of the marketing strategy (promotion intensity, time period selection, target customer group, etc.) are encoded as strategy feature vectors as the state space of the reinforcement learning model. Through the Monte Carlo tree search method, the model can efficiently explore in the huge strategy space and find the optimal strategy combination.

[0110] In order to store and distribute marketing strategies securely, this embodiment designs a multi-level data security protection mechanism. User identification information is irreversibly encrypted using secure hash algorithms such as SHA-256, and policy parameters are reversibly encrypted using the AES encryption algorithm. In the data desensitization process, dynamic masking technology is used to display different degrees of information detail according to the access rights of users at different levels.

[0111] This embodiment uses a distributed storage architecture to ensure high availability and scalability of data. The encrypted policy data is stored in multiple storage nodes through a consistent hashing algorithm to achieve load balancing. At the same time, the Raft protocol is used to ensure data consistency, and the reliability of data is ensured through a multi-copy mechanism.

[0112] In the policy distribution link, this embodiment builds an efficient policy distribution channel based on distributed message middleware. The asynchronous transmission of policy data is realized through the message queue service to ensure reliable message delivery under network fluctuations. The theme subscription mode is adopted to enable different store terminals to receive relevant marketing strategies on demand.

[0113] This embodiment designs a complete effect evaluation and feedback optimization mechanism. The execution effect data of the marketing strategy is collected in real time at the store terminal, including key indicators such as sales conversion rate, change in average customer price and member repurchase rate. These indicators are processed in a standardized manner through a preset data format to ensure data consistency and comparability.

[0114] In the data transmission link, this embodiment builds a reliable data pipeline and realizes the real-time transmission of effect evaluation indicators through streaming processing technology. The incremental synchronization mechanism is adopted to only transmit the changed data to improve the transmission efficiency. At the same time, the network resource utilization is optimized through data compression and batch transmission strategy.

[0115] This embodiment continuously optimizes model performance through online learning. Sample data consisting of strategy parameters and execution effects are input into the model optimization module, and the model parameters are adjusted through gradient updating. The model can adaptively learn the optimal strategy in different scenarios and improve marketing effects.

[0116] Through the above innovative solution, this embodiment realizes intelligent marketing strategy generation and optimization. The solution can dynamically adjust the marketing strategy according to real-time market feedback to ensure the effectiveness and pertinence of the strategy. At the same time, through a strict data security protection mechanism and an efficient distribution system, the safe and reliable execution of the marketing strategy is guaranteed, providing merchants with a full range of marketing support services. In actual application, the solution significantly improves marketing efficiency, enhances user purchasing experience, and promotes the improvement of merchants' operating efficiency.

[0117] In order to break through the limitations of traditional single data analysis and provide a comprehensive data processing solution for the precision marketing of smart stores, the present application provides an embodiment of a smart store data processing device based on multidimensional data analysis for realizing all or part of the contents of the smart store data processing method based on multidimensional data analysis, see Figure 8 The smart store data processing device based on multidimensional data analysis specifically includes the following contents: The multidimensional data fusion module 10 is used to collect multi-channel sensor data, smart terminal device data and social media data in the store, perform time stamp alignment and spatial coordinate mapping on the sensor data and the smart terminal device data to obtain a spatiotemporal feature sequence, perform correlation analysis on the spatiotemporal feature sequence to obtain a feature weight coefficient, and perform multidimensional data fusion on the spatiotemporal feature sequence and the social media data based on the feature weight coefficient to generate a fused data set; The multidimensional data processing module 20 is used to perform data cleaning and standardization processing on the fused data set to obtain sample data, divide the sample data into a training data set and a validation data set according to a preset ratio, use the training data set to train a product recommendation model based on a convolutional neural network and a recurrent neural network, use the validation data set to perform performance evaluation on the trained model, automatically tune the model parameters based on the evaluation results and perform incremental learning updates, use the updated model for product preference prediction to generate a personalized product recommendation list, and perform time series analysis on the fused data set to obtain store operation trend data; The intelligent recommendation module 30 is used to generate an initial marketing strategy plan based on the personalized product recommendation list and the store operation trend data, input the initial marketing strategy plan into a pre-trained marketing decision optimization model to perform strategy evaluation and optimization to obtain a personalized marketing strategy, perform data desensitization and encryption on the personalized marketing strategy and then store it in a cloud data center, distribute the personalized marketing strategy to store terminals for execution through a message queue, and collect marketing strategy execution effect data for model optimization and updating.

