An e-commerce demand analysis method based on a large language model

Abstract

The application provides an e-commerce demand analysis method and system based on a large language model, which optimizes product characteristics and market strategies by analyzing user-generated content (UGC). The method includes: first, collecting UGC data such as product reviews and user Q&A from e-commerce platforms. Then, using a large language model to deeply analyze these data, extract product attributes and sentiment information, and convert them into structured data. Next, generate a dual importance matrix at the market level to evaluate the importance of product attributes in purchase decision and post-purchase satisfaction. Further, for the products of the merchant, calculate the actual performance value of the attributes, generate a dual importance-performance matrix at the product level, and identify optimization needs. Finally, the analysis results are visualized and a report is generated to provide market insights and decision support for merchants. This method helps merchants accurately position product optimization direction and improve market competitiveness.

Description

TECHNICAL FIELD

[0001] The present application relates to e-commerce big data technology and natural language processing (NLP) technology, and particularly relates to an e-commerce demand analysis method based on a large language model. BACKGROUND

[0002] In today's fierce market competition environment, understanding user demand has become the core of enterprises to improve the competitiveness of products and services. Especially in the e-commerce field, user-generated content (UGC) as an important demand feedback channel is increasingly valued by enterprises. Compared with traditional offline research, user-generated content has the advantages of wide coverage, less time limit and strong real-time performance, which can more intuitively reflect consumers' actual needs, improvement suggestions and future expectations. The comments and feedback information generated by users on e-commerce platforms can provide valuable insights into products and services, especially in product design and optimization, customer experience improvement, etc.

[0003] With the continuous development of big data technology and natural language processing (NLP) technology, enterprises can extract valuable demand information from massive user-generated content, thereby providing data support for product innovation and service optimization. However, how to effectively mine user demand, evaluate its priority, and accurately interpret consumer attitudes remains a difficult problem to be solved. In view of this problem, how to help enterprises efficiently and accurately conduct demand analysis through innovative methodology and technical means has become an important topic in theoretical research and practical exploration.

[0004] In recent years, natural language processing technology based on large language models has developed rapidly and has become a powerful tool for processing various language tasks. Large language models have strong context understanding capabilities and are widely used in text classification, sentiment analysis, information extraction and other fields, showing strong language processing capabilities. In the e-commerce field, text analysis technology based on large language models can accurately extract key information from complex user-generated content. Large language models can capture subtle language features in large-scale data through deep learning technology, identify potential patterns in text, and help enterprises automatically extract user demand points from user-generated content such as comments and questions. Based on this technology, enterprises can quickly obtain user demand dynamics from data and optimize products and services accordingly.

[0005] In the aspect of demand mining based on user-generated content, existing technical solutions mainly focus on two core steps: comment data processing and demand mining analysis, that is, first processing the comment data into structured data, and then presenting the demand analysis results through statistical analysis methods.

[0006] In terms of review data processing, current research on product demand mining based on online reviews mainly relies on keyword extraction techniques and sentiment analysis methods. Common keyword extraction methods include TF-IDF and LDA topic models, which extract key information from reviews by counting word frequencies or mining latent topics. In addition, some researchers use deep learning models such as BERT to perform sentiment analysis and information extraction using pre-trained language models to extract user sentiment and focus points.

[0007] In terms of demand mining analysis, current research mainly relies on the Kano model and the IPA (Importance-Performance Analysis) method to evaluate the impact of product attributes on user satisfaction and their importance. The Kano model classifies product features into basic, expected, and exciting types based on user feedback, thereby analyzing the impact of these features on satisfaction. The IPA method helps businesses identify the most valuable features to consumers by comparing product attribute performance with user expectations, thereby optimizing resource allocation.

[0008] Traditional views often consider customer satisfaction as a linear function of product or service quality, i.e., the higher the perceived quality, the higher the satisfaction. However, simply improving product attributes may not effectively improve customer satisfaction, as the nature of customer demand may not be linear or symmetric. Academic research has shown that there may be non-linear or asymmetric characteristics between attribute performance and satisfaction. Kano et al. proposed the Kano model in 1984 to analyze this relationship and classify demands. Although the Kano model does not directly measure the importance of attributes, it can provide a basis for prioritization based on the degree of impact on satisfaction or dissatisfaction.

[0009] The Kano model classifies product attributes into five categories (see Figure 1 ): exciting demand (A), expected demand (O), basic demand (M), indifferent demand (I), and reverse demand (R), corresponding to different satisfaction impact mechanisms. Exciting demand refers to features that exceed user expectations, which can greatly improve satisfaction, while their absence does not cause dissatisfaction. Expected demand is positively correlated with satisfaction, users have clear expectations for it, and the better the performance, the higher the satisfaction. Basic demand is the user's basic expectation, although its good performance does not significantly improve satisfaction, but its absence will cause strong dissatisfaction. Indifferent demand has little effect on satisfaction, and users usually have a neutral attitude towards its existence or absence. Reverse demand refers to users' negative attitude towards the existence of certain features, which may reduce satisfaction.

