Big data analysis method based on e-commerce platform and AI system

Abstract

The invention relates to the technical field of electronic shopping, in particular to a big data analysis method based on an e-commerce platform and an AI system, and the method comprises the following steps: collecting a click rate, a search frequency and a purchase record of a user on the e-commerce platform, carrying out the subdivision of the click rate, the search frequency and the purchase record, and creating a data set; according to the invention, by integrating the equipment fingerprint technology and the social media data, the analysis precision of the user behavior data is improved, and the capabilities of user identity recognition and behavior pattern analysis are further optimized. In-depth analysis is carried out on user interaction data such as click rate, search frequency and purchase records, consumer behavior patterns and preference prediction are refined through clustering and decision tree algorithms, and sentiment analysis and keyword extraction are carried out on social media posting data. Therefore, the enterprise can monitor the market dynamic state in real time and quickly adjust the marketing strategy. And the accuracy of market prediction and the pertinence of marketing activities are improved.

Description

Technical Field

[0001] The present invention relates to the field of electronic shopping technology, and in particular to a big data analysis method and an AI system based on an e-commerce platform. Background Art

[0002] The e-shopping technology field involves the use of information technology and Internet platforms to facilitate the sale of goods and services. The technical foundation includes website design, online payment processing, data encryption, and user interface optimization to ensure a smooth and secure user experience.

[0003] Although existing technologies provide basic e-shopping functions and user experience optimization, they have obvious shortcomings in large-scale real-time data processing and analysis. The lack of monitoring and analysis of social media dynamics makes it difficult to fully predict market trends or effectively evaluate the impact of marketing strategies. This limitation may lead to missed market opportunities and insufficient response to changes in consumer demand, especially in a highly competitive e-commerce environment, which may lead to customer churn and reduced sales efficiency. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a big data analysis method and AI system based on an e-commerce platform.

[0005] In order to achieve the above purpose, the present invention adopts the following technical solution, which is a big data analysis method based on an e-commerce platform, comprising the following steps: Collecting click-through rates, search frequencies, and purchase records of users on the e-commerce platform, segmenting the click-through rates, search frequencies, and purchase records to create a data set, using device fingerprints and social media links to identify user identities and associate attributes with the data set, generating a user identity link set, analyzing the user identity link set, distinguishing consumer behavior patterns through clustering, and obtaining consumer behavior groups; Based on the consumer behavior groups, the shopping frequency and product type of each data point are attributed calculated through a decision tree, the user's preference basis for future products is established, and a preference prediction model is generated; the product demand probability is analyzed, the demand for each product category is calculated, and the market demand prediction result is obtained; Based on the market demand forecast results, market trend change forecasts are performed, and sales strategies are formulated for each product in combination with past sales data and market trend change forecast results, generating sales strategy adjustment suggestions; Extract posting data from social media platforms, perform sentiment analysis and keyword extraction on the posting data, determine the emotional tendency and topic relevance of the posting data, estimate the impact of the posting data on product sales, and obtain impact results.

[0006] Preferably, the steps of obtaining the user identity link set are: Collect the click-through rate, search frequency, and purchase history of users on the e-commerce platform, segment and create data sets, and obtain user behavior data sets; Based on the user behavior dataset, the identity recognition index of each user is calculated using device fingerprints and social media links. The calculation formula is:

[0007] in, is the user identification index, and They are device fingerprint score and social media link score, and is the adjustment coefficient; According to each user's identity identification index, a user identity link set is generated.

[0008] Preferably, the steps of obtaining the consumer behavior group are: Based on the user identity link set, applying a clustering algorithm to analyze user data to generate a clustering analysis result; According to the cluster analysis results, user behavior patterns are distinguished, and users are grouped according to behavior similarities to obtain consumer behavior groups.