[0118] From the above description, it can be seen that the smart store data processing device based on multidimensional data analysis provided by the embodiment of the present application can innovatively realize the alignment and mapping of spatiotemporal feature sequences by collecting sensor, smart terminal and social media data. Multidimensional data fusion is performed based on feature weight coefficients, and a product recommendation model is constructed using convolutional neural networks and recurrent neural networks. The system continuously optimizes model performance through incremental learning to achieve personalized product recommendations and store operation trend analysis. At the same time, a marketing decision optimization model is introduced to intelligently evaluate and dynamically optimize marketing strategies, and the model is continuously updated through feedback data. This method breaks through the limitations of traditional single data analysis and provides a comprehensive data processing solution for the precision marketing of smart stores.

[0119] From the hardware level, in order to break through the limitations of traditional single data analysis and provide a comprehensive data processing solution for the precision marketing of smart stores, the present application provides an embodiment of an electronic device for implementing all or part of the contents of the smart store data processing method based on multi-dimensional data analysis, and the electronic device specifically includes the following contents: Processor, memory, communication interface and bus; wherein the processor, memory and communication interface communicate with each other through the bus; the communication interface is used to realize information transmission between the smart store data processing device based on multidimensional data analysis and the core business system, user terminal and related database and other related equipment; the logic controller can be a desktop computer, a tablet computer and a mobile terminal, etc., but the present embodiment is not limited thereto. In the present embodiment, the logic controller can be implemented with reference to the embodiment of the smart store data processing method based on multidimensional data analysis and the embodiment of the smart store data processing device based on multidimensional data analysis in the embodiment, and the contents thereof are incorporated herein, and the repeated parts are not repeated.

[0120] It is understandable that the user terminal may include a smart phone, a tablet electronic device, a network set-top box, a portable computer, a desktop computer, a personal digital assistant (PDA), a vehicle-mounted device, a smart wearable device, etc. Among them, the smart wearable device may include smart glasses, a smart watch, a smart bracelet, etc.

[0121] In practical applications, part of the smart store data processing method based on multidimensional data analysis can be executed on the electronic device side as described above, or all operations can be completed in the client device. The specific selection can be based on the processing capability of the client device and the limitations of the user's usage scenario. This application does not limit this. If all operations are completed in the client device, the client device may also include a processor.

[0122] The client device may have a communication module (i.e., a communication unit) that can communicate with a remote server to achieve data transmission with the server. The server may include a server on the task scheduling center side, and other implementation scenarios may also include a server on an intermediate platform, such as a server on a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, or a server cluster consisting of multiple servers, or a server structure of a distributed device.

[0123] Fig. 9 FIG. 9 is a schematic block diagram of the system structure of the electronic device 9600 according to an embodiment of the present application. Fig. 9 As shown, the electronic device 9600 may include a central processor 9100 and a memory 9140; the memory 9140 is coupled to the central processor 9100. It is worth noting that Fig. 9 is exemplary; other types of structures may also be used to supplement or replace this structure to implement telecommunication functions or other functions.

[0124] In one embodiment, the function of the smart store data processing method based on multi-dimensional data analysis can be integrated into the central processor 9100. The central processor 9100 can be configured to perform the following control: Step S101: Collect multiple sensor data, smart terminal device data and social media data in the store, perform timestamp alignment and spatial coordinate mapping on the sensor data and the smart terminal device data to obtain a spatiotemporal feature sequence, perform correlation analysis on the spatiotemporal feature sequence to obtain a feature weight coefficient, and perform multi-dimensional data fusion on the spatiotemporal feature sequence and the social media data based on the feature weight coefficient to generate a fused data set; Step S102: performing data cleaning and standardization processing on the fused data set to obtain sample data, dividing the sample data into a training data set and a validation data set according to a preset ratio, using the training data set to train a product recommendation model based on a convolutional neural network and a recurrent neural network, using the validation data set to perform performance evaluation on the trained model, automatically tuning the model parameters based on the evaluation results and performing incremental learning updates, using the updated model to predict product preferences and generate a personalized product recommendation list, and performing time series analysis on the fused data set to obtain store operation trend data; Step S103: Generate an initial marketing strategy plan based on the personalized product recommendation list and the store operation trend data, input the initial marketing strategy plan into a pre-trained marketing decision optimization model for strategy evaluation and optimization to obtain a personalized marketing strategy, perform data desensitization and encryption on the personalized marketing strategy and then store it in a cloud data center, distribute the personalized marketing strategy to store terminals for execution through a message queue, and collect marketing strategy execution effect data for model optimization and update.

