E-commerce platform information analysis method and system based on artificial intelligence

Abstract

The invention discloses an electronic commerce platform information analysis method and system based on artificial intelligence, and relates to the technical field of information analysis, and the method comprises the steps: obtaining user behavior data based on an electronic commerce platform, calculating prior distribution based on a user behavior change rate, and combining with initial likelihood distribution calculated through a probability density function, constructing an objective function to optimally calculate short-term, medium-term and long-term behavior windows, calculating power spectral density of a time sequence to perform clustering analysis, optimally calculating the three windows, and determining different characteristics of user behaviors through kernel density estimation; a multi-head attention mechanism is adopted to fuse different features of the user. According to the method, the time sequence is converted into the frequency domain space through calculation of the power spectral density, the main periodic mode of the user behavior is revealed, behavior prediction is more interpretable, and optimization can be carried out according to the actual frequency characteristics of the user behavior by constructing the frequency and power density distribution of the power spectral density.

Description

Technical Field

[0001] The present invention relates to the technical field of information analysis, and particularly to an information analysis method and system for an e-commerce platform based on artificial intelligence. Background Art

[0002] With the booming development of e-commerce, user behavior analysis has become one of the key technologies for improving the user experience of e-commerce platforms and optimizing business decisions. Analysis methods based on user behavior data have been continuously improved. Especially in the application of recommendation systems, content-based recommendation methods focus on analyzing the attribute information of goods and making recommendations based on the degree of matching with user interests. However, they also lack the precise capture of long-term and short-term changes in users' dynamic interests. Therefore, how to deeply analyze user behavior patterns and optimize recommendation systems to better adapt to changes in user needs has become a current research hotspot; Although existing technical solutions have solved the problems of user behavior modeling and optimization of recommendation systems to a certain extent, there are still many limitations. Traditional time window partitioning methods usually use fixed time windows to analyze user behavior changes. However, this method fails to fully consider the dynamic characteristics and multi-scale changes of user behavior, resulting in insufficient flexibility and accuracy in analyzing user interests. In addition, existing recommendation systems mostly rely on single-dimensional user interest modeling, lacking the comprehensive integration of different user characteristics and multi-level information extraction, resulting in limited improvement in personalized recommendation effects. Especially when dealing with short-term, medium-term, and long-term user behaviors, traditional methods often ignore the dynamic adjustment of behavior change rates and fail to accurately capture user interest fluctuations at different time scales. Summary of the Invention

[0003] In view of the above existing problems, the present invention is proposed.

[0004] Therefore, the present invention provides an information analysis method for an e-commerce platform based on artificial intelligence to solve the problems that the dynamic characteristics and multi-scale changes of user behavior are not fully considered, resulting in insufficient flexibility and accuracy in analyzing user interests. In addition, existing recommendation systems mostly rely on single-dimensional user interest modeling, lacking the comprehensive integration of different user characteristics and multi-level information extraction, resulting in limited improvement in personalized recommendation effects. Especially when dealing with short-term, medium-term, and long-term user behaviors, traditional methods often ignore the dynamic adjustment of behavior change rates and fail to accurately capture user interest fluctuations at different time scales.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides an information analysis method for an e-commerce platform based on artificial intelligence, which includes: Obtain user behavior data based on an e-commerce platform, calculate the prior distribution based on the user behavior change rate, combine it with the preliminary likelihood distribution calculated through the probability density function, construct an objective function to optimize the calculation of short-term, medium-term, and long-term behavior windows, calculate the power spectral density of the time series for clustering analysis, optimize the calculation of the three windows, and determine different characteristics of user behavior through kernel density estimation; Adopt a multi-head attention mechanism to fuse different characteristics of users, construct a user interest graph, calculate the partial correlation coefficient using multiple regression analysis, and use the Peter-Clark algorithm to calculate the causal weight between users and commodities, and update the user interest characteristics through heterogeneous GNN; Determine the interest distribution of users for commodities, calculate the causal effect value through user interest characteristics, and perform the final recommendation score calculation; Generate a commodity recommendation list, and perform encryption processing and secure transmission on the data.

