E-commerce economic management data analysis method and system

Abstract

The invention discloses an e-commerce economic management data analysis method and system, and relates to the technical field of data analysis, after a dynamic graph is constructed, the popularity score of each node is calculated, the node with the maximum popularity score is highlighted in the dynamic graph, each node is randomly distributed to a plurality of communities, and the dynamic graph is established; performing iteration after maximizing the modularity of each community until the modularity converges, displaying a commodity community graph through a dynamic graph, dividing the dynamic graph into a plurality of time steps, generating embedded representation for the graph of each time step, calculating the graph change between adjacent time steps, and identifying the change trend of a competition pattern in the dynamic graph; and generating decision suggestions according to the optimized dynamic graph. The analysis system maps information such as a commodity association relationship and a popularity score into a time step, dynamically reflects market changes, quantifies a competition pattern change trend, provides a basis for quick response to market fluctuation, and provides an accurate market coping strategy based on an optimized dynamic graph.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to an economic management data analysis method and system for e-commerce. Background Art

[0002] With the rapid development of e-commerce, there are many types of goods on the platform, user behaviors are complex, and market competition is increasingly fierce. Enterprises not only need to attract users through innovative marketing strategies, but also need to improve operational efficiency, achieve cost control and maximize profits. In this context, economic management data analysis systems have gradually become an important tool for decision-making support for e-commerce companies.

[0003] The prior art has the following defects: 1. Existing popularity evaluation methods usually rely on single dimensions such as sales volume and click volume, ignoring the synergistic relationship between products (such as common clicks or keyword similarity). In actual scenarios, the popularity of a product may be affected by related products (for example, the promotion of a certain product may increase the popularity of other related products). Static evaluation cannot reflect this complex network effect. Long-tail products (low sales volume or click volume but strong correlation) are often ignored in existing models, and market potential cannot be fully evaluated; 2. In the face of sudden market changes (such as competitor promotions and new product launches), existing technologies cannot quickly adjust strategies and are prone to losing short-term competitive advantages. Existing analysis models usually lack attention to long-term trends and optimization capabilities, resulting in strategy formulation biased towards short-term interests and ignoring long-term sustainable development. How to achieve a dynamic balance between short-term and long-term goals currently lacks effective modeling and analysis tool support.

[0004] Based on this, the present invention proposes an economic management data analysis method and system for e-commerce, which maps information such as commodity associations and popularity scores into time steps, dynamically reflects market changes, quantifies the changing trends of the competition landscape, provides a basis for rapid response to market fluctuations, and provides accurate market response strategies based on the optimized dynamic graph. Summary of the invention

[0005] The purpose of the present invention is to provide an economic management data analysis method and system for e-commerce to solve the shortcomings of the background technology.

[0006] In order to achieve the above object, the present invention provides the following technical solution: an e-commerce economic management data analysis method, the analysis method comprising the following steps: The analysis system obtains product data and user behavior data from the e-commerce platform, builds a dynamic graph based on the product data and user behavior data, calculates the popularity score of each node based on the PageRank algorithm, and highlights the node with the highest popularity score in the dynamic graph; Each node is randomly assigned to multiple communities, and the modularity of each community is maximized and then iterated until the modularity converges, and the commodity community graph is displayed through a dynamic graph; The dynamic graph is divided into multiple time steps, and the DynGEM embedding algorithm is used to generate an embedded representation for the graph of each time step. The graph changes between adjacent time steps are calculated, the changing trend of the competition landscape in the dynamic graph is identified, and decision recommendations are generated based on the optimized dynamic graph.

[0007] In a preferred embodiment, each node is randomly assigned to multiple communities, and the modularity of each community is maximized and then iterated until the modularity converges and the commodity community graph is displayed through a dynamic graph, including the following steps: In the initial state, each node is randomly assigned to multiple communities, and the modularity of the current community division results is calculated; In each round of iteration, each node is moved from the current community to the community where the neighboring node is located. When the number of iterations is equal to the number threshold, all community division results are output, and the community division result with the largest modularity is selected for use.

[0008] In a preferred embodiment, the modularity of the current community division result is calculated, and the expression is: , where is the modularity, is the total number of edges in the dynamic graph, is the edge weight between node i and node j, is the number of edges of node i, is the number of edges of node j, is the indicator function. If node i and node j belong to the same community, is 1, otherwise is 0, is the sum of the purchase rates of all nodes in the community, is the maximum number of nodes in the community, , is the adjustment coefficient, and , Both are greater than 0.