[0125] From the above description, it can be seen that the electronic device provided in the embodiment of the present application innovatively realizes the alignment and mapping of spatiotemporal feature sequences by collecting sensor, smart terminal and social media data. Multidimensional data fusion is performed based on the feature weight coefficient, and a product recommendation model is constructed using convolutional neural network and recurrent neural network. The system continuously optimizes the model performance through incremental learning to achieve personalized product recommendations and store operation trend analysis. At the same time, a marketing decision optimization model is introduced to intelligently evaluate and dynamically optimize marketing strategies, and the model is continuously updated through feedback data. This method breaks through the limitations of traditional single data analysis and provides a comprehensive data processing solution for the precision marketing of smart stores.

[0126] In another embodiment, the smart store data processing device based on multidimensional data analysis can be configured separately from the central processing unit 9100. For example, the smart store data processing device based on multidimensional data analysis can be configured as a chip connected to the central processing unit 9100, and the function of the smart store data processing method based on multidimensional data analysis can be realized through the control of the central processing unit.

[0127] like Fig. 9 As shown, the electronic device 9600 may also include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It is worth noting that the electronic device 9600 does not necessarily have to include Fig. 9 In addition, the electronic device 9600 may also include Fig. 9 For components not shown, reference may be made to the prior art.

[0128] like Fig. 9 As shown, the central processing unit 9100 is sometimes also referred to as a controller or an operation control, and may include a microprocessor or other processor device and / or logic device. The central processing unit 9100 receives input and controls the operation of various components of the electronic device 9600.

[0129] The memory 9140 may be, for example, one or more of a cache, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory or other suitable devices. The above-mentioned information related to the failure may be stored, and a program for executing the relevant information may also be stored. The CPU 9100 may execute the program stored in the memory 9140 to implement information storage or processing, etc.

[0130] The input unit 9120 provides input to the central processing unit 9100. The input unit 9120 is, for example, a key or a touch input device. The power supply 9170 is used to provide power to the electronic device 9600. The display 9160 is used to display display objects such as images and texts. The display may be, for example, an LCD display, but is not limited thereto.

[0131] The memory 9140 may be a solid-state memory, such as a read-only memory (ROM), a random access memory (RAM), a SIM card, etc. It may also be a memory that saves information even when the power is off, can be selectively erased, and is provided with more data, examples of which are sometimes referred to as EPROMs, etc. The memory 9140 may also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 may include an application / function storage unit 9142, which is used to store application programs and function programs or processes for executing the operation of the electronic device 9600 through the central processor 9100.

[0132] The memory 9140 may also include a data storage unit 9143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 9144 of the memory 9140 may include various drivers for communication functions of the electronic device and / or for executing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0133] The communication module 9110 is a transmitter / receiver that sends and receives signals via the antenna 9111. The communication module 9110 (transmitter / receiver) is coupled to the central processor 9100 to provide input signals and receive output signals, which may be the same as the case of a conventional mobile communication terminal.

[0134] Based on different communication technologies, multiple communication modules 9110 may be provided in the same electronic device, such as a cellular network module, a Bluetooth module and / or a wireless LAN module. The communication module 9110 (transmitter / receiver) is also coupled to a speaker 9131 and a microphone 9132 via an audio processor 9130 to provide an audio output via the speaker 9131 and receive an audio input from the microphone 9132, thereby realizing a common telecommunication function. The audio processor 9130 may include any suitable buffer, decoder, amplifier, etc. In addition, the audio processor 9130 is also coupled to the central processor 9100, so that recording can be performed on the local machine through the microphone 9132, and the sound stored on the local machine can be played through the speaker 9131.