[0006] As a preferred solution of the information analysis method for an e-commerce platform based on artificial intelligence according to the present invention, wherein: the determination of different characteristics of user behavior includes constructing a variational Bayesian time window VBTW model based on user behavior data, calculating the user behavior change rate at different times based on the window size, and performing change rate segmentation of the time window on the change rate sequence, calculating the mean and variance for each change rate window, and using the variational Bayesian method to calculate the prior distribution of the time window; Calculate the preliminary likelihood distribution through the probability density function, use the product of the prior distribution and the preliminary likelihood distribution as the objective function, calculate the objective function value for each candidate time window, and use the gradient ascent method to optimize and maximize the objective function; According to the branches of the objective function in different window intervals, respectively determine the smallest optimal window as the short-term behavior window, determine the largest optimal window as the long-term behavior window, and take the mean of the sum of the long-term behavior window and the short-term behavior window as the medium-term behavior window; Use the autoregressive AR model to model user behavior data, calculate the autocorrelation function ACF of the time series, and calculate the power spectral density through Fourier transform; Construct the frequency and power density based on the calculation of the power spectral density, use K-Means clustering for clustering analysis, use the long-term behavior window, the short-term behavior window, and the medium-term behavior window as the clustering targets, optimize the calculation by calculating the loss function, and optimize the calculation with the three behavior windows as the number of clusters, and respectively extract the windows at the maximum frequency, minimum frequency, and medium frequency of the change frequency f as the optimized short-term window, long-term window, and medium-term window; Adopt a Gaussian kernel to calculate the kernel density estimation KDE of the short-term window, long-term window, and medium-term window, which are respectively used as the short-term, medium-term, and long-term characteristics of user behavior.

[0007] As a preferred solution of the information analysis method for an e-commerce platform based on artificial intelligence according to the present invention, wherein: updating the user interest characteristics includes using a Transformer to adopt a multi-head attention mechanism to fuse the short-term characteristics, medium-term characteristics and long-term characteristics of the user as comprehensive behavior characteristics; Taking the comprehensive behavior characteristics as user nodes, the commodity texts, images and historical interaction behavior information of user interactions as commodity nodes, and the category, brand and price range information of the commodity as attribute nodes, and taking the relationships between user nodes and commodity nodes, commodity nodes and attribute nodes, and user nodes and attribute nodes as edges, to construct a user interest graph; Based on the edges between user nodes and commodity nodes, respectively determine the comprehensive behavior characteristics as user behavior data U, the commodity texts, images and historical interaction behavior information as commodity feature data, and the promotion influence and advertisement exposure times as control variable data F; Adopt multiple regression analysis to calculate the partial correlation coefficient, and adopt the Peter-Clark algorithm to calculate the conditional partial correlation coefficient of the edge between the user node and the commodity node, and take the conditional partial correlation coefficient as the causal weight of the edge between the user node and the commodity node; Propagate the relationships of users, commodities and attributes through heterogeneous GNNs. The message passing between the user node c and the commodity node v is carried out through the causal weight and attention mechanism of the edge between the user node and the commodity node to update the user interest representation.

[0008] As a preferred solution of the information analysis method for an e-commerce platform based on artificial intelligence according to the present invention, wherein: calculating the final recommendation score includes, based on the updated user interest characteristics, using Softmax normalization to calculate the interest distribution of the user for each commodity attribute, and calculating the causal effect; Calculate the final recommendation score by combining the product of the causal effect value of a single commodity and the similarity between the user interest and the commodity with the ratio of the remaining commodities.

[0009] As a preferred solution of the information analysis method for an e-commerce platform based on artificial intelligence according to the present invention, wherein: generating a commodity recommendation list includes sorting the recommendation scores of all commodities in descending order to generate a final recommendation list, and at the same time generating a commodity classification recommendation list based on the attribute node information of the commodity, including the category, brand and price range information of the commodity.

[0010] As a preferred solution of the information analysis method for an e-commerce platform based on artificial intelligence according to the present invention, wherein: the user behavior data is obtained based on the e-commerce platform, including obtaining the behavioral time series data of the user based on the e-commerce platform, and collecting the behavioral data of the user within the time window, including the number of browsing behaviors, the number of click behaviors, and the number of purchase behaviors, performing data standardization processing, and constructing a user behavior matrix, where each row corresponds to the user behavior data at different times within the time window.

[0011] As a preferred solution of the information analysis method for an e-commerce platform based on artificial intelligence according to the present invention, wherein: the data is encrypted and securely transmitted, including encrypting and storing the recommendation list data and the user interest feature data of the user through AES-256, managing user permissions using RBAC, and performing secure data transmission using HTTPS, and adding noise to the user behavior data for privacy storage.