[0009] In a preferred embodiment, the dynamic graph is divided into multiple time steps, and the DynGEM embedding algorithm is used to generate an embedded representation for the graph of each time step, and the graph changes between adjacent time steps are calculated to identify the changing trend of the competition pattern in the dynamic graph, including the following steps: The dynamic graph is divided into multiple time steps according to the data collection frequency of the e-commerce platform. The graph in each time step represents the nodes and user behavior data in that time step. The initial graph embedding is performed on the dynamic graph of the first time step to generate a low-dimensional vector representation of the node. DynGEM generates an embedded representation for the graph at each subsequent time step through recursive adjustment. It performs iterative optimization based on the addition and deletion of nodes and the change of edge weights. Each node generates a vector of fixed dimension, which represents its characteristics in the current time step. The embedded vector reflects the characteristics of the node and its importance in the network. Compare the Euclidean distance or cosine similarity of the node embedding vectors in adjacent time steps to evaluate node changes. For edges, calculate the absolute difference or ratio of weight changes. If the embedding of a node changes significantly, it indicates that the competitive position of the node has changed. If the weight of the edge between two nodes changes significantly, it indicates that the association between the two nodes has changed. According to the changing trajectory of nodes in the embedding space, the popularity trend and community position of the nodes are predicted. Through time series analysis and embedding representation, nodes with continuous growth or decline are identified, and the dynamic graph is adjusted based on the changing trend.

[0010] In a preferred embodiment, predicting the popularity trend and community position of a node according to the change trajectory of the node in the embedding space includes the following steps: In each time step, the embedding vector of the node is generated. The embedding vector reflects the popularity trend and community position of the node. By comparing the embedding representations of adjacent time steps, the change amplitude of each node is calculated. The expression is: , where is the variation range of node v, is the feature vector of node v at time t, is the feature vector of node v at time t+1, represents the cosine distance; A time series is constructed for the popularity score of each node to reflect the change of its popularity over time, and the time series model is used to predict the popularity score: , where Score the heat of node v at time t+1, Represents a prediction function, which is used to predict future heat based on past heat sequences. represents the historical heat score of node v from time 1 to t, represents the historical heat score of node v at the i-th time step; According to the prediction results, the node heat changes are classified, and in each time step, the community to which the node belongs is recorded, and the trajectory of community changes is analyzed: , where is the degree of community change of node v at time steps t and t+1, represents the cosine similarity function, represents the set of communities to which node v belongs at time step t, represents the set of communities to which node v belongs at time step t+1.

[0011] In a preferred embodiment, the heat score of each node is calculated based on the PageRank algorithm, and the node with the largest heat score is highlighted in the dynamic graph, including the following steps: Assume that there are N nodes in the dynamic graph, and the initial heat score of each node is set equal. The heat score of each node is iteratively updated according to the PageRank algorithm. The expression is: , where is the heat score of node i, is the damping factor, and , is the heat score of node j, represents the number of edges starting from node j, Represents the set of all nodes pointing to node i; When any heat score change is less than or equal to the change threshold, the convergence condition is judged to be met. After the heat scores of all nodes are output, all nodes are sorted from large to small according to the heat scores, and a heat table is generated. The heat table is mapped to the dynamic graph, and different nodes are marked with different colors.

[0012] In a preferred embodiment, the analysis system obtains product data and user behavior data through an e-commerce platform, and constructs a dynamic graph based on the product data and user behavior data, including the following steps: Obtain product category, sales volume, and price information, record user click behavior, including the number of common clicks between products and purchase paths, remove abnormal or incomplete data, integrate static product data with dynamic user behavior data, and build a global association relationship between products; Each product corresponds to a node. The node attributes include sales volume, category, and price. The edge weight is calculated based on the user behavior data to obtain the information of the two nodes connected by the edge. The node information includes the number of clicks by the same user and the text similarity. The number of clicks by the same user and the text similarity are normalized so that the value range of the number of clicks by the same user and the text similarity is mapped to [0,1]. The normalized number of clicks by the same user and the text similarity are summed to obtain the edge weight. Construct an initial dynamic graph, where nodes are commodities, edges are associations, and dynamic time steps are defined as different time windows.