[0135] The embodiments of the present application also provide a computer-readable storage medium capable of implementing all the steps of the smart store data processing method based on multidimensional data analysis in the above embodiments, where the execution subject is a server or a client. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, all the steps of the smart store data processing method based on multidimensional data analysis in the above embodiments are implemented. For example, when the processor executes the computer program, the following steps are implemented: Step S101: Collect multiple sensor data, smart terminal device data and social media data in the store, perform timestamp alignment and spatial coordinate mapping on the sensor data and the smart terminal device data to obtain a spatiotemporal feature sequence, perform correlation analysis on the spatiotemporal feature sequence to obtain a feature weight coefficient, and perform multi-dimensional data fusion on the spatiotemporal feature sequence and the social media data based on the feature weight coefficient to generate a fused data set; Step S102: performing data cleaning and standardization processing on the fused data set to obtain sample data, dividing the sample data into a training data set and a validation data set according to a preset ratio, using the training data set to train a product recommendation model based on a convolutional neural network and a recurrent neural network, using the validation data set to perform performance evaluation on the trained model, automatically tuning the model parameters based on the evaluation results and performing incremental learning updates, using the updated model to predict product preferences and generate a personalized product recommendation list, and performing time series analysis on the fused data set to obtain store operation trend data; Step S103: Generate an initial marketing strategy plan based on the personalized product recommendation list and the store operation trend data, input the initial marketing strategy plan into a pre-trained marketing decision optimization model for strategy evaluation and optimization to obtain a personalized marketing strategy, perform data desensitization and encryption on the personalized marketing strategy and then store it in a cloud data center, distribute the personalized marketing strategy to store terminals for execution through a message queue, and collect marketing strategy execution effect data for model optimization and update.

[0136] From the above description, it can be seen that the computer-readable storage medium provided in the embodiment of the present application innovatively realizes the alignment and mapping of spatiotemporal feature sequences by collecting sensor, smart terminal and social media data. Multidimensional data fusion is performed based on feature weight coefficients, and a product recommendation model is constructed using convolutional neural networks and recurrent neural networks. The system continuously optimizes model performance through incremental learning to achieve personalized product recommendations and store operation trend analysis. At the same time, a marketing decision optimization model is introduced to intelligently evaluate and dynamically optimize marketing strategies, and the model is continuously updated through feedback data. This method breaks through the limitations of traditional single data analysis and provides a comprehensive data processing solution for the precision marketing of smart stores.

[0137] The embodiments of the present application also provide a computer program product capable of implementing all the steps of the smart store data processing method based on multidimensional data analysis in the above embodiments, where the execution subject is a server or a client. When the computer program / instruction is executed by a processor, the steps of the smart store data processing method based on multidimensional data analysis are implemented. For example, the computer program / instruction implements the following steps: Step S101: Collect multiple sensor data, smart terminal device data and social media data in the store, perform timestamp alignment and spatial coordinate mapping on the sensor data and the smart terminal device data to obtain a spatiotemporal feature sequence, perform correlation analysis on the spatiotemporal feature sequence to obtain a feature weight coefficient, and perform multi-dimensional data fusion on the spatiotemporal feature sequence and the social media data based on the feature weight coefficient to generate a fused data set; Step S102: performing data cleaning and standardization processing on the fused data set to obtain sample data, dividing the sample data into a training data set and a validation data set according to a preset ratio, using the training data set to train a product recommendation model based on a convolutional neural network and a recurrent neural network, using the validation data set to perform performance evaluation on the trained model, automatically tuning the model parameters based on the evaluation results and performing incremental learning updates, using the updated model to predict product preferences and generate a personalized product recommendation list, and performing time series analysis on the fused data set to obtain store operation trend data; Step S103: Generate an initial marketing strategy plan based on the personalized product recommendation list and the store operation trend data, input the initial marketing strategy plan into a pre-trained marketing decision optimization model for strategy evaluation and optimization to obtain a personalized marketing strategy, perform data desensitization and encryption on the personalized marketing strategy and then store it in a cloud data center, distribute the personalized marketing strategy to store terminals for execution through a message queue, and collect marketing strategy execution effect data for model optimization and update.

[0138] From the above description, it can be seen that the computer program product provided in the embodiment of the present application innovatively realizes the alignment and mapping of spatiotemporal feature sequences by collecting sensor, smart terminal and social media data. Multidimensional data fusion is performed based on feature weight coefficients, and a product recommendation model is constructed using convolutional neural networks and recurrent neural networks. The system continuously optimizes model performance through incremental learning to achieve personalized product recommendations and store operation trend analysis. At the same time, a marketing decision optimization model is introduced to intelligently evaluate and dynamically optimize marketing strategies, and the model is continuously updated through feedback data. This method breaks through the limitations of traditional single data analysis and provides a comprehensive data processing solution for the precision marketing of smart stores.