[0012] In a second aspect, the present invention provides a system for an information analysis method for an e-commerce platform based on artificial intelligence, including a user behavior data processing module, which obtains the browsing, clicking, purchasing and other behavior data of the user from the e-commerce platform, and calculates the user behavior change rate; a time window optimization module, which constructs an objective function based on the prior distribution and the preliminary likelihood distribution, calculates the short-term, medium-term and long-term behavior windows, and optimizes the window division through power spectral density analysis and clustering; a user behavior feature extraction module, which calculates the user behavior features under different time windows using kernel density estimation, and fuses the short-term, medium-term and long-term behavior patterns through a multi-head attention mechanism; a user interest graph construction module, which constructs a heterogeneous graph containing the relationships between users, products and attributes based on the user behavior features, and calculates the causal weights between users and products; a GNN propagation module, which calculates the partial correlation coefficient using multiple regression analysis, calculates the causal weights using the Peter-Clark algorithm, and updates the user interest features through heterogeneous GNN propagation; a recommendation score calculation module, which calculates the recommendation scores of the user for the products based on the user interest distribution and the causal effect values, and performs normalization processing to ensure the rationality of sorting, sorts the calculated recommendation scores, and generates a recommendation list; a data security protection module, which adds noise to the data to protect privacy.

[0013] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and wherein: when the computer program is executed by the processor, any step of the information analysis method for an e-commerce platform based on artificial intelligence as described in the first aspect of the present invention is implemented.

[0014] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the information analysis method for an e-commerce platform based on artificial intelligence as described in the first aspect of the present invention is implemented.

[0015] The beneficial effects of the present invention are as follows: By calculating the power spectral density, the time series is transformed into the frequency domain space, revealing the main periodic patterns of user behavior, making the behavior prediction more interpretable. By constructing the frequency and power density distributions of the power spectral density, optimization can be carried out according to the actual frequency characteristics of user behavior, ensuring the scientificity and rationality of behavior analysis. When calculating the kernel density estimation for the extracted short, medium, and long-term behavior windows, due to the true probability distribution of the data, it is avoided to assume that user behavior must follow a certain specific distribution. By combining the causal effect value of a single commodity with the user interest and commodity similarity, and normalizing the ratio with all candidate commodities, it can be ensured that the final recommendation score not only considers the causal relationship but also the matching degree between user interest and the commodity itself. Description of the Drawings

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0017] Figure 1 It is a schematic flowchart of the information analysis method for an e-commerce platform based on artificial intelligence in Embodiment 1.

[0018] Figure 2 It is a schematic structural diagram of the information analysis system for an e-commerce platform based on artificial intelligence in Embodiment 1. Detailed Embodiments

[0019] To make the above objects, features, and advantages of the present invention more obvious and understandable, the detailed embodiments of the present invention will be described in detail below with reference to the drawings in the specification.

[0020] Many specific details are set forth in the following description in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0021] Second, the "one embodiment" or "embodiment" referred to herein means a specific feature, structure, or characteristic that may be included in at least one implementation manner of the present invention. The "in one embodiment" that appears in different places in this specification does not all refer to the same embodiment, nor is it an embodiment that is separate or selectively mutually exclusive with other embodiments.

[0022] Embodiment 1, referring to From Figure 1 to Figure 2 , is the first embodiment of the present invention. This embodiment provides an information analysis method for an e-commerce platform based on artificial intelligence, including the following steps: S1. Obtain user behavior data based on the e-commerce platform, calculate the prior distribution based on the user behavior change rate, combine it with the preliminary likelihood distribution calculated through the probability density function, construct an objective function to optimize and calculate short-term, medium-term, and long-term behavior windows, calculate the power spectral density of the time series for clustering analysis, optimize and calculate the three windows, and determine different characteristics of user behavior through kernel density estimation; Preferably, obtaining user behavior data based on the e-commerce platform includes obtaining the user's behavior time series data based on the e-commerce platform, and collecting the user's behavior data within the time window, including the number of browsing behaviors, the number of click behaviors, and the number of purchase behaviors, performing data standardization processing, and constructing a user behavior matrix, where each row corresponds to the user behavior data at different times within the time window By obtaining the user's behavior time series data based on the e-commerce platform and collecting the user's behavior data within the time window, including the number of browsing, clicking, and purchasing times, the changing trend of the user's interests can be comprehensively captured, enabling the recommendation system to perform personalized optimization based on the user's actual behavior pattern. Through data standardization processing, the differences on different user behavior scales can be eliminated, avoiding data deviation caused by different user activity levels, making the behavior data of different users comparable on the same scale, improving the stability and generalization ability of the model. By constructing a user behavior matrix, the user behavior data can be input into the subsequent modeling and analysis processes in a structured manner, ensuring that the recommendation system can efficiently process time series information, thereby improving the accuracy of recommendations and the accuracy of user interest prediction.