[0009] Preferably, the step of obtaining the preference prediction model is: According to the consumer behavior groups, the decision tree is used to perform attribute analysis on the shopping frequency and product type of each data point, and then the product preference index of each consumer is calculated. The calculation formula is:

[0010] in, Indicates The product preference index of consumers, For the The shopping frequency of each data point, is the score of the corresponding product type, is the total number of data points; Based on the product preference index, a preference basis for each consumer for future products is established, and a preference prediction model is generated.

[0011] Preferably, the steps for obtaining the market demand forecast result are: Extract past sales data and user rating information for each product category to obtain a basic set of data analysis; The demand probability is calculated for each product category in the data analysis basic set, and the calculation formula is:

[0012] in, Indicates The demand probability of product categories, For the The sum of the scores of product categories in the data analysis base set, is the total number of product categories, is the sum of the ratings for other product categories; According to the demand probability of each product category, the expected market demand is calculated in combination with the preference prediction model to obtain a market demand prediction result.

[0013] Preferably, the steps for obtaining the sales strategy adjustment suggestion are: Integrate the market demand forecast results with past sales data, analyze the sales trends and market demand differences of product categories, and generate a basic data set; Through time series analysis, based on the basic data set, the future demand changes and market dynamics of each product category are predicted to generate market trend change prediction results; Based on the market trend change forecast results, combined with product characteristics and market competition conditions, sales strategy adjustment suggestions are customized for each product, and the sales strategy adjustment suggestions include adjusting pricing, promotional activities and inventory strategies.

[0014] Preferably, the steps of obtaining the impact result are: Extracting posting data of related products from social media platforms to obtain an original posting dataset; The original posting data set is subjected to text analysis, and the sentiment of each posting data is manually labeled as positive, neutral or negative, and the sentiment score of each post is calculated using the following formula:

[0015] in, represents the sentiment score, For the Manually annotated sentiment scores of posts, is the user interaction factor of the post, is the time factor, is the total number of posts analyzed; The impact results on product sales are generated based on the sentiment score of each posting data and the frequency of occurrence of keywords.

[0016] The present invention provides an AI system, comprising: The user behavior analysis module is used to extract user click-through rates, search frequencies, and purchase records from e-commerce platforms, calculate the activity frequency and purchase habits of each user, identify user identities and associate attributes based on device fingerprints and social media links, and generate user identity link sets; through cluster analysis of user identity link sets, consumer behavior patterns are distinguished and consumer behavior groups are obtained; The preference prediction module is used to calculate the shopping frequency and product type of the data points in each consumer behavior group based on consumer behavior groups, establish a preference basis, and generate a preference prediction model; analyze the demand probability of each product category to obtain market demand prediction results; The market strategy module analyzes changes in market trends and adjusts sales strategies based on the market demand forecast results; extracts posting data from social media platforms, performs sentiment analysis and keyword extraction, estimates the impact on product sales, and generates sales strategy adjustment suggestions.

[0017] Compared with the prior art, the advantages and positive effects of the present invention are: The present invention improves the analysis accuracy of user behavior data by integrating device fingerprint technology and social media data, and further optimizes the ability of user identity recognition and behavior pattern analysis. It conducts in-depth analysis of user interaction data such as click-through rate, search frequency and purchase history, and refines consumer behavior patterns and preference predictions through clustering and decision tree algorithms. In addition, it also includes sentiment analysis and keyword extraction of social media posting data, which enables companies to monitor market dynamics in real time and quickly adjust marketing strategies. It improves the accuracy of market forecasts and the pertinence of marketing activities, thereby enhancing the company's response speed to market changes and the accuracy of strategy adjustments. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the steps of the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0020] See also Figure 1 The present invention provides a technical solution, a big data analysis method based on an e-commerce platform, comprising the following steps: Collect the click-through rate, search frequency and purchase history of users on the e-commerce platform, segment the click-through rate, search frequency and purchase history and create a data set, use device fingerprints and social media links to identify users and associate attributes in the data set, generate a user identity link set, analyze the user identity link set, distinguish consumer behavior patterns through clustering, and obtain consumer behavior groups; Based on consumer behavior groups, the shopping frequency and product type of each data point are calculated through decision trees to establish the user's preference basis for future products and generate a preference prediction model; the product demand probability is analyzed to calculate the demand for each product category and obtain the market demand forecast results; According to the market demand forecast results, we will perform market trend change forecasts, combine past sales data with the market trend change forecast results, formulate sales strategies for each product, and generate sales strategy adjustment suggestions; Extract posting data from social media platforms, perform sentiment analysis and keyword extraction on the posting data, determine the emotional tendency and topic relevance of the posting data, estimate the impact of the posting data on product sales, and obtain the impact results.