[0139] It should be understood by those skilled in the art that embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0140] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (apparatus), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0141] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0142] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0143] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A smart store data processing method based on multidimensional data analysis, characterized in that: The method comprises: Collecting multiple sensor data, smart terminal device data and social media data in the store, performing timestamp alignment and spatial coordinate mapping on the sensor data and the smart terminal device data to obtain a spatiotemporal feature sequence, performing correlation analysis on the spatiotemporal feature sequence to obtain a feature weight coefficient, and performing multi-dimensional data fusion on the spatiotemporal feature sequence and the social media data based on the feature weight coefficient to generate a fused data set; The fused data set is cleaned and standardized to obtain sample data, the sample data is divided into a training data set and a validation data set according to a preset ratio, the training data set is used to train a product recommendation model based on a convolutional neural network and a recurrent neural network, the validation data set is used to perform performance evaluation on the trained model, the model parameters are automatically tuned based on the evaluation results and incremental learning updates are performed, the updated model is used for product preference prediction to generate a personalized product recommendation list, and the fused data set is subjected to time series analysis to obtain store operation trend data; An initial marketing strategy plan is generated based on the personalized product recommendation list and the store operation trend data, and the initial marketing strategy plan is input into a pre-trained marketing decision optimization model for strategy evaluation and optimization to obtain a personalized marketing strategy. The personalized marketing strategy is desensitized and encrypted before being stored in a cloud data center. The personalized marketing strategy is distributed to store terminals for execution through a message queue, and marketing strategy execution effect data is collected for model optimization and updating.

2. The smart store data processing method based on multidimensional data analysis according to claim 1 is characterized in that: The step of performing timestamp alignment and spatial coordinate mapping on the sensor data and the smart terminal device data to obtain a spatiotemporal feature sequence, and performing correlation analysis on the spatiotemporal feature sequence to obtain a feature weight coefficient includes: Establish a spatiotemporal reference coordinate system based on the collection time and device location information in the sensor data, map the sensor data to the time axis according to the timestamp marked with the collection time, convert the device number in the smart terminal device data into a corresponding spatial coordinate point and map it to the spatial axis, and resample the data with different sampling frequencies through bilinear interpolation to generate a spatiotemporal feature sequence; The Pearson correlation coefficient is used to calculate the correlation between the features in the spatiotemporal feature sequence, a feature correlation matrix is ​​constructed, the contribution of each feature to the target variable is calculated based on the feature correlation matrix, and the contribution is normalized to obtain a feature weight coefficient.

3. The smart store data processing method based on multidimensional data analysis according to claim 1 is characterized in that: The performing multi-dimensional data fusion on the spatiotemporal feature sequence and the social media data based on the feature weight coefficient to generate a fused data set includes: The feature weight coefficient is weighted and calculated with the corresponding spatiotemporal feature sequence to obtain a weighted feature value, the social media data is subjected to topic classification and sentiment analysis to obtain a text feature vector, and the weighted feature value is aligned with the text feature vector in the time dimension according to the timestamp mark in the spatiotemporal feature sequence to construct a multidimensional feature tensor; A tensor decomposition algorithm is used to reduce the dimension of the multidimensional feature tensor to obtain a feature subspace, and a local linear embedding algorithm is used to map the heterogeneous data in the feature subspace to a unified feature space to generate a fused data set.

4. The smart store data processing method based on multidimensional data analysis according to claim 1 is characterized in that: The fusion data set is cleaned and standardized to obtain sample data, the sample data is divided into a training data set and a verification data set according to a preset ratio, the training data set is used to train a product recommendation model based on a convolutional neural network and a recurrent neural network, the verification data set is used to perform performance evaluation on the trained model, and the model parameters are automatically tuned and incrementally updated based on the evaluation results, including: The missing values ​​in the fused data set are supplemented by the K nearest neighbor interpolation method, the outlier detection algorithm is used to remove abnormal data points, the numerical features are normalized to the maximum and minimum values, and the categorical features are converted to one-hot encoding to obtain sample data, and the sample data are randomly divided into a training data set and a validation data set in a ratio of 8:2; A convolutional neural network is used to extract spatial features in the training data set, and the extracted spatial features are input into a recurrent neural network for time series modeling. The network parameters are optimized by a back propagation algorithm to obtain an initial product recommendation model. The accuracy, recall rate and F1 score of the model are calculated using the validation data set. The learning rate, batch size and number of network layers of the initial product recommendation model are tuned based on a grid search algorithm, and the sliding window method is used to perform incremental training on newly added data samples to update the model parameters.