[0023] Preferably, determining different characteristics of user behavior includes constructing a variational Bayesian time window VBTW model based on user behavior data, calculating the user behavior change rate at different times based on the window size, performing change rate segmentation of the time window for the change rate sequence, calculating the mean and variance for each change rate window, and using the variational Bayesian method to calculate the prior distribution of the time window, expressed as: ; Where represents the prior distribution of the change rate window, Represents the calculation of the Gaussian prior distribution, Represents the mean of the rate-of-change window, Represents the variance of the rate-of-change window, Represents the regularization term, where Represents the penalty factor for window growth, usually set to a small constant; Through the calculation of the probability density function as the preliminary likelihood distribution, and taking the product of the prior distribution and the preliminary likelihood distribution as the objective function, and calculating the objective function value of each candidate time window, using the gradient ascent method to optimize and maximize the objective function, expressed as: ; ; Where Represents the user behavior data at time t, Represents the maximum time step, Represents the optimal time window function value, Represents the probability of observing the behavior under the given window W; According to the branches of the objective function in different window intervals, respectively determine the smallest optimal window as the short-term behavior window, determine the largest optimal window as the long-term behavior window, and take the mean of the sum of the long-term behavior window and the short-term behavior window as the medium-term behavior window; Use the autoregressive AR model to model the user behavior data, (predict the current behavior based on past data), and calculate the autocorrelation function ACF of the time series, and calculate the power spectral density through Fourier transform, expressed as: ; ; ; Where Represents the autoregressive order, Represents the autoregressive coefficient (confirmed by the least squares method), Represents the user behavior data at time t-i, Represents the error term, (can be assumed to be Gaussian noise), Represents the autoregressive order at lag k time, Represents the global mean of the sequence, Represents the power density at different behavior change frequencies f, j represents the imaginary unit; Construct the frequency and power density based on the calculation of the power spectral density, perform clustering analysis using K-Means clustering, use the long-term behavior window, short-term behavior window, and medium-term behavior window as the clustering targets, perform optimization calculations by calculating the loss function, and perform optimization calculations with the three behavior windows as the number of clusters. Extract the windows at the maximum frequency, minimum frequency, and medium frequency of the changing frequency f respectively, and use them as the optimized short-term window, long-term window, and medium-term window respectively; Calculate the kernel density estimation KDE of the short-term window, long-term window, and medium-term window using a Gaussian kernel, and use them as the short-term feature, medium-term feature, and long-term feature of user behavior respectively, which are expressed as: ; where represents the kernel density estimation value of the observed behavior data under the fixed window W, represents the number of historical data points within the time window, represents the bandwidth parameter (the optimal bandwidth is determined using the Silverman rule), K represents the kernel function, represents the i-th historical data point within the time window.

[0024] By calculating the user behavior change rate, avoid the information loss or lag problems caused by using a fixed time window. Calculate the prior distribution of the time window using the variational Bayesian method to effectively avoid the instability of time window partitioning caused by data sparsity or abnormal distribution, and improve the robustness of the model. By combining the prior information with the likelihood distribution of the observed data, a more reasonable window partitioning strategy can be provided to ensure that the window selection is not interfered by individual abnormal behaviors. Using the autoregressive (AR) model can extract the temporal dependence of user behavior, and calculating the autocorrelation function can measure the correlation of behavior data at different lag times; Transform the time series into the frequency domain space through the calculation of the power spectral density, reveal the main periodic patterns of user behavior, and make behavior prediction more interpretable. By constructing the frequency and power density distribution of the power spectral density and using K-Means clustering to optimize the time window, the time window partitioning no longer depends on fixed rules, but can be optimized according to the actual frequency characteristics of user behavior, ensuring the scientificity and rationality of behavior analysis. Make the extracted short-term, medium-term, and long-term behavior windows the true probability distribution of the data under the calculation of kernel density estimation, and avoid assuming that user behavior must follow a certain specific distribution.