[0021] Specifically, by integrating device fingerprint technology and social media data, the analysis accuracy of user behavior data has been improved, and the ability to identify user identities and analyze behavior patterns has been further optimized. In-depth analysis of user interaction data such as click-through rate, search frequency, and purchase records is performed, and consumer behavior patterns and preference predictions are refined through clustering and decision tree algorithms. In addition, sentiment analysis and keyword extraction of social media posting data are also included, which enables companies to monitor market dynamics in real time and quickly adjust marketing strategies. The accuracy of market forecasts and the targeting of marketing activities have been improved, thereby enhancing the company's response speed to market changes and the accuracy of strategy adjustments.

[0022] The steps to obtain the user identity link set are: Collect the click-through rate, search frequency, and purchase history of users on the e-commerce platform, segment and create data sets, and obtain user behavior data sets; Based on the user behavior dataset, the identity recognition index of each user is calculated using device fingerprints and social media links. The calculation formula is:

[0023] in, is the user identification index, and They are device fingerprint score and social media link score, and is the adjustment coefficient.

[0024] According to each user's identity identification index, a user identity link set is generated.

[0025] Specifically, the click-through rate, search frequency and purchase history of users on the e-commerce platform are collected. These data are used to create a preliminary user behavior dataset, which is the basis for analyzing user purchase tendencies and behavior patterns. Based on these data, the specific behaviors of user groups can be refined, such as users who browse frequently but have low purchase rates, or active users with high conversion rates. Through this refinement, marketing strategies can be further optimized to meet the needs of different user groups, thereby improving users' purchase conversion rates and obtaining a refined user behavior dataset.

[0026] formula The benefit is that by adjusting the weight coefficient and , which can flexibly reflect the relative importance of device fingerprints and social media links in user identity authentication. Such weight adjustment helps to accurately determine the user's true identity; parameter is the user identification index, which is calculated by the weighted average of the device fingerprint score and the social media link score; the parameter The device fingerprint score is obtained by analyzing the physical and behavioral characteristics of the user's device; parameter is the social media link score, which is obtained by evaluating the user's activity and interactivity on the social media platform; parameter The weight coefficient of the device fingerprint score is set based on the importance of the device fingerprint in identity authentication; parameter The weight coefficient of the social media link score is determined according to the influence of social media activities on user identity authentication.

[0027] Calculation process: Device fingerprint score of a user , social media link score , weight coefficient and , then the identity index The calculation of is as follows:

[0028] The result shows that after combining the device fingerprint and social media link scores, the user's identity recognition index is 0.5167, which indicates that the user's real identity has a high verification credibility. This value in the user identity link set can be used as a basis for further analysis and user behavior prediction.

[0029] Based on each user's identity identification index, a user identity link set is generated. This link set is a combination of user identity data and attribute association information. This information is used for further behavior pattern analysis and formulation of marketing strategies. Through the user identity link set, the company can more accurately identify target customer groups, achieve more efficient advertising and product recommendations, thereby increasing operating income. The final user identity link set will be directly applied to the customer relationship management system.

[0030] The steps to obtain consumer behavior groups are: Based on the user identity link set, a clustering algorithm is applied to analyze user data and generate clustering analysis results; According to the cluster analysis results, user behavior patterns are distinguished, and users are grouped according to behavior similarities to obtain consumer behavior groups.