[0013] An economic management data analysis system for e-commerce, comprising a graph construction module, a heat ranking module, and a trend analysis module; Graph construction module: obtains product data and user behavior data through the e-commerce platform, and builds dynamic graphs based on the product data and user behavior data; Hotness ranking module: Calculates the hotness score of each node based on the PageRank algorithm, and highlights the node with the highest hotness score in the dynamic graph; Trend analysis module: Each node is randomly assigned to multiple communities, and the modularity of each community is maximized before iteration. After the modularity converges, the commodity community graph is displayed through a dynamic graph. The dynamic graph is divided into multiple time steps, and the DynGEM embedding algorithm is used to generate an embedded representation for the graph of each time step. The graph changes between adjacent time steps are calculated, the changing trend of the competition landscape in the dynamic graph is identified, and decision recommendations are generated based on the optimized dynamic graph.

[0014] In the above technical solution, the technical effects and advantages provided by the present invention are: The present invention constructs a dynamic graph based on commodity data and user behavior data, calculates the heat score of each node, and highlights the node with the largest heat score in the dynamic graph. Each node is randomly assigned to multiple communities, and iterates after maximizing the modularity of each community until the modularity converges, and then displays the commodity community graph through a dynamic graph. The dynamic graph is divided into multiple time steps, and the DynGEM embedding algorithm is used to generate an embedded representation for the graph of each time step, calculates the graph changes between adjacent time steps, identifies the changing trend of the competitive landscape in the dynamic graph, and generates decision suggestions based on the optimized dynamic graph. The analysis system maps information such as commodity associations and heat scores to time steps, dynamically reflects market changes, quantifies the changing trend of the competitive landscape, provides a basis for rapid response to market fluctuations, and provides accurate market response strategies based on the optimized dynamic graph. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] 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 will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0016] Figure 1 The present invention is a flow chart of the method. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] Example 1: Please refer to Figure 1 As shown, the present embodiment provides an e-commerce economic management data analysis method, the analysis method comprising the following steps: The analysis system obtains product data and user behavior data from the e-commerce platform. Product data includes categories, sales, etc. User behavior data includes click associations between products. For example, a user may click on product B after purchasing product A. A dynamic graph is constructed based on the product data and user behavior data. The nodes in the dynamic graph are products, and the edges are the associations between products (such as the number of common clicks, similar keywords). The heat score of each node is calculated based on the PageRank algorithm, and the node with the largest heat score is highlighted in the dynamic graph (for example, in a prominent color such as red). Each node is randomly assigned to multiple communities, and the modularity of each community is maximized and then iterated until the modularity converges. The product community graph is displayed through a dynamic graph, and the dynamic graph is divided into multiple time steps. The DynGEM embedding algorithm is used to generate an embedded representation for the graph of each time step, the graph changes between adjacent time steps are calculated, the changing trend of the competitive landscape in the dynamic graph is identified, and decision recommendations are generated based on the optimized dynamic graph.

[0019] This application builds a dynamic graph based on commodity data and user behavior data, calculates the heat score of each node, and highlights the node with the largest heat score in the dynamic graph. Each node is randomly assigned to multiple communities, and iterates after maximizing the modularity of each community until the modularity converges. The commodity community graph is displayed through a dynamic graph, and the dynamic graph is divided into multiple time steps. The DynGEM embedding algorithm is used to generate an embedded representation for the graph of each time step, calculate the graph changes between adjacent time steps, identify the changing trend of the competitive landscape in the dynamic graph, and generate decision recommendations based on the optimized dynamic graph. The analysis system maps information such as commodity associations and heat scores to time steps, dynamically reflects market changes, quantifies the changing trend of the competitive landscape, provides a basis for rapid response to market fluctuations, and provides accurate market response strategies based on the optimized dynamic graph.