[0025] S2. Use the multi-head attention mechanism to fuse different features of the user, construct a user interest graph, calculate the partial correlation coefficient using multiple regression analysis, and use the Peter-Clark algorithm to calculate the causal weight between the user and the commodity, and update the user interest features through heterogeneous GNN; Preferably, updating the user interest characteristics includes using a Transformer with a multi-head attention mechanism to fuse the short-term, medium-term, and long-term characteristics of the user as comprehensive behavior characteristics; Taking the comprehensive behavior characteristics as user nodes, the commodity texts, images, and historical interaction behavior information of user interactions as commodity nodes, and the category, brand, and price range information of the commodity as attribute nodes, and taking the relationships between user nodes and commodity nodes, commodity nodes and attribute nodes, and user nodes and attribute nodes as edges, construct a user interest graph; Based on the edges between user nodes and commodity nodes, respectively determine the comprehensive behavior characteristics as user behavior data U, the commodity texts, images, and historical interaction behavior information as commodity feature data, and the promotion impact and advertisement exposure times as control variable data F; Use multiple regression analysis to calculate the partial correlation coefficient, and use the Peter-Clark algorithm to calculate the conditional partial correlation coefficient of the edge between the user node and the commodity node, and take the conditional partial correlation coefficient as the causal weight of the edge between the user node and the commodity node, expressed as: ; ; Where represents the partial correlation coefficient of user behavior data, commodity feature data, and control variable data, , and respectively represent the covariance of user behavior and commodity features, user behavior and a single control variable, and a single control variable and commodity features, and respectively represent the variances of user behavior and commodity features, represents the conditional partial correlation coefficient of user behavior data, commodity feature data, and control variable data, represents the sample size, represents the number of control variable data; Propagate the relationships of users, commodities, and attributes through heterogeneous GNNs. The causal weight and attention mechanism of the edge between the user node c and the commodity node v are used for message passing to update the user interest representation, expressed as: ; ; Where represents the transposed calculation of the feature data of the user node, represents the similarity between the user node and the commodity node calculated based on the trainable attention weight matrix G, represents the adjacent commodity node, Represents the updated user interest characteristics, Represents the ReLU activation function, Represents the set of adjacent products of the user node, that is, all products that have interacted with user c, Represents the attention weight between the user node and the product node, Represents the causal weight between the user node and the product node, Represents the feature data of the user node, Represents the feature data of the product node.

[0026] Fuse user features through Transformer, take the comprehensive behavior features as user nodes, and construct product nodes by combining the text, images and historical interaction behavior information of products. At the same time, construct attribute nodes using the category, brand, and price range information of products, so that the user interest graph can contain the complete user-product-attribute ternary relationship, improve the expression ability of the graph structure, and enable the recommendation system to still have strong generalization ability in the cold start scenario. In addition, introducing product attribute information as attribute nodes enables the model to learn more fine-grained product features, thereby improving the interpretability and accuracy of recommendations; Based on using comprehensive behavior features as user behavior data, the text, images and historical interaction behaviors of products as product feature data, and introducing promotion influence and advertisement exposure times as control variables, causal inference can remove external interference factors, ensuring that the user interest expression learned by the model is driven by real user preferences rather than short-term interference from external factors, thereby reducing the bias of the recommendation system and improving the recommendation quality; The calculation method of the conditional partial correlation coefficient can quantify the influence degree of different control variables on user behavior and product features, so as to improve the long-term benefits of the recommendation system. Message passing is carried out through the causal weight and attention mechanism between users and products in the heterogeneous GNN to improve the accuracy of user interest modeling. At the same time, the attention mechanism can dynamically adjust the intensity of information propagation according to the weights of different user-product relationships, enabling the model to more accurately capture the key features of user interests.

[0027] The attention weight between the user node and the product node is calculated by a trainable attention weight matrix, enabling the information transmission between different user-product pairs to be weighted according to similarity rather than average propagation, thereby improving the model's ability to capture changes in user interests. The introduction of causal weights enables the model to model the relationship between user behavior and product features as a real causal relationship, rather than just a simple matching relationship between covariates, thus ensuring that the recommendation system can provide more interpretable recommendation results.