[0031] Specifically, the user identity link set is received and organized. This data set integrates user identity authentication information and its related attribute data. First, these data must be preprocessed to ensure data quality and consistency. The preprocessing steps include missing value processing, outlier detection, and data normalization. The key to this step is to optimize the structure of the data to adapt to the subsequent clustering algorithm. The preprocessed data set will be used for clustering analysis, such as K-means clustering, because of its high efficiency and scalability in user behavior analysis. Before the algorithm is implemented, the number of clusters needs to be determined, which is usually determined by the distribution characteristics of the data itself, such as using the elbow rule to determine the optimal number of clusters, and then the algorithm is applied to iteratively process the user data until the cluster center is stable and the cluster analysis results are generated.

[0032] Based on the generated cluster analysis results, further refine the analysis to distinguish the behavior patterns of each user. The core of this step is to understand the commonalities of user behavior in each cluster and its internal connections. The analysis work includes but is not limited to evaluating the frequency, continuity and correlation of user behavior with specific products or services in each cluster. Each cluster represents one or more specific consumer behavior patterns, which reflect the behavioral characteristics and consumption tendencies of users on the platform. By analyzing these patterns in detail, users can be effectively grouped according to behavioral similarities, thereby obtaining segmented consumer behavior groups. These groups provide a basis for subsequent market strategies, ensuring that marketing activities can be more accurately targeted at specific user groups.

[0033] The steps to obtain the preference prediction model are: According to the consumer behavior groups, the decision tree is used to analyze the shopping frequency and product type of each data point, and then the product preference index of each consumer is calculated. The calculation formula is:

[0034] in, Indicates The product preference index of consumers, For the The shopping frequency of each data point, is the score of the corresponding product type, is the total number of data points.

[0035] Based on the product preference index, the basis for each consumer’s preference for future products is established, and a preference prediction model is generated.

[0036] Specifically, after the data is imported, the data format is checked first to ensure that the date and timestamp formats of all records are correct. Then, any null values ​​or outliers in the purchase history are identified and processed. For missing shopping frequency information, it is estimated by finding records of other dates for the same consumer. For product type data, the consistency of its classification labels is checked to ensure that the same product type is not misclassified due to improper naming. After these steps, the accuracy and consistency of the data set are confirmed to generate a complete consumer behavior data set.

[0037] In the formula, the parameter is the shopping frequency, obtained through user purchase data collected from e-commerce platforms; parameter Product type score, determined based on product market positioning and sales data; parameters is the total number of data points, i.e., the number of user interactions, directly obtained from the dataset; formula The calculation process: Set is the shopping frequency. For a specific consumer, the shopping frequency is [3,2,4], which indicates the number of times the three products are purchased; is the product type score, which is [1, 2, 1.5], indicating the preference weight of different products; is the total number of data points, here it is 3; the specific calculation process is:

[0038] The results show that consumers’ average product preference index is 4.33, reflecting their average preference strength for the types of products they purchase.

[0039] Based on the decision tree algorithm, we first determine the input order and processing level of each variable. Shopping frequency and product type are the main variables. These variables are directly input into the model, and then through a series of node decision processes, each node represents an attribute decision point. These points are divided based on the specific values ​​in the data set, thus forming different paths to represent different consumer behavior patterns. In this way, the model can predict consumers' possible purchasing preferences based on existing data and generate a preference prediction model.

[0040] The steps to obtain market demand forecast results are: Past sales data and user rating information for each product category are extracted to obtain a basic set for data analysis.

[0041] The demand probability is calculated for each product category in the data analysis base set, and the calculation formula is:

[0042] in, Indicates The demand probability of product categories, For the The sum of the scores of product categories in the data analysis base set, is the total number of product categories, The sum of the ratings for other product categories.

[0043] According to the demand probability of each product category, the expected market demand is calculated in combination with the preference prediction model to obtain the market demand forecast result.