[0020] Embodiment 2: The analysis system obtains product data and user behavior data through the e-commerce platform. The product data includes categories, sales, etc. The user behavior data includes click associations between products. For example, after purchasing product A, a user may click product B. A dynamic graph is constructed based on the product data and user behavior data. The nodes in the dynamic graph are products, and the edges are associations between products (such as the number of common clicks, similar keywords), including the following steps: Obtain static information such as product category, sales volume, price, etc., record user click behavior, such as the number of common clicks on product A and product B, purchase path, etc., remove abnormal or incomplete data (such as product delisting or insufficient click records), integrate static product data with dynamic user behavior data, and build a global association relationship between products; Each product corresponds to a node. The node attributes include sales volume, category, price, etc. The relationship between products is calculated based on user behavior data. The information of the two nodes connected by the edge is obtained, including the number of clicks by the same user and the text similarity. The number of clicks by the same user and the text similarity are normalized so that the value range of the number of clicks by the same user and the text similarity is mapped to [0,1]. The normalized number of clicks by the same user and the text similarity are summed to obtain the edge weight; Construct the initial dynamic graph, where nodes are products, edges are relationships, and dynamic time steps are defined as different time windows (such as days, weeks, and months).

[0021] The heat score of each node is calculated based on the PageRank algorithm, and the node with the largest heat score is highlighted in the dynamic graph (for example, displayed in a highlight color such as red), including the following steps: Assume that there are N nodes in the dynamic graph, and the initial heat score of each node is set equal. The heat score of each node is iteratively updated according to the PageRank algorithm. The expression is: , where is the heat score of node i, is the damping factor, and , is the heat score of node j, represents the number of edges starting from node j, Represents the set of all nodes pointing to node i (i.e. nodes with edges pointing to node i); When any heat score change is less than or equal to the change threshold, it is judged that the convergence condition is met. After outputting the heat scores of all nodes, all nodes are sorted from large to small according to the heat scores to generate a heat table. The heat table is mapped to the dynamic graph, and different nodes are marked with different colors. Nodes with high heat (such as product A) are marked in red, and nodes with low heat (such as product C) are marked in gray.

[0022] Highlighting high-profile products (such as A) can help e-commerce platforms identify which products are receiving more attention in the market. Product A may be the focus of the current market and needs special attention; For highly popular products (such as Product A): Strengthen marketing strategies: For example, increase the sales of Product A through recommendation systems, discounts and promotions, etc. Pay attention to market reactions: Monitor user feedback on Product A and adjust strategies in a timely manner.

[0023] For low-profile products (such as product C): consider reducing inventory or optimizing promotion strategies to increase the exposure of product C. Further analysis of the reasons for low-profile products may be due to pricing, description, or keyword optimization issues.

[0024] Each node is randomly assigned to multiple communities, and the modularity of each community is maximized and then iterated until the modularity converges and the commodity community graph is displayed through a dynamic graph, including the following steps: In the initial state, each node is randomly assigned to multiple communities, which avoids dividing all products into fixed communities at the beginning, increases algorithm flexibility, and provides space for subsequent optimization; Calculate the modularity of the current community division result, the expression is: , where is the modularity, is the total number of edges in the dynamic graph, is the edge weight between node i and node j, is the number of edges of node i, is the number of edges of node j, is the indicator function. If node i and node j belong to the same community, is 1, otherwise is 0, is the sum of the purchase rates of all nodes in the community, is the maximum number of nodes in the community (that is, after obtaining the number of nodes in each community, select the maximum number of nodes), , is the adjustment coefficient, and , All are greater than 0; By adjusting the community division of nodes, the nodes within the community are connected as much as possible, while the nodes between communities are connected as little as possible, and the number of nodes in the community is as small as possible. The larger the modularity, the better the community structure, the stronger the association within the community, and the weaker the association between communities.

[0025] In each round of iteration, try to move each node from the current community to the community where the neighbor node is located, and select the operation that can maximize the modularity gain. When the number of iterations equals the number threshold, output all community division results, and select the community division result that maximizes the modularity for use.

[0026] 1) Relationship between the total purchase rate of all nodes in the community and modularity: Assume that the purchase rate of each node (product) refers to the purchase probability or purchase frequency of the product on the e-commerce platform. In community division, the sum of the purchase rates of all nodes in the community is the cumulative value of the purchase rates of all products, indicating the overall popularity of the products in the community.

[0027] The core of modularity is to measure the difference between the degree of connection between nodes within a community and the degree of connection between communities. If the purchase rate of nodes (products) in a community is high, it means that the products within the community may have strong market appeal and consumer purchase intention, which will affect the edge weight within the community and the strength of association between nodes.