5. The smart store data processing method based on multidimensional data analysis according to claim 1 is characterized in that: The updated model is used for product preference prediction to generate a personalized product recommendation list, and the fused data set is subjected to time series analysis to obtain store operation trend data, including: Input the user's historical behavior data into the updated product recommendation model, extract the user's interest feature vector, calculate the similarity matrix between the user and the product through the collaborative filtering algorithm, sort the products according to the similarity score and select the products with the highest score to build a personalized recommendation list; The fused data set is divided into time series according to hours and dates, and the exponential smoothing method is used to perform trend decomposition on the customer flow and sales data. The cyclical and seasonal characteristics in the data are extracted through the autoregressive model, and the decomposed time series data is modeled and analyzed with the long short-term memory network to obtain the store operation trend data.

6. The method for processing smart store data based on multidimensional data analysis according to claim 1, characterized in that: The generating of an initial marketing strategy plan based on the personalized product recommendation list and the store operation trend data, and inputting the initial marketing strategy plan into a pre-trained marketing decision optimization model for strategy evaluation and optimization to obtain a personalized marketing strategy, includes: Based on the commodity categories and sales forecast values ​​in the personalized commodity recommendation list, combined with the customer flow fluctuation pattern and consumer portrait distribution in the store operation trend data, a decision tree algorithm is used to build a commodity promotion rule library, and an initial marketing strategy plan including promotion intensity, promotion period and target customer group is generated through rule matching; The initial marketing strategy plan is converted into a strategy feature vector and input into a marketing decision optimization model built based on reinforcement learning. The strategy parameters are explored through the Monte Carlo tree search method, the strategy benefit value is calculated based on historical marketing effect data, and the strategy gradient algorithm is used to maximize the benefit value to obtain a personalized marketing strategy.

7. The smart store data processing method based on multidimensional data analysis according to claim 1 is characterized in that: The personalized marketing strategy is desensitized and encrypted and then stored in a cloud data center, the personalized marketing strategy is distributed to the store terminal for execution through a message queue, and the marketing strategy execution effect data is collected for model optimization and update, including: Hash encryption is performed on the user identification information in the personalized marketing strategy, the policy parameters are reversibly encrypted, the sensitive product information is desensitized using data masking technology, the processed policy data is written to multiple storage nodes in the cloud data center in a distributed storage manner, a policy distribution channel is established based on the distributed message middleware, and the encrypted marketing policy data packet is pushed to the message queue server; Collect sales conversion rate, average order value change and member repurchase rate data of marketing strategies in different time periods from store terminals, parse the collected data into effect evaluation indicators in a preset format, and transmit the effect evaluation indicators back to the model optimization module through the data pipeline to establish sample data pairs containing strategy parameters and execution effects for model parameter updating.

8. A smart store data processing device based on multidimensional data analysis, characterized in that: The device comprises: A multidimensional data fusion module is used to collect multi-channel sensor data, smart terminal device data and social media data in the store, perform timestamp alignment and spatial coordinate mapping on the sensor data and the smart terminal device data to obtain a spatiotemporal feature sequence, perform correlation analysis on the spatiotemporal feature sequence to obtain a feature weight coefficient, and perform multidimensional data fusion on the spatiotemporal feature sequence and the social media data based on the feature weight coefficient to generate a fused data set; A multidimensional data processing module, for performing data cleaning and standardization processing on the fused data set to obtain sample data, dividing the sample data into a training data set and a validation data set according to a preset ratio, using the training data set to train a product recommendation model based on a convolutional neural network and a recurrent neural network, using the validation data set to perform performance evaluation on the trained model, automatically tuning model parameters based on the evaluation results and performing incremental learning updates, using the updated model to predict product preferences to generate a personalized product recommendation list, and performing time series analysis on the fused data set to obtain store operation trend data; An intelligent recommendation module is used to generate an initial marketing strategy plan based on the personalized product recommendation list and the store operation trend data, input the initial marketing strategy plan into a pre-trained marketing decision optimization model for strategy evaluation and optimization to obtain a personalized marketing strategy, perform data desensitization and encryption on the personalized marketing strategy and then store it in a cloud data center, distribute the personalized marketing strategy to store terminals for execution through a message queue, and collect marketing strategy execution effect data for model optimization and update.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the smart store data processing method based on multidimensional data analysis described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the smart store data processing method based on multidimensional data analysis described in any one of claims 1 to 7 are implemented.

Images