[0028] S3. Determine the interest distribution of users for products, calculate the causal effect value through user interest characteristics, and perform the calculation of the final recommendation score; Preferably, calculating the final recommendation score includes, based on the updated user interest characteristics, using Softmax normalization to calculate the interest distribution of users for each product attribute , representing the interest probability distribution of user c for the specific attribute a in the attribute node A of product v, and calculating the causal effect, expressed as: ; Where represents the causal effect value between the user and the product, represents the probability that product node v is purchased by behavior L when user c has a certain interest a; Combine the product of the causal effect value of a single product and the similarity between user interest and the product with the ratio of the remaining products to perform the calculation of the final recommendation score, expressed as: ; Where represents the recommendation score of user c for product node v, represents the causal effect value between the user and the adjacent product, represents the cosine similarity calculation.

[0029] By using Softmax normalization to calculate the interest distribution of users for each product attribute based on the updated user interest characteristics, the preferences of users for different product attributes can be quantified, so as to ensure that the recommendation system can accurately understand the user's interest structure. By calculating the causal effect value of users for products, the model not only depends on historical interaction data, but can combine the potential interest preferences of users to measure the attractiveness of products to users, so as to eliminate the interest deviation problem caused by the exposure mechanism of the recommendation system; By combining the product of the causal effect value of a single product and the similarity between user interest and the product, and normalizing its ratio with all candidate products, it can be ensured that the final recommendation score not only considers the causal relationship, but also considers the matching degree between user interest and the product itself. By using the cosine similarity to calculate the matching degree between user interest and product characteristics, it can be ensured that the model will not cause recommendation deviation due to the high causal effect of a certain product, but can comprehensively consider the overall interest characteristics of users, so that the recommendation result achieves a balance between accuracy and personalization.

[0030] S4. Generate a product recommendation list, and perform encryption processing and secure transmission on the data; Preferably, a product recommendation list is generated, including sorting the recommendation scores of all products in descending order to generate a final recommendation list, and at the same time generating a product classification recommendation list based on the attribute node information of the products, including the category, brand, and price range information of the products.

[0031] By sorting the recommendation scores of all products in descending order, it can be ensured that the final recommendation list preferentially includes the products that best match the user's interests, thereby improving the accuracy of recommendations and user satisfaction. Based on the attribute node information of the products, including the category, brand, and price range information of the products, a product classification recommendation list is generated, enabling the recommendation system to provide more structured and diverse recommendation results to meet the preference needs of different users.

[0032] Furthermore, data encryption processing and secure transmission are performed, including encrypting and storing the recommendation list data and the user interest feature data of users through AES-256, managing user permissions using RBAC, and performing secure data transmission using HTTPS, and adding noise to the user behavior data for privacy storage.

[0033] By encrypting and storing the recommendation list data and the user interest feature data through AES-256, it is ensured that sensitive data is not leaked during storage, improving the data security of the system. Using RBAC (Role-Based Access Control) to manage user permissions can effectively limit the access scope of different levels of users to data. Using HTTPS (TLS encryption protocol) for secure data transmission can prevent data from being intercepted, tampered with, or forged during transmission. By adding noise to the user behavior data for privacy storage, the personal behavior patterns of users can be protected and the individual identity can be prevented from being inferred.

[0034] This embodiment also provides a system for an information analysis method of an e-commerce platform based on artificial intelligence, including A user behavior data processing module that obtains the browsing, clicking, purchasing, and other behavior data of users from the e-commerce platform and calculates the user behavior change rate; A time window optimization module that constructs an objective function based on the prior distribution and the preliminary likelihood distribution, calculates short-term, medium-term, and long-term behavior windows, and optimizes the window division through power spectral density analysis and clustering; A user behavior feature extraction module that calculates the user behavior features under different time windows using kernel density estimation and fuses short-term, medium-term, and long-term behavior patterns through a multi-head attention mechanism; A user interest graph construction module that constructs a heterogeneous graph containing the relationships between users, products, and attributes based on the user behavior features and calculates the causal weights of user-product; The GNN propagation module calculates the partial correlation coefficient using multiple regression analysis, calculates the causal weight using the Peter-Clark algorithm, and updates the user interest feature through heterogeneous GNN propagation; The recommended score calculation module calculates the recommended score of the user for the product based on the user interest distribution and the causal effect value, and performs normalization to ensure the rationality of sorting. The calculated recommended scores are sorted to generate a recommended list; The data security protection module adds noise to the data to protect privacy.