[0044] Specifically, extracting historical sales data for each product category involves summarizing the sales volume of the past five years by product category from the company's sales database. User rating information is obtained by obtaining the scores and number of comments from the user feedback section of the online platform through the API. The data includes the number of units sold and the average rating of each product. These data are calculated through the arithmetic mean and sum to obtain the comprehensive score and sales index of each product, generating a basic set for data analysis. This set includes fields such as product number, comprehensive score, and sales index.

[0045] formula The benefit of is that it reflects the demand probability between product categories through relative proportions, not only considering the total score of each product, but also highlighting the impact of competition intensity through the proportion of relative scores, thus enhancing the perception of competition dynamics; parameter For the The sum of the scores of a product category in the data analysis base set is obtained by extracting all user scores of this product category from the database and summing them up; parameter is the sum of the scores of other product categories, which is also calculated by querying the database and accumulating the score data; parameter is the total number of product categories, obtained by counting the number of all independent product categories in the database.

[0046] Calculation process: Assume that the total scores of product A and product B are 200 and 300 respectively, and calculate the demand probability of product A:

[0047] The results show that under the current market conditions, the demand probability of product A is 40%. If the demand probability is greater than 0.5, it means that the demand is high, and if it is less than 0.5, it means that the demand is average. This model helps merchants adjust their inventory strategies and marketing priorities.

[0048] When applying the preference prediction model, first define the model's input variables as user ratings and product types, and the output variable as predicted demand; process the data, including filtering, sorting, and grouping operations, so that user preference data can be mapped to specific product types; finally, the output includes the predicted market demand for each product. This value reflects the expected sales volume and helps companies make adjustments to their production and marketing strategies.

[0049] The steps to obtain sales strategy adjustment suggestions are as follows: Integrate market demand forecast results with past sales data, analyze sales trends and market demand differences of product categories, and generate basic data sets.

[0050] Through time series analysis, we predict future demand changes and market dynamics for each product category based on the basic data set and generate market trend change forecasts.

[0051] According to the forecast results of market trend changes, combined with product characteristics and market competition conditions, customized sales strategy adjustment suggestions are made for each product. Sales strategy adjustment suggestions include adjustments to pricing, promotional activities and inventory strategies.

[0052] Specifically, we collect and integrate market demand forecast results and past sales data, and analyze the sales trends and market demand differences of each product category. First, we extract the sales records of each product category in the past year from the sales database, including sales volume, sales revenue and return rate. Then, we compare these historical data with the current market demand forecast, identify product categories with obvious demand increase or decrease trends, and analyze the correlation between these trends and market promotion activities, seasonal changes and other factors. Through this analysis process, we form a basic data set that reflects the sales trends of product categories and the differences in market demand.

[0053] Based on the basic data set generated above, time series analysis methods are used to predict future market dynamics and changes in demand for each product category. This process includes importing the time series data in the basic data set into the analysis tool, using a linear regression model to estimate the trend component of the time series, analyzing the seasonal fluctuations of each product category, and adjusting the model parameters to accurately match the actual sales trends of historical data. Finally, through these analyses and calculations, market trend change forecasts that describe future demand changes and market dynamics for each product category are generated.

[0054] According to the forecast results of market trend changes, combined with the characteristics of each product category and the current market competition situation, sales strategy adjustment suggestions are formulated. Specific operations include analyzing the profit margin, market share and competition intensity of each product, adjusting the pricing strategy according to the forecast results, optimizing promotional activities to enhance market appeal, and adjusting the inventory strategy to cope with the predicted market demand. These adjustment suggestions take into account cost control and market competitiveness, and aim to optimize the market performance of the product. The generated sales strategy adjustment suggestions include specific pricing, promotion and inventory operation suggestions.

[0055] The steps that affect the results are: Extracting posting data of related products from social media platforms to obtain an original posting dataset; Perform text analysis on the original posting data set, manually annotate the sentiment of each posting data as positive, neutral or negative, and calculate the sentiment score of each post. The calculation formula is:

[0056] in, represents the sentiment score, For the Manually annotated sentiment scores of posts, is the user interaction factor of the post, is the time factor, is the total number of posts analyzed; Based on the sentiment score of each post and the frequency of occurrence of keywords, the impact results on product sales are generated.