[0010] The IPA method (Importance-Performance Analysis) was proposed by Martilla and James in 1977 as a tool for classifying product or service attributes and determining optimization priorities. This method constructs a two-dimensional matrix with the horizontal axis representing product performance and the vertical axis representing user-perceived importance, and divides attributes into different quadrants (see Figure 2 ). In the Q1 (Keep Up) quadrant, high-importance and high-performance attributes are the core strengths that the business should maintain, and their stability should be ensured to avoid performance decline. In the Q2 (Improve First) quadrant, high-importance but low-performance attributes need to be optimized first, as these factors are crucial to users but currently underperforming, so resources should be invested to improve customer satisfaction. In the Q3 (Irrelevant) quadrant, low-importance and low-performance attributes have limited impact on customers, so resources can be allocated to more critical areas rather than focusing on improving these features. Finally, the Q4 (Reduce Appropriately) quadrant contains low-importance but high-performance attributes, which, although performing well, are not highly focused on by users, so resources can be reduced to avoid waste.

[0011] Current online review-based product demand mining methods mainly include two core parts: review data processing and demand mining analysis. However, this process has the following significant limitations, which limit the accurate understanding of user demand by businesses:

[0012] 1. Incomplete structured review data processing

[0013] Existing methods often use keyword extraction techniques (such as TF-IDF, LDA topic model, etc.) for review analysis. Although these methods have lower computational costs and can quickly process large-scale review data, they have obvious limitations. Especially in the increasingly homogenized market environment, relying solely on keyword frequency for analysis can overlook low-frequency but critical product features, which are often important for product differentiation. Therefore, traditional methods may lead to incomplete demand identification, which can affect product optimization direction.

[0014] 2. Error accumulation in sentiment analysis

[0015] Existing sentiment analysis methods usually use a pipeline approach, i.e., aspect word extraction is performed first, and then sentiment classification is performed separately. However, this step-by-step processing approach can generate errors at each stage and amplify them in subsequent processing, ultimately affecting the accuracy of overall demand analysis. For example, if aspect word extraction fails to accurately identify key attributes in user reviews, the subsequent sentiment analysis results may also be biased, leading to biased understanding of user demand by businesses.

[0016] 3. Single perspective, ignoring pre-purchase demand analysis

[0017] Current demand analysis methods mainly rely on post-purchase user reviews, ignoring the pre-purchase focus and decision factors of consumers. However, in actual online shopping scenarios, consumers' purchase decisions are influenced by the expectancy theory, that is, forming expectations based on product descriptions, user reviews, and other information before purchase, while post-purchase reviews reflect the gap between expectations and actual experience. Therefore, demand analysis based only on post-purchase reviews may not fully capture consumers' true needs, and may even lead businesses to focus on the wrong direction when optimizing products. For example, some key attributes may be the core focus of consumers before purchase, but if the performance of the attribute meets expectations, consumers may not mention it in reviews, resulting in post-purchase reviews not fully reflecting true needs. Therefore, the lack of analysis of pre-purchase focus limits businesses' understanding of consumers' needs.

[0018] It should be noted that the information disclosed in the above background section is only for understanding the background of the present application, and therefore can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0019] The main purpose of the present application is to overcome the defects existing in the background art, and to provide an e-commerce demand analysis method based on a large language model, which optimizes the accuracy and comprehensiveness of demand analysis by introducing a dual perspective before and after purchase.

[0020] To achieve the above purpose, the present application adopts the following technical solutions:

[0021] An e-commerce demand analysis method based on a large language model, comprising the following steps:

[0022] S1, data collection: collecting user-generated content (UGC) of self-owned products and target competitors of merchants from e-commerce platforms, including product review data and user Q&A data;

[0023] S2, data processing and analysis: using a large language model to process and analyze the collected user-generated content, extracting product attributes and sentiment information therefrom, and converting the extracted information into structured data;

[0024] S3, generation of dual importance matrix at market level: based on the extracted structured data, calculating the pre-purchase attention and post-purchase importance of each product attribute, and generating a dual importance matrix at market level by combining the attention and importance, for evaluating the relative importance of different product attributes in consumer purchase decisions and satisfaction;

[0025] S4, product-level dual importance-performance matrix generation: combining the market-level dual importance matrix, calculate the actual performance value of each product attribute for the merchant's own products, and generate a product-level dual importance-performance matrix to identify product features that need to be optimized;

[0026] S5, result presentation and report generation: present the generated dual importance matrix and dual importance-performance matrix through visualization, and generate an analysis report to provide market insights and product optimization decision support for the merchant.

[0027] Further, in step S2, when processing and analyzing user-generated content using a large language model, the following steps are included:

[0028] Predefine the attribute list of product categories;

[0029] For each comment, use a preset prompt template to guide the large language model to identify the attributes involved in the comment and their corresponding sentiment polarity;

[0030] Convert the identified attributes and sentiment information into structured data.

[0031] Further, in step S3, when calculating the pre-purchase attention of each product attribute, the following steps are included:

[0032] Analyze user question data in the user question and answer module to count the mention frequency of each attribute;

[0033] Calculate the pre-purchase attention of each attribute based on the mention frequency.

[0034] Further, in step S3, when calculating the post-purchase importance of each product attribute, the following steps are included:

[0035] Based on the comment data, use a joint analysis method to model through multivariate regression analysis to calculate the influence of each attribute on customer satisfaction;

[0036] According to the difference of regression coefficients, determine the post-purchase importance of each attribute.