[0028] Communities with a high total purchase rate: When the total purchase rate of all products in a community is high, it means that the products in the community may be more popular and users pay more attention to them. Therefore, the nodes in the community will have more edge connections (such as joint purchases, recommendations, etc.), and the connections within the community will be closer. The value of modularity depends on the density of edge weights within the community. When the total purchase rate of a community is high, the edge weights within the community will increase, and the connections within the community will be closer, thereby improving modularity.

[0029] Communities with low total purchase rates: On the contrary, if the total purchase rate of goods within a community is low, it means that these goods have weak market appeal, and the goods within the community may have low correlation, resulting in weak connections within the community. In this case, there may not be many connections between nodes within the community, and the modularity of the community may be low.

[0030] A community with a higher sum of purchase rates generally means that the nodes within the community are more closely connected to each other, thus improving modularity. A community with a lower sum of purchase rates may result in weaker connections between nodes within the community, thus reducing modularity.

[0031] 2) Relationship between the maximum number of nodes in a community and modularity: Communities with a small maximum number of nodes: If the maximum number of nodes is small, it means that all communities are evenly divided and there is no obvious overly large community. Modularity is not penalized, and a higher modularity indicates that the current division is more reasonable. Communities with a large maximum number of nodes: If there are one or more communities with a very large number of nodes, the modularity may be artificially high, because overly large communities tend to contain a large number of connections, resulting in higher edge weights within the community. After introducing the penalty factor, communities with a large maximum number of nodes will be significantly penalized, and the modularity will decrease, making them more likely to be reasonably divided.

[0032] Too large a single community: If a community contains most of the nodes (for example, the entire graph is divided into one community), although the original modularity may be high, it is not very meaningful in practice. After introducing the penalty factor, this partition will lead to extremely low modularity, thus promoting a finer-grained partition.

[0033] The dynamic graph is divided into multiple time steps, and the DynGEM embedding algorithm is used to generate an embedded representation for the graph of each time step. The graph changes between adjacent time steps are calculated, and the changing trend of the competition landscape in the dynamic graph is identified. Decision recommendations are generated based on the optimized dynamic graph, including the following steps: Divide the dynamic graph into multiple time steps according to the data collection frequency of the e-commerce platform (such as daily, weekly or monthly). The graph in each time step represents the product and user behavior data in that time period. Perform initial graph embedding on the dynamic graph of the first time step to generate a low-dimensional vector representation of the node. The embedding representation dimension is configured by the system and is adaptively selected based on the number of nodes or edges. DynGEM generates an embedded representation for the graph at each subsequent time step through recursive adjustment. The embedding algorithm iteratively optimizes according to the addition and deletion of nodes and the change of edge weights to maintain the continuity of the embedding representation and ensure that the new embedding is comparable with the embedding of the previous time step. Each node generates a vector of fixed dimension, which represents its characteristics in the current time step. The embedding vector reflects the characteristics of the product and its importance in the network (such as the strength of association). Compare the Euclidean distance or cosine similarity of the node embedding vectors in adjacent time steps to evaluate node changes. For edges (association between products), calculate the absolute difference or ratio of weight changes. If the embedding of a node changes significantly, it indicates that the competitive position of the product has changed significantly (such as an increase or decrease in popularity). If the weight of the edge between two products changes significantly, it indicates that the association between the two products (such as the trend of users purchasing together) has changed significantly. Based on the changes in nodes and edges of the entire network, a visualization chart of changes in the global competitive landscape is generated, and newly emerged high-weight nodes or edges are analyzed to identify products that have risen rapidly recently and their competitors. Other nodes (products) that these products are close to in the embedding space are found, and potential competitors or substitutes are analyzed. Through changes in embedding representations, it is detected which communities (product categories) have the most intense competition, and edges or nodes that change frequently in multiple time steps are found. Hot products in the market or combinations of related products are determined. According to the change trajectory of nodes in the embedding space, the popularity trend of products and their possible community positions are predicted. Through time series analysis and embedding representations, products that are continuously growing or may decline are identified.

[0034] Adjust the dynamic graph based on the changing trends (such as redefining edge weights or node importance).

[0035] Highlight key nodes (such as fast-rising items) and edges (such as emerging relationships).

[0036] Visualize the final dynamic map and intuitively display the competitive landscape through visual elements such as color and size.