[0057] Specifically, we extract posting data of related products from social media platforms, including the text content of posts and user interaction data (such as the number of likes and comments), to obtain the original posting dataset. We then access specific platform data through public social media APIs, filter posts, and select posts containing specific keywords, such as product names or related events. We then collect the publication dates, number of likes, number of comments, and number of reposts of these posts. We then organize these data using statistical software to establish a data foundation for subsequent sentiment analysis and impact assessment.

[0058] formula The benefit is that by weighting user interaction data and post timeliness, the dynamics and accuracy of sentiment analysis are enhanced, making the results more time-relevant; Calculation process: First calculate the sentiment score of each post , each rating is annotated, and then a user interaction factor is calculated for each post , which is a function of the number of likes and comments, and then introduces the time factor , which is the gap between the publishing time and the current time, is used to adjust the real-time influence of the post. This formula is applied to all posts. There are 5 posts with sentiment scores of 2, 3, 2, 4, and 5, user interaction factors of 100, 150, 120, 130, and 160, and time factors of 0.8, 0.9, 0.85, 0.95, and 0.9. Substitute them into the formula for calculation. The calculation process is as follows:

[0059] The calculation will show the overall trend of social media sentiment scores.

[0060] Based on the sentiment score of each post and the frequency of keyword occurrence, we evaluate how the sentiment and theme of social media posts affect the market performance of products, generate predicted impact results on product sales, compare the sentiment score with sales data, identify the correlation between sentiment highs and sales highs, and thus derive a specific assessment of the impact of post sentiment on sales. Ultimately, we obtain an impact assessment report that includes specific data analysis results and recommended market strategy adjustments, providing a guide to action for the marketing department.

[0061] The present invention provides an AI system, comprising: The user behavior analysis module is used to extract user click-through rates, search frequencies, and purchase records from e-commerce platforms, calculate the activity frequency and purchase habits of each user, identify user identities and associate attributes based on device fingerprints and social media links, and generate user identity link sets; through cluster analysis of user identity link sets, consumer behavior patterns are distinguished and consumer behavior groups are obtained; The preference prediction module is used to calculate the shopping frequency and product type of the data points in each consumer behavior group based on consumer behavior groups, establish a preference basis, and generate a preference prediction model; analyze the demand probability of each product category to obtain market demand prediction results; The market strategy module analyzes changes in market trends and adjusts sales strategies based on market demand forecast results; extracts posting data from social media platforms, performs sentiment analysis and keyword extraction, estimates the impact on product sales, and generates sales strategy adjustment recommendations.

[0062] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A big data analysis method based on an e-commerce platform, characterized in that: The following steps are involved: Collecting click-through rates, search frequencies, and purchase records of users on the e-commerce platform, segmenting the click-through rates, search frequencies, and purchase records to create a data set, using device fingerprints and social media links to identify user identities and associate attributes with the data set, generating a user identity link set, analyzing the user identity link set, distinguishing consumer behavior patterns through clustering, and obtaining consumer behavior groups; Based on the consumer behavior groups, the shopping frequency and product type of each data point are attributed calculated through a decision tree, the user's preference basis for future products is established, and a preference prediction model is generated; the product demand probability is analyzed, the demand for each product category is calculated, and the market demand prediction result is obtained; Based on the market demand forecast results, market trend change forecasts are performed, and sales strategies are formulated for each product in combination with past sales data and market trend change forecast results, generating sales strategy adjustment suggestions; Extract posting data from social media platforms, perform sentiment analysis and keyword extraction on the posting data, determine the emotional tendency and topic relevance of the posting data, estimate the impact of the posting data on product sales, and obtain impact results.