[0035] This embodiment also provides a computer device, which is applicable to the case of the information analysis method of an e-commerce platform based on artificial intelligence, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the information analysis method of the e-commerce platform based on artificial intelligence as proposed in the above embodiment.

[0036] The computer device can be a terminal. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be achieved through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the outer shell of the computer device, or an external keyboard, a touchpad, or a mouse, etc.

[0037] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for analyzing e-commerce platform information based on artificial intelligence as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, abbreviated as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, abbreviated as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, abbreviated as EPROM), programmable read-only memory (Programmable Red-Only Memory, abbreviated as PROM), read-only memory (Read-Only Memory, abbreviated as ROM), magnetic memory, flash memory, magnetic disk or optical disc.

[0038] In summary, the present invention transforms the time series into the frequency domain space by calculating the power spectral density, revealing the main periodic patterns of user behavior, making the behavior prediction more interpretable. By constructing the frequency and power density distribution of the power spectral density, it can be optimized according to the actual frequency characteristics of user behavior, ensuring the scientificity and rationality of behavior analysis. Under the calculation of kernel density estimation for the extracted short-, medium-, and long-term behavior windows, due to the true probability distribution of the data, it avoids assuming that user behavior must follow a certain specific distribution. By combining the causal effect value of a single commodity with the similarity between user interest and the commodity, and normalizing the ratio of it to all candidate commodities, it can ensure that the final recommendation score not only considers the causal relationship but also the matching degree between user interest and the commodity itself.

[0039] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. An information analysis method for an e-commerce platform based on artificial intelligence, characterized in that, Including: Obtain user behavior data based on the e-commerce platform, calculate the prior distribution based on the user behavior change rate, combine it with the preliminary likelihood distribution calculated through the probability density function, construct an objective function to optimize the calculation of short-term, medium-term, and long-term behavior windows, calculate the power spectral density of the time series for clustering analysis, optimize the calculation of the three windows, and determine different characteristics of user behavior through kernel density estimation; Adopt the multi-head attention mechanism to fuse different characteristics of users, construct a user interest graph, calculate the partial correlation coefficient using multiple regression analysis, and use the Peter-Clark algorithm to calculate the causal weight between users and commodities, and update the user interest characteristics through heterogeneous GNN; Determine the interest distribution of users for commodities, calculate the causal effect value through user interest characteristics, and perform the final recommendation score calculation; Generate a commodity recommendation list, and perform encryption processing and secure transmission on the data.

2. The information analysis method for an e-commerce platform based on artificial intelligence according to claim 1, characterized in that: The determination of different characteristics of user behavior Including: Based on user behavior data, construct a variational Bayesian time window (VBTW) model, calculate the user behavior change rate at different times based on the window size, perform change rate segmentation of the time window on the change rate sequence, calculate the mean and variance for each change rate window, and use the variational Bayesian method to calculate the prior distribution of the time window; Calculate it as the preliminary likelihood distribution through the probability density function, take the product of the prior distribution and the preliminary likelihood distribution as the objective function, calculate the objective function value for each candidate time window, and use the gradient ascent method to optimize and maximize the objective function; According to the branches of the objective function in different window intervals, respectively determine the smallest optimal window as the short-term behavior window, determine the largest optimal window as the long-term behavior window, and take the mean of the sum of the long-term behavior window and the short-term behavior window as the medium-term behavior window; Use the autoregressive (AR) model to model user behavior data, calculate the autocorrelation function (ACF) of the time series, and calculate the power spectral density through Fourier transform; Construct the frequency and power density based on the calculation of the power spectral density, use K-Means clustering for clustering analysis, use the long-term behavior window, the short-term behavior window, and the medium-term behavior window as the clustering targets, optimize the calculation by calculating the loss function, and optimize the calculation with the three behavior windows as the number of clusters, and extract the windows at the maximum frequency, minimum frequency, and medium frequency of the change frequency f respectively as the optimized short-term window, long-term window, and medium-term window; Adopt Gaussian kernels to calculate the kernel density estimation (KDE) of the short-term window, long-term window, and medium-term window, respectively as the short-term, medium-term, and long-term characteristics of user behavior.