[0037] Generate decision recommendations: Suggestions on competitive strategy: Adjust your product pricing and promotion strategy for key products of competitors. Develop targeted marketing activities for rapidly emerging competitor products.

[0038] Optimize product recommendation strategy: Based on dynamic graph embedding, optimize the recommendation algorithm and give priority to products with low competition but great potential.

[0039] Inventory and supply chain optimization: For products whose popularity is rising rapidly, increase inventory in advance. For products whose popularity is declining, optimize inventory to reduce waste.

[0040] New product launch strategy: Based on the blank areas in the competitive landscape, explore potential market demand and design and launch new products.

[0041] Regular iteration: Repeat the above process regularly to dynamically track market changes. Generate real-time updated decision suggestions based on new dynamic graphs.

[0042] According to the changing trajectory of nodes in the embedding space, the popularity trend and community position of nodes are predicted. Through time series analysis and embedding representation, nodes with continuous growth or decline are identified, including the following steps: In each time step, the embedding vector of the node is generated. The embedding vector reflects the popularity trend and community position of the node. By comparing the embedding representations of adjacent time steps, the change amplitude of each node is calculated. The expression is: , where is the variation range of node v, is the feature vector of node v at time t, is the feature vector of node v at time t+1, represents the cosine distance, and a trajectory with a large change in the node indicates that its role in the community or network has changed significantly; Use clustering algorithms to automatically divide nodes into three categories: growth, stability, and decline according to the distribution of node change amplitudes. Build a time series for each node's popularity (such as associated edge weight and click-through rate) to reflect changes in its popularity over time. Use a time series model to predict the popularity: , where: Score the heat of node v at time t+1, Represents the historical heat score of node v from time 1 to t, each value Represents the historical heat score of the node at the i-th time step.

[0043] Represents a prediction function, which is used to predict future popularity based on past popularity sequences. It can be a time series model (such as ARIMA, LSTM, etc.) or other regression prediction methods. The prediction method belongs to the existing technology. The example is as follows: Collect the heat data of each node v in the past several time periods, such as the past 24 hours or the past 7 days, to form a heat time series , normalize the heat sequence, for example, compress the data to the [0,1][0,1][0,1] interval through Min-Max normalization, construct training samples in a sliding window manner, for example, use a heat sequence of length t to predict the heat value at the next moment.

[0044] Constructing LSTM neural network structure: Input layer: accepts a heat sequence of t time steps; LSTM layer: several hidden layer units can be set; Output layer: outputs a scalar .

[0045] Use historical data to train the model and minimize the predicted heat score The error between the actual heat score and the actual heat score (such as mean square error MSE).

[0046] Input the latest t heat values ​​into the LSTM model and output the predicted value. If it has been normalized, denormalize it to get the real heat value.

[0047] According to the prediction results, the popularity of the classification nodes changes: Continuous growth nodes: The popularity increases significantly, and the future trend is upward. Continuous decline nodes: The popularity continues to decrease, and attention or optimization is needed. Stable nodes: There is no significant change in popularity, and the influence is stable; In each time step, record the community to which the node belongs and analyze the trajectory of community changes: , where is the degree of community change of node v at time steps t and t+1, represents the cosine similarity function, represents the set of communities to which node v belongs at time step t, represents the set of communities to which node v belongs at time step t+1; Growth node identification: The popularity has increased significantly, the embedded features have changed greatly, and they tend to be close to other high-popularity nodes. Typical features: There are many newly added associated edges, and the popularity trend forecast shows an increase.

[0048] Decay node identification: The popularity has dropped significantly, and the embedded features tend to move away from the central community or hot products. Typical features: Loss of associated edges and low changes in the embedded vector.

[0049] Stable node identification: The changes in popularity and community position are small and continue to remain within the current influence range.

[0050] Optimize node heat display: Use more prominent colors (such as red) for growth nodes to indicate that they are potential contention hotspots. Use gray or light colors for decay nodes to reduce their visual prominence.

[0051] Redefine community division: According to the predicted community position adjustment, frequently transferred nodes are moved from the original community to the new community. By dynamically adjusting the modularity optimization function, the community division is ensured to be more realistic.

[0052] Update edge weights: Based on the predicted changes in node associations, dynamically adjust edge weights to highlight the associations between emerging commodities. For disappeared associations, weaken or remove the corresponding edges.