[0037] Further, in step S3, when generating a market-level dual importance matrix, the following steps are included:

[0038] Normalize the pre-purchase attention and post-purchase importance;

[0039] Draw the dual importance matrix with post-purchase importance as the horizontal axis and pre-purchase attention as the vertical axis;

[0040] According to the mean value in the matrix, four quadrants are divided, which correspond to core features, focus features, low-value features and potential optimization features respectively.

[0041] Further, in the step S4, when calculating the actual performance value of each product attribute, the following steps are specifically included:

[0042] Based on the comment data, the proportion of positive sentiment of each attribute is calculated as the actual performance value of the attribute.

[0043] Further, in the step S4, generating the dual importance-performance matrix at the product level specifically includes the following steps:

[0044] Determine the mean value of the pre-purchase importance and post-purchase importance of each attribute as the benchmark for evaluating the relative importance of the attribute; determine the weight adjustment factor of each attribute using the conditional judgment mechanism, which is dynamically adjusted based on the comparison results of the pre-purchase importance and post-purchase importance of the attribute with their respective mean values; according to the weighted calculation of the weight adjustment factor, pre-purchase importance and post-purchase importance of each attribute, the dual importance of each attribute is obtained.

[0045] Draw the dual importance-performance matrix at the product level with the dual importance of the attribute as the vertical axis and the actual performance value as the horizontal axis; wherein, the dual importance-performance matrix is divided into five modules, and the characteristics of each module are as follows:

[0046] The first module contains attributes with high dual importance and high actual performance value, indicating that these attributes have a large impact on consumer decision-making and perform well, and should be maintained;

[0047] The second module contains attributes with high dual importance but low actual performance value, indicating that these attributes have a large impact on consumer decision-making but perform poorly, and need to be improved first;

[0048] The third module contains attributes with high dual importance but low actual performance value, indicating that these attributes have some impact on consumers but do not perform as expected, and can be maintained at a low priority for optimization;

[0049] The fourth module contains attributes with low dual importance and low actual performance value, indicating that these attributes have little impact on consumer decision-making and perform generally, and can be maintained as unimportant;

[0050] The fifth module contains attributes with low dual importance but high actual performance value, indicating that these attributes have little impact on consumer decision-making but perform well, and can appropriately reduce investment.

[0051] Further, in the step S4, the dual importance of each attribute is calculated by weighting based on the pre-purchase attention, post-purchase importance, and a weight adjustment factor adjusted based on a condition judgment; and the condition judgment mechanism is specifically: when the pre-purchase importance and the post-purchase importance of the attribute are both higher than the average, a higher weight adjustment factor is given; when any importance is higher than the average, a medium weight adjustment factor is given; and when both are lower than the average, a lower weight adjustment factor is given.

[0052] Further, in the step S5, the result presentation and report generation specifically include the following steps:

[0053] The dual importance matrix at the market level and the dual importance-performance matrix at the product level are displayed through a visualization tool;

[0054] An analysis report is generated to provide market insight and product optimization decision support.

[0055] An e-commerce demand analysis system based on a large language model includes the following steps:

[0056] A data collection module: collects user-generated content (UGC) of self-owned products and target competitors from an e-commerce platform, including product review data and user Q&A data;

[0057] A data processing and analysis module: processes and analyzes the collected user-generated content using a large language model, extracts product attributes and sentiment information therefrom, and converts the extracted information into structured data;

[0058] A dual importance matrix generation module at the market level: based on the extracted structured data, the pre-purchase attention and the post-purchase importance of each product attribute are calculated, and a dual importance matrix at the market level is generated by combining the attention and the importance, for evaluating the relative importance of different product attributes in consumer purchase decision and satisfaction;

[0059] A dual importance-performance matrix generation module at the product level: based on the dual importance matrix at the market level, the actual performance value of each product attribute is calculated for the self-owned products, and a dual importance-performance matrix at the product level is generated, for identifying product features that need to be optimized;

[0060] A result presentation and report generation module: the generated dual importance matrix and dual importance-performance matrix are presented in a visual manner, and an analysis report is generated to provide market insight and product optimization decision support for the merchant.

[0061] The present application has the following beneficial effects:

[0062] The application provides an e-commerce demand analysis method and system based on a large language model, which breaks through the limitation of traditional methods relying only on post-purchase feedback, comprehensively identifies and analyzes key demands from the dual perspectives of consumers before and after purchase, and significantly improves the accuracy of decision support. The system collects product review data of the merchant's own products and target competitors and user question and answer data of the "Ask Everyone" module, performs text analysis based on a large language model, extracts product attributes and sentiment information, and converts them into structured data. Then, the system calculates the attention of users before purchase to each attribute, evaluates the influence of each attribute on customer satisfaction, generates a dual importance matrix at the market level, and reveals the role of each attribute in consumer purchase decision and post-purchase satisfaction. At the same time, based on the review data, the system calculates the actual performance of the product attributes, combines the dual importance index, and generates a dual importance-performance matrix at the product level to accurately position the product features that need to be optimized.