3. The information analysis method for an e-commerce platform based on artificial intelligence according to claim 2, wherein: The update of user interest characteristics includes using Transformer to adopt the multi-head attention mechanism to fuse the short-term, medium-term, and long-term characteristics of users as the comprehensive behavior characteristics; Construct a user interest graph by taking the comprehensive behavior characteristics as user nodes, the commodity texts, images, and historical interaction behavior information of user interactions as commodity nodes, and the category, brand, and price range information of commodities as attribute nodes, and taking the relationships between user nodes and commodity nodes, commodity nodes and attribute nodes, and user nodes and attribute nodes as edges; Based on the edges between user nodes and commodity nodes, respectively determine the comprehensive behavior characteristics as user behavior data U, the commodity texts, images, and historical interaction behavior information as commodity feature data, and the promotion influence and advertisement exposure times as control variable data F; Use multiple regression analysis to calculate the partial correlation coefficient, and use the Peter-Clark algorithm to calculate the conditional partial correlation coefficient of the edge between the user node and the commodity node, and take the conditional partial correlation coefficient as the causal weight of the edge between the user node and the commodity node; Propagate the relationships of users, commodities, and attributes through heterogeneous GNN. Message passing is carried out between the user node c and the commodity node v through the causal weight and attention mechanism of the edge between the user node and the commodity node to update the user interest representation.

4. The information analysis method for an e-commerce platform based on artificial intelligence according to claim 3, characterized in that: The calculation of the final recommendation score includes, based on the updated user interest characteristics, using Softmax normalization to calculate the interest distribution of the user for each commodity attribute, and calculating the causal effect; Calculate the final recommendation score by combining the product of the causal effect value of a single commodity and the similarity between the user interest and the commodity with the ratio of the remaining commodities.

5. The information analysis method for an e-commerce platform based on artificial intelligence according to claim 4, characterized in that: The generation of the commodity recommendation list includes sorting the recommendation scores of all commodities in descending order to generate the final recommendation list, and at the same time generating a commodity classification recommendation list based on the attribute node information of the commodity, including the category, brand, and price range information of the commodity.

6. The information analysis method for an e-commerce platform based on artificial intelligence according to claim 5, characterized in that: The acquisition of user behavior data based on the e-commerce platform includes obtaining the behavioral time series data of the user based on the e-commerce platform, and collecting the behavioral data of the user within the time window, including the number of browsing behaviors, the number of click behaviors, and the number of purchase behaviors, performing data standardization processing, and constructing a user behavior matrix, where each row corresponds to the user behavior data at different times within the time window.

7. The information analysis method of the e-commerce platform based on artificial intelligence according to claim 6, characterized in that: The encryption processing and secure transmission of the data include encrypting and storing the recommendation list data and the user interest feature data of the user through AES-256, managing user permissions using RBAC, and performing secure data transmission using HTTPS, and adding noise to the user behavior data for private storage.

8. A system for an information analysis method of an e-commerce platform based on artificial intelligence, based on the information analysis method of an e-commerce platform based on artificial intelligence according to any one of claims 1 to 7, characterized in that: Including, A user behavior data processing module that obtains user behavior data from the e-commerce platform and calculates the user behavior change rate; A time window optimization module that constructs an objective function based on the prior distribution and the preliminary likelihood distribution, calculates short-term, medium-term, and long-term behavior windows, and optimizes the window division through power spectral density analysis and clustering; A user behavior feature extraction module that uses kernel density estimation to calculate the user behavior characteristics under different time windows, and fuses short-term, medium-term, and long-term behavior patterns through a multi-head attention mechanism; A user interest graph construction module that constructs a heterogeneous graph containing the relationships of users, commodities, and attributes based on user behavior characteristics, and calculates the causal weight between users and commodities; The GNN propagation module calculates the partial correlation coefficient using multiple regression analysis, calculates the causal weight using the Peter-Clark algorithm, and updates the user interest feature through heterogeneous GNN propagation; The recommended score calculation module calculates the recommended score of the user for the commodity based on the user interest distribution and the causal effect value, and performs normalization processing to ensure the rationality of sorting, sorts the calculated recommended scores, and generates a recommendation list; The data security protection module adds noise to the data to protect privacy.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the information analysis method for the e-commerce platform based on artificial intelligence according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the information analysis method for the e-commerce platform based on artificial intelligence according to any one of claims 1 to 7.

Images