2. The big data analysis method based on the e-commerce platform according to claim 1 is characterized in that: The steps for obtaining the user identity link set are: Collect the click-through rate, search frequency, and purchase history of users on the e-commerce platform, segment and create data sets, and obtain user behavior data sets; Based on the user behavior dataset, the identity recognition index of each user is calculated using device fingerprints and social media links. The calculation formula is: in, is the user identification index, and They are device fingerprint score and social media link score, and is the adjustment coefficient; According to each user's identity identification index, a user identity link set is generated.

3. The big data analysis method based on the e-commerce platform according to claim 1 is characterized in that: The steps for obtaining the consumer behavior group are: Based on the user identity link set, applying a clustering algorithm to analyze user data to generate a clustering analysis result; According to the cluster analysis results, user behavior patterns are distinguished, and users are grouped according to behavior similarities to obtain consumer behavior groups.

4. The big data analysis method based on the e-commerce platform according to claim 1 is characterized in that: The steps for obtaining the preference prediction model are: According to the consumer behavior groups, the decision tree is used to perform attribute analysis on the shopping frequency and product type of each data point, and then the product preference index of each consumer is calculated. The calculation formula is: in, Indicates The product preference index of consumers, For the The shopping frequency of each data point, is the score of the corresponding product type, is the total number of data points; Based on the product preference index, a preference basis for each consumer for future products is established, and a preference prediction model is generated.

5. The big data analysis method based on the e-commerce platform according to claim 1 is characterized in that: The steps for obtaining the market demand forecast result are: Extract past sales data and user rating information for each product category to obtain a basic set of data analysis; The demand probability is calculated for each product category in the data analysis basic set, and the calculation formula is: in, Indicates The demand probability of product categories, For the The sum of the scores of product categories in the data analysis base set, is the total number of product categories, is the sum of the ratings for other product categories; According to the demand probability of each product category, the expected market demand is calculated in combination with the preference prediction model to obtain a market demand prediction result.

6. The big data analysis method based on the e-commerce platform according to claim 1 is characterized in that: The steps for obtaining the sales strategy adjustment suggestion are as follows: Integrate the market demand forecast results with past sales data, analyze the sales trends and market demand differences of product categories, and generate a basic data set; Through time series analysis, based on the basic data set, the future demand changes and market dynamics of each product category are predicted to generate market trend change prediction results; Based on the market trend change forecast results, combined with product characteristics and market competition conditions, sales strategy adjustment suggestions are customized for each product, and the sales strategy adjustment suggestions include adjusting pricing, promotional activities and inventory strategies.

7. The big data analysis method based on the e-commerce platform according to claim 1 is characterized in that: The steps for obtaining the impact results are: Extracting posting data of related products from social media platforms to obtain an original posting dataset; The original posting data set is subjected to text analysis, and the sentiment of each posting data is manually labeled as positive, neutral or negative, and the sentiment score of each post is calculated using the following formula: in, represents the sentiment score, For the Manually annotated sentiment scores of posts, is the user interaction factor of the post, is the time factor, is the total number of posts analyzed; The impact results on product sales are generated based on the sentiment score of each posting data and the frequency of occurrence of keywords.

8. An AI system according to any one of claims 1 to 7, characterized in that: include: The user behavior analysis module is used to extract user click-through rates, search frequencies, and purchase records from e-commerce platforms, calculate the activity frequency and purchase habits of each user, identify user identities and associate attributes based on device fingerprints and social media links, and generate user identity link sets; through cluster analysis of user identity link sets, consumer behavior patterns are distinguished and consumer behavior groups are obtained; The preference prediction module is used to calculate the shopping frequency and product type of the data points in each consumer behavior group based on the consumer behavior group, establish the preference basis, and generate the preference prediction model; Analyze the demand probability of each product category to obtain market demand forecast results; The market strategy module analyzes market trend changes and adjusts sales strategies based on the market demand forecast results; Extract posting data from social media platforms, perform sentiment analysis and keyword extraction, estimate the impact on product sales, and generate sales strategy adjustment recommendations.

Images