[0053] Visualize the dynamic graph after adjustment: Node size: Displayed by predicted popularity (the higher the popularity, the larger the node).

[0054] Edge weights: Adjust edge width or transparency based on predicted association strength.

[0055] Time series changes: Display the dynamic changes of nodes and communities frame by frame through dynamic graphs.

[0056] Embodiment 3: The economic management data analysis system for e-commerce described in this embodiment includes a graph construction module, a heat ranking module, and a trend analysis module; Graph construction module: obtains product data and user behavior data through the e-commerce platform. Product data includes categories, sales, etc. User behavior data includes click associations between products. For example, after purchasing product A, a user may click product B. A dynamic graph is constructed based on the product data and user behavior data. The nodes in the dynamic graph are products, and the edges are the associations between products (such as the number of common clicks, similar keywords). The dynamic graph is sent to the popularity ranking module; Heat ranking module: Calculate the heat score of each node based on the PageRank algorithm, and highlight the node with the highest heat score in the dynamic graph (for example, in red or other prominent colors), and send the optimized dynamic graph to the trend analysis module; Trend analysis module: Each node is randomly assigned to multiple communities, and the modularity of each community is maximized before iteration. After the modularity converges, the commodity community graph is displayed through a dynamic graph. The dynamic graph is divided into multiple time steps, and the DynGEM embedding algorithm is used to generate an embedded representation for the graph of each time step. The graph changes between adjacent time steps are calculated, the changing trend of the competition landscape in the dynamic graph is identified, and decision recommendations are generated based on the optimized dynamic graph.

[0057] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0058] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0059] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. An economic management data analysis method for e-commerce, characterized by: The analytical method comprises the following steps: The analysis system obtains product data and user behavior data from the e-commerce platform, builds a dynamic graph based on the product data and user behavior data, calculates the popularity score of each node based on the PageRank algorithm, and highlights the node with the highest popularity score in the dynamic graph; Each node is randomly assigned to multiple communities, and the modularity of each community is maximized and then iterated until the modularity converges, and the commodity community graph is displayed through a dynamic graph; The dynamic graph is divided into multiple time steps, and the DynGEM embedding algorithm is used to generate an embedded representation for the graph of each time step. The graph changes between adjacent time steps are calculated, the changing trend of the competition landscape in the dynamic graph is identified, and decision recommendations are generated based on the optimized dynamic graph.

2. The method for analyzing economic management data for e-commerce according to claim 1, characterized in that: Each node is randomly assigned to multiple communities, and the modularity of each community is maximized and then iterated until the modularity converges and the commodity community graph is displayed through a dynamic graph, including the following steps: In the initial state, each node is randomly assigned to multiple communities, and the modularity of the current community division results is calculated; In each round of iteration, each node is moved from the current community to the community where the neighboring node is located. When the number of iterations is equal to the number threshold, all community division results are output, and the community division result with the largest modularity is selected for use.

3. The method for analyzing economic management data for e-commerce according to claim 2, characterized in that: Calculate the modularity of the current community division result, the expression is: , where is the modularity, is the total number of edges in the dynamic graph, is the edge weight between node i and node j, is the number of edges of node i, is the number of edges of node j, is the indicator function. If node i and node j belong to the same community, is 1, otherwise is 0, is the sum of the purchase rates of all nodes in the community, is the maximum number of nodes in the community, , is the adjustment coefficient, and , Both are greater than 0.

4. The method for analyzing economic management data for e-commerce according to claim 3, characterized in that: The dynamic graph is divided into multiple time steps, and the DynGEM embedding algorithm is used to generate an embedded representation for the graph of each time step, calculate the graph changes between adjacent time steps, and identify the changing trend of the competition landscape in the dynamic graph, including the following steps: The dynamic graph is divided into multiple time steps according to the data collection frequency of the e-commerce platform. The graph in each time step represents the nodes and user behavior data in that time step. The initial graph embedding is performed on the dynamic graph of the first time step to generate a low-dimensional vector representation of the node. DynGEM generates an embedded representation for the graph at each subsequent time step through recursive adjustment. It performs iterative optimization based on the addition and deletion of nodes and the change of edge weights. Each node generates a vector of fixed dimension, which represents its characteristics in the current time step. The embedded vector reflects the characteristics of the node and its importance in the network. Compare the Euclidean distance or cosine similarity of the node embedding vectors in adjacent time steps to evaluate node changes. For edges, calculate the absolute difference or ratio of weight changes. If the embedding of a node changes significantly, it indicates that the competitive position of the node has changed. If the weight of the edge between two nodes changes, it indicates that the association between the two nodes has changed. According to the changing trajectory of nodes in the embedding space, the popularity trend and community position of the nodes are predicted. Through time series analysis and embedding representation, nodes with continuous growth or decline are identified, and the dynamic graph is adjusted based on the changing trend.