[0063] The e-commerce demand analysis method and system based on a large language model provided by the application optimizes the accuracy and comprehensiveness of demand analysis by introducing the dual perspectives before and after purchase. Specifically:

[0064] 1. Improve the accuracy of structured processing

[0065] The application uses a large language model (such as GPT, BERT) to perform deep semantic understanding and processing on user-generated content (UGC), avoiding the error accumulation problem in traditional pipeline methods. The large language model can more accurately extract product attributes and sentiment information from reviews, especially for low-frequency but important features, thereby improving the comprehensiveness and accuracy of demand analysis and providing higher-quality data support for merchants.

[0066] 2. Build a dual-perspective demand analysis framework

[0067] The application innovatively proposes a demand analysis method combining the perspectives before and after purchase, which comprehensively reveals the differences in consumer behavior during the purchase decision stage and the feedback stage. By calculating the pre-purchase attention (based on "Ask Everyone" data) and the influence of post-purchase attributes on user satisfaction (based on review sentiment analysis), a dual importance matrix at the market level is constructed to quantify the influence of different product attributes on consumer decision-making and satisfaction. At the same time, further combined with the actual performance of the product, a dual importance-performance matrix at the product level is generated to help merchants accurately identify which product attributes need to be optimized and provide scientific optimization directions.

[0068] 3. Provide accurate market insights and optimization strategy support

[0069] The analysis framework of the present application not only supports market-level demand comparison, helps enterprises identify key demand characteristics in the industry, and provides clear direction for new product design and market promotion, but also focuses on the importance of single product attributes and performance relationships, providing scientific basis for business resource allocation, product optimization and market competition strategy formulation. Through this method, businesses can more accurately understand the dynamic needs of consumers, improve market response capabilities, and enhance the market adaptability and competitiveness of products.

[0070] The present application overcomes the shortcomings of traditional single-dimensional analysis methods by integrating market-level and product-level demand analysis and proposing an innovative visualization solution, providing more comprehensive and accurate market insights for businesses, supporting scientific decision-making for product optimization, resource allocation and market promotion, and enhancing the market adaptability and competitiveness of products.

[0071] Other benefits of the embodiments of the present application will be further described below. BRIEF DESCRIPTION OF DRAWINGS

[0072] Figure 1 Existing Kano model schematic diagram.

[0073] Figure 2 Existing IPA method schematic diagram.

[0074] Figure 3 The system block diagram of the e-commerce demand analysis system based on large language model of the embodiments of the present application.

[0075] Figure 4 The dual importance matrix of the market level of the embodiments of the present application.

[0076] Figure 5 The dual importance-performance matrix of the product level of the intermediate process in the embodiments of the present application.

[0077] Figure 6 The final version of the dual importance-performance matrix of the product level of the embodiments of the present application. DETAILED DESCRIPTION

[0078] The following describes the embodiments of the present application in detail. It should be emphasized that the following description is only exemplary and is not intended to limit the scope of the present application and its applications.

[0079] In addition, the terms "first", "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the embodiments of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0080] Existing e-commerce demand analysis methods usually focus on user feedback after purchase, ignoring the consumer's concerns and decision-making process before purchase. This single-dimensional analysis method often cannot fully reflect the real needs of consumers, leading to an inaccurate understanding of market trends by businesses, affecting the effectiveness of product optimization and market promotion. The present invention proposes an innovative method and system for demand analysis based on e-commerce user-generated content (UGC) to help businesses. Unlike existing technologies, the present invention considers user needs from both pre-purchase and post-purchase stages, combining the powerful semantic understanding ability of large language models to accurately analyze pre-purchase concerns and post-purchase satisfaction influencing factors, overcoming the limitations of traditional methods that rely solely on post-purchase reviews. By constructing dual importance matrices at the market and product levels, the present invention provides more comprehensive and accurate demand analysis for businesses, helping them not only identify real feedback from consumers after purchase but also understand potential concerns before purchase.

[0081] Referring to Figures 3 to 6 The present invention provides an e-commerce demand analysis method based on a large language model, comprising the following steps:

[0082] Step S1, data collection: collecting user-generated content (UGC) of self-owned products and target competitors from e-commerce platforms, including product review data and user Q&A data;

[0083] Step S2, data processing and analysis: using a large language model to process and analyze the collected user-generated content, extracting product attributes and their sentiment information, and converting the extracted information into structured data;

[0084] Step S3, generation of market-level dual importance matrix: based on the extracted structured data, calculate the pre-purchase attention (as pre-purchase importance) and post-purchase importance of each product attribute, and generate a market-level dual importance matrix combining the attention and importance, for evaluating the relative importance of different product attributes in consumer purchase decision and satisfaction;

[0085] Step S4, generation of product-level dual importance-performance matrix: combining the market-level dual importance matrix, calculate the actual performance value of each product attribute for self-owned products, and generate a product-level dual importance-performance matrix for identifying product features that need to be optimized;

[0086] Step S5, result presentation and report generation: presenting the generated dual importance matrix and dual importance-performance matrix through visualization, and generating an analysis report to provide market insights and product optimization decision support for businesses.

[0087] In the preferred embodiment, in step S2, when the large language model is used to process and analyze the user-generated content, the following steps are included: predefining an attribute list of product categories; for each comment, using a preset prompt template to guide the large language model to identify the attributes involved in the comment and their corresponding sentiment polarity; converting the identified attributes and sentiment information into structured data.

[0088] In the preferred embodiment, in step S3, when calculating the pre-purchase attention of each product attribute, the following steps are included: analyzing user question data in the user question and answer module, and counting the mention frequency of each attribute; based on the mention frequency, calculating the pre-purchase attention of each attribute. When calculating the post-purchase importance of each product attribute, the following steps are included: based on the comment data, using a joint analysis method, modeling through multivariate regression analysis, calculating the influence degree of each attribute on customer satisfaction; according to the difference of regression coefficients, determining the post-purchase importance of each attribute. When generating the dual importance matrix at the market level, the following steps are included: normalizing the pre-purchase attention and post-purchase importance; taking the post-purchase importance as the horizontal axis and the pre-purchase attention as the vertical axis, drawing the dual importance matrix; according to the mean value in the matrix, dividing it into four quadrants, respectively corresponding to core features, attention point features, low value features and potential optimization features.

[0089] In some embodiments, in step S4, when generating the dual importance-performance matrix at the product level, the following steps are included: combining the dual importance matrix at the market level, calculating the dual importance weighted value of each attribute; taking the dual importance weighted value as the vertical axis and the actual performance value as the horizontal axis, drawing the dual importance-performance matrix; according to the mean value in the matrix, multiple quadrants can be divided to correspond to different product attribute categories or strategy guidance, as an example, four quadrants, respectively corresponding to continue to maintain demand, improve demand, observe demand and potential demand.

[0090] In the preferred embodiment, in the step S4, when calculating the actual performance value of each product attribute, the proportion of positive sentiment of each attribute is calculated based on the review data as the actual performance value of the attribute. Generating the dual importance-performance matrix at the product level specifically includes the following steps: determining the mean of the pre-purchase importance and post-purchase importance of each attribute as the benchmark for evaluating the relative importance of the attribute; using a conditional judgment mechanism to determine the weight adjustment factor of each attribute, which is dynamically adjusted based on the comparison results of the pre-purchase importance and post-purchase importance of the attribute with the respective mean; obtaining the dual importance of each attribute through weighted calculation based on the weight adjustment factor, pre-purchase importance and post-purchase importance of each attribute; drawing the dual importance-performance matrix at the product level with the dual importance of the attribute as the vertical axis and the actual performance value as the horizontal axis; wherein the dual importance-performance matrix is divided into five modules, and each module has the following characteristics: the first module contains attributes with high dual importance and high actual performance value, indicating that these attributes have a large impact on consumer decision-making and perform well, and should be maintained; the second module contains attributes with high dual importance but low actual performance value, indicating that these attributes have a large impact on consumer decision-making but perform poorly, and need to be improved first; the third module contains attributes with high dual importance but low actual performance value, indicating that these attributes have some impact on consumers but do not perform as expected, and can be kept at a low optimization priority; the fourth module contains attributes with low dual importance and low actual performance value, indicating that these attributes have a small impact on consumer decision-making and perform generally, and can be kept as unimportant; the fifth module contains attributes with low dual importance but high actual performance value, indicating that these attributes have a small impact on consumer decision-making but perform well, and can be appropriately reduced in investment.

[0091] In the preferred embodiment, the dual importance of each attribute is obtained through weighted calculation based on the pre-purchase importance, post-purchase importance and weight adjustment factor adjusted based on the conditional judgment; the conditional judgment mechanism specifically includes: when the pre-purchase importance and post-purchase importance of an attribute are both higher than their mean, a higher weight adjustment factor (θ) is assigned to highlight the importance of the attribute; when either importance is higher than its mean, a medium weight adjustment factor (θ) is assigned to show that the attribute is important in a certain stage; when both are lower than their mean, a lower weight adjustment factor (θ) is assigned to indicate that the attribute has a smaller impact on overall demand.

[0092] In the step S5, the result presentation and report generation specifically includes the following steps: displaying the dual importance matrix at the market level and the dual importance-performance matrix at the product level through a visualization tool; generating an analysis report to provide market insights and product optimization decision support.

[0093] The embodiment of the application also provides an e-commerce demand analysis system based on a large language model, comprising the following steps:

[0094] A data acquisition module: user generated content (UGC) of a self-owned product of a merchant and a target competitor is collected from an e-commerce platform, including product review data and user Q&A data;

[0095] A data processing and analysis module: the collected user generated content is processed and analyzed by using a large language model, product attributes and sentiment information thereof are extracted, and the extracted information is converted into structured data;

[0096] A double importance matrix generation module at a market level: based on the extracted structured data, pre-purchase attention and post-purchase importance of each product attribute are calculated, and a double importance matrix at a market level is generated by combining the attention and the importance, which is used to evaluate the relative importance of different product attributes in consumer purchase decision and satisfaction;

[0097] A double importance-performance matrix generation module at a product level: the actual performance value of each product attribute is calculated for the self-owned product of the merchant by combining the double importance matrix at the market level, and a double importance-performance matrix at a product level is generated, which is used to identify product features that need to be optimized;

[0098] A result presentation and report generation module: the generated double importance matrix and double importance-performance matrix are presented in a visual manner, and an analysis report is generated, which provides market insights and product optimization decision support for the merchant.