5. The method for analyzing economic management data for e-commerce according to claim 4, characterized in that: According to the change trajectory of the node in the embedding space, the popularity trend and community position of the node are predicted, including the following steps: In each time step, the embedding vector of the node is generated. The embedding vector reflects the popularity trend and community position of the node. By comparing the embedding representations of adjacent time steps, the change amplitude of each node is calculated. The expression is: , where is the variation range of node v, is the feature vector of node v at time t, is the feature vector of node v at time t+1, represents the cosine distance; A time series is constructed for the popularity score of each node to reflect the change of its popularity over time, and the time series model is used to predict the popularity score: , where Score the heat of node v at time t+1, Represents a prediction function, which is used to predict future heat based on past heat sequences. represents the historical heat score of node v from time 1 to t, represents the historical heat score of node v at the i-th time step; According to the prediction results, the node heat changes are classified, and in each time step, the community to which the node belongs is recorded, and the trajectory of community changes is analyzed: , where is the degree of community change of node v at time steps t and t+1, represents the cosine similarity function, represents the set of communities to which node v belongs at time step t, Represents the set of communities to which node v belongs at time step t+1.

6. The method for analyzing economic management data for e-commerce according to claim 5, characterized in that: The heat score of each node is calculated based on the PageRank algorithm, and the node with the largest heat score is highlighted in the dynamic graph, including the following steps: Assume that there are N nodes in the dynamic graph, and the initial heat score of each node is set equal. The heat score of each node is iteratively updated according to the PageRank algorithm. The expression is: , where is the heat score of node i, is the damping factor, and =0.85, is the heat score of node j, represents the number of edges starting from node j, Represents the set of all nodes pointing to node i; When any heat score change is less than or equal to the change threshold, the convergence condition is judged to be met. After the heat scores of all nodes are output, all nodes are sorted from large to small according to the heat scores to generate a heat table. The heat table is mapped to the dynamic graph, and different nodes are marked with different colors.

7. The method for analyzing economic management data for e-commerce according to claim 6, characterized in that: The analysis system obtains product data and user behavior data from the e-commerce platform, and builds a dynamic graph based on the product data and user behavior data, including the following steps: Obtain product category, sales volume, and price information, record user click behavior, including the number of common clicks between products and purchase paths, remove abnormal or incomplete data, integrate static product data with dynamic user behavior data, and build a global association relationship between products; Each product corresponds to a node. The node attributes include sales volume, category, and price. The edge weight is calculated based on the user behavior data to obtain the information of the two nodes connected by the edge. The node information includes the number of clicks by the same user and the text similarity. The number of clicks by the same user and the text similarity are normalized so that the value range of the number of clicks by the same user and the text similarity is mapped to [0,1]. The normalized number of clicks by the same user and the text similarity are summed to obtain the edge weight. Construct an initial dynamic graph, where nodes are commodities, edges are associations, and dynamic time steps are defined as different time windows.

8. An economic management data analysis system for e-commerce, used to implement the analysis method according to any one of claims 1 to 7, characterized in that: Including graph construction module, heat sorting module, and trend analysis module; Graph construction module: obtains product data and user behavior data through the e-commerce platform, and builds dynamic graphs based on the product data and user behavior data; Hotness ranking module: Calculates the hotness score of each node based on the PageRank algorithm, and highlights the node with the highest hotness score in the dynamic graph; Trend analysis module: Each node is randomly assigned to multiple communities, and the modularity of each community is maximized before iteration. After the modularity converges, the commodity community graph is displayed through a dynamic graph. The dynamic graph is divided into multiple time steps, and the DynGEM embedding algorithm is used to generate an embedded representation for the graph of each time step. The graph changes between adjacent time steps are calculated, the changing trend of the competition landscape in the dynamic graph is identified, and decision recommendations are generated based on the optimized dynamic graph.

Images