[0099] The specific embodiments of the application are further described below.

[0100] System framework

[0101] The technical solution of the application comprises the following main modules:

[0102] 1. Data acquisition module

[0103] This module is responsible for obtaining product review data and review star ratings of the merchant's product and the target competitor, as well as user Q&A data in the "Ask Everyone" module. By scraping product reviews and user Q&A content from e-commerce platforms, rich data support is provided for subsequent demand analysis.

[0104] 2. Data processing and analysis module

[0105] The module uses a large language model to deeply process the collected user-generated content (UGC), extracts the product attributes and sentiment information involved in each comment, and extracts the attributes involved in the "Ask Everyone" questions. These information is organized into structured data. This module is the core of demand analysis, ensuring that high-quality, in-depth valuable insight information is extracted from unstructured data.

[0106] 3. Market-level dual importance matrix generation module

[0107] This module combines the attention and satisfaction of users before and after purchase, and uses the data of the merchant's own products and similar competitors to draw a market-level dual importance matrix. The specific steps are as follows:

[0108] Calculation of attention before purchase: By analyzing the user questions in "Ask Everyone", the user's attention to each attribute before purchase is measured (for example, Taobao and Jingdong platforms).

[0109] Calculation of importance after purchase: Based on comment data and comment star ratings, a joint analysis method is used to identify the impact of each attribute on overall customer satisfaction.

[0110] Drawing of market-level dual importance matrix: The horizontal axis of this matrix represents the impact of each attribute on overall customer satisfaction, and the vertical axis represents the user's attention before purchase, forming a market-dimension dual importance matrix, helping the merchant to fully understand market demand.

[0111] 4. Product-level dual importance-performance matrix generation module

[0112] Based on the market-level dual importance matrix, the actual performance of each attribute is calculated for the merchant's own product, and combined with the dual importance matrix, a product-dimension performance analysis matrix is generated. The specific steps are as follows:

[0113] Calculation of attribute performance: Based on comment data, the performance value of each attribute is calculated, such as the actual effect of the product, quality, etc.

[0114] Calculation of dual importance weighted value: Combined with the data in the market-level dual importance matrix, the weighted value of each attribute is calculated.

[0115] Drawing of product-level dual importance-performance matrix: The horizontal axis of this matrix represents the actual performance value of each attribute, and the vertical axis represents the weighted value of the dual importance, finally forming a product-dimension dual importance-performance matrix, helping the merchant to identify the priority of product optimization.

[0116] 5. Result presentation and report generation module

[0117] The module presents the analysis results to the merchant through visualization, including a dual importance matrix at the market level and a dual importance-performance matrix at the product level. Through these matrices, the merchant can clearly identify key needs and provide precise support for product optimization decisions based on the analysis results.

[0118] Technical implementation examples

[0119] 1. Data collection and preprocessing:

[0120] The data collection module obtains review data and "Ask Everyone" module data from e-commerce platforms through API or crawler technology. The number of crawls is determined by the merchant's preset time range or data volume to ensure that the collected data covers sufficient sample size. When processing review data, the system first performs data cleaning, removing duplicate reviews, removing reviews longer than 100 characters (usually template comments), and excluding other irrelevant data (such as garbled code, repeated output, etc.) to ensure data quality and analysis accuracy.

[0121] 2. Information extraction of text data:

[0122] The data processing module uses a generative large language model (such as DeepSeek or GPT) to perform in-depth analysis of text data. First, the system will predefine a list of attributes related to different product categories, such as "nutritional ingredients," "product taste," "shelf life" for plant milk, and "product texture," "functional effects" for toner. Then, for each review, the system uses a pre-defined prompt template to guide the large language model to identify the relevant attributes mentioned in the review and their corresponding sentiment polarity (such as positive, neutral, negative). In each prompt template, the system provides few-shot learning examples to enhance the accuracy of attribute recognition. Each review may contain multiple attributes and sentiment information. For example, the comment "This product has a long shelf life and tastes good" will be parsed as {[shelf life, positive}, {product taste, positive}]. Similarly, for "Ask Everyone" data, the system will focus on extracting attribute information, ultimately generating structured attribute and sentiment analysis results.

[0123] 3. Generation of dual importance matrix:

[0124] This part is based on the merchant's own products and the review data of similar competitors to measure market conditions and draw a dual importance matrix at the market level. It mainly includes the following three steps: calculation of pre-purchase importance, calculation of post-purchase importance, and generation of the final dual importance matrix at the market level.

[0125] Calculation of pre-purchase importance:

[0126] The importance before purchase is calculated by the "Ask Everyone" data of the e-commerce platform. This data reflects the product attributes that users actively pay attention to before purchase. We measure the attention of users to each attribute by analyzing the frequency of mentioning each attribute in the question data. Specifically, we can use two measurement methods:

[0127] Wide frequency proportion: Calculate the proportion of the number of mentions of a certain attribute to the total number of mentions of all attributes.

[0128] Relative frequency proportion: Measure the overall mention frequency of a certain attribute in all products, and further investigate whether the attribute is mainly concentrated in a few products or evenly distributed in multiple products.

[0129] According to the proportion of mention frequency, we get the pre-purchase attention of each attribute, and calculate the importance of the attribute before purchase pre_importance.

[0130] Calculation of importance after purchase:

[0131] The importance after purchase is measured by user review data, and the joint analysis method is used to determine the relative contribution of each attribute to customer satisfaction. Through multivariate regression analysis modeling, the satisfaction and dissatisfaction of attributes in the review and the star rating are used as data input to calculate the influence of each attribute on customer satisfaction. The basic form of the model is as follows:

[0132]

[0133] Where y j represents the star rating of review j (i.e. customer satisfaction), and are the regression coefficients of attribute i, representing the influence of attribute i on customer satisfaction in the case of satisfaction (positive) and dissatisfaction (negative). and record the satisfaction and dissatisfaction of attribute i in review j, and a is the intercept term of the model, representing the baseline value of user rating when all attribute sentiment values are 0, which is estimated by ordinary least squares (OLS) and automatically calculated by the regression model.

[0134] The difference between the two regression coefficients of a certain attribute i is used to measure the influence of each attribute on satisfaction. If the difference between the two regression coefficients of a certain attribute i is larger, it means that the performance of this attribute in the review significantly affects the user's sentiment, and it is considered a key attribute. Finally, by calculating the difference between the positive and negative influences of each attribute, we get the importance of each attribute after purchase post_importance.

[0135] Drawing of double importance matrix at market level:

[0136] According to the previous calculation steps, pre_importance and post_importance of each attribute at the market level are obtained. The horizontal axis represents the impact of the attribute on customer satisfaction after purchase, and the vertical axis represents the user's attention before purchase. Through normalization processing, two values corresponding to each attribute are obtained, and the mean value in the matrix is marked with two dotted lines, representing the average level of the market. According to these two dotted lines, the matrix is divided into four quadrants, and finally a double importance matrix at the market level is formed as shown in Figure 4 , providing guidance for businesses on demand priorities.

[0137] Demand analysis of the four quadrants:

[0138] (1) Q1 quadrant: Core features

[0139] Features in this quadrant have important influence on both consumer purchase decision and post-purchase satisfaction. They are the most concerned characteristics of consumers when choosing products, and are also the key factors that ultimately affect customer satisfaction. Enterprises should prioritize optimizing these features to meet the core needs of consumers.

[0140] For example: In a plant milk product, "nutritional ingredients" may be frequently mentioned by consumers before purchase, and also show strong emotional feedback in post-purchase comments. By improving the performance of this feature, such as adding more nutritional ingredients or highlighting its health benefits, the consumer's purchase decision and satisfaction can be effectively improved.

[0141] (2) Q2 quadrant: Attention features

[0142] These features have high attention before purchase, but have little impact on customer satisfaction. Such attributes are usually of interest to consumers before purchase, but their actual effect may not be as expected to affect overall satisfaction. Businesses can improve the performance of these features while maintaining reasonable cost investment.

[0143] For example: The "packaging design" of plant milk may attract high attention from consumers before purchase, but if the packaging design does not significantly affect the function or taste of the product in actual use, its contribution to customer satisfaction is small.

[0144] (3) Q3 quadrant: Low-value features

[0145] Features in this quadrant have low attention before purchase and little impact on customer satisfaction. Generally, these features can be considered moderately in product design and optimization, but not too many resources should be invested. For example: In plant milk, "product appearance" may have low mention frequency before purchase, and its impact on the actual experience of consumers is small, so it can be considered as a secondary optimization feature.

[0146] (4) Q4 quadrant: potential optimization features

[0147] These features have lower pre-purchase attention but have a greater impact on customer satisfaction. Merchants can enhance the overall satisfaction of products by optimizing these features, especially in areas that users do not expect.

[0148] For example: The "shelf life" of plant milk may have lower pre-purchase attention, but if consumers find that the product has a short shelf life during use, it may have a greater negative impact on their satisfaction. Merchants can improve the performance of this feature by extending the shelf life or improving storage methods to enhance the consumer experience.

[0149] Comparative analysis

[0150] Unlike previous methods of identifying important features based only on post-purchase indicators (post_importance), the invention classifies features based on a dual perspective of pre-purchase and post-purchase (post_importance and pre_importance), providing a more comprehensive classification. In traditional methods, features are usually divided into core features and low-value features, only considering the needs of consumers after purchase. However, by combining pre-purchase and post-purchase perspectives, the invention not only identifies core features, but also further subdivides them into attention features, low-value features, and potential optimization features, providing more accurate demand analysis. This dual perspective analysis method enables merchants to better understand consumer needs, optimize product characteristics, and improve the accuracy and market adaptability of decision-making.

[0151] Table 1: Comparison of dual importance matrix and single importance features

[0152]

[0153] 4. Construction of product-level dual importance-performance matrix:

[0154] Based on the market-level dual importance matrix, the system calculates the actual performance of each attribute based on the merchant's own product review data, and combines the dual importance index to draw a product-level dual importance-performance matrix. This process focuses on the merchant's product itself, and the specific steps are as follows:

[0155] Dual importance calculation:

[0156] First, define several variables:

[0157] pre_importance i : pre-purchase importance of attribute (i)

[0158] post_importancei : Importance of attribute (i) after purchase

[0159] pre_importancce_mean: Mean of importance of all attributes before purchase

[0160] post_importance_mean: Mean of importance of all attributes after purchase

[0161] θ: Weight based on conditional judgment adjustment, taking values 0, 1 or 2. The calculation formula is as follows:

[0162]

[0163] w1 + w2 = 1

[0164] The calculation of θ is as follows:

[0165]

[0166] Explanation:

[0167] Weighted calculation: First, combine the importance before and after purchase through the weighting factors w1 and w2, both of which are by default one-half, which can be selected by the user.

[0168] θ value adjustment: Adjust the weight according to the performance of the attribute before and after purchase through the conditional judgment θ. Specifically, the value of θ is obtained by comparing the pre-purchase attention of attribute i with the pre-purchase mean of all attributes and the post-purchase importance with the post-purchase mean:

[0169] When the importance of attribute i before and after purchase is higher than the mean, θ is 2, indicating that the attribute has a greater impact on the merchant;

[0170] When one of the dimensions is higher than the mean, θ is 1, indicating that the attribute is relatively important;

[0171] When both dimensions are lower than the mean, θ is 0, indicating that the attribute has a smaller impact on demand.

[0172] The calculated dual_importance combines the influence of features before and after purchase and is weighted according to its importance.

[0173] Product performance calculation

[0174] The performance of the product is measured by the proportion of positive sentiment in the review data, which is referred to as performance. This sentiment proportion represents the proportion of positive feedback for each attribute in the total feedback as an indicator of user satisfaction. A higher proportion of positive sentiment indicates better performance of the attribute, and vice versa.

[0175] Drawing a dual importance-representation matrix

[0176] Based on the calculated dual_importance and the performance of each attribute, these metrics are normalized to create a product-level dual importance-performance matrix. For example... Figure 5 As shown, the horizontal axis of the matrix represents the actual performance value of each attribute, and the vertical axis represents the weighted double importance value.

[0177] exist Figure 5 In the graph, the dashed line on the horizontal axis represents the mean performance, while the vertical axis has two dashed lines: one representing one-third and the other two-thirds. These two dashed lines correspond to the calculation criteria in the dual importance calculation formula. These dashed lines divide the vertical axis into three regions, reflecting the importance of the attribute before and after the purchase.

[0178] Specifically, the vertical axis measures dual importance and is divided into three regions:

[0179] The first area: The attributes of this area have a lower than average importance both before and after purchase.

[0180] The second part: the attributes of this area, at least one of which has a greater than average importance before purchase and after purchase;

[0181] The third section: The attributes of this area have a greater than average importance both before and after purchase.

[0182] To ensure that each attribute accurately belongs to its corresponding region and that attributes within the same region can be reasonably compared, the dual importance calculation formula weights the two importance values ​​and multiplies them by one-third to prevent any attribute from deviating from its category.

[0183] Next, a detailed explanation will follow. Figure 5 Each quadrant in the table and the quadrants described in Table 2 ( Figure 5 (and Table 2 as an intermediate process of the preferred embodiment). Each quadrant is labeled in counter-clockwise order, and the quadrant division is based on whether the attribute exceeds the mean. Specifically:

[0184] Table 2. Matrix of Dual Importance-Representation Methods

[0185]

[0186] Figure 5 A preliminary version of the dual importance-performance matrix at the product level is shown, which can serve as an intermediate process for a preferred embodiment (described in further detail below).

[0187] Q1 quadrant: attributes in this quadrant perform well and are both pre-purchase and post-purchase importance. Attributes above the two-thirds line indicate that the pre-purchase and post-purchase importance of the attribute both exceed the mean.

[0188] Q2 quadrant: attributes in this quadrant perform poorly, but are both pre-purchase and post-purchase importance. That is, although consumers pay great attention to the attribute before purchase and it has a greater impact on satisfaction, the attribute fails to achieve the expected performance.

[0189] Q3 quadrant: attributes in this quadrant perform poorly and only one of the pre-purchase or post-purchase importance exceeds the mean. This indicates that the attribute, although it has a greater impact in one aspect, fails to meet the needs of users in overall performance.

[0190] Q4 quadrant: attributes in this quadrant perform poorly and both pre-purchase and post-purchase importance are low.

[0191] Q5 quadrant: attributes in this quadrant perform well, but both pre-purchase and post-purchase importance are low. These attributes, although they do well in actual performance, have less impact on consumer purchase decisions and satisfaction, and may not be fully exploited or disseminated.

[0192] Q6 quadrant: attributes in this quadrant perform well, but only one of the pre-purchase and post-purchase importance is high.

[0193] In the case of effective resources, merchants need to clarify the importance of demand to determine priorities, thereby helping to evaluate and optimize resource allocation. By calculating the performance of each attribute at the product level, and combining the double importance calculation formula, the final double importance-performance matrix is obtained. On this basis, the preferred embodiment combines the original Q1 and Q6 into one module, corresponding to the "continue to maintain" strategy, in order to better optimize resource allocation. Therefore, the preferred embodiment finally divides the features into five modules. Figure 6 The final version of the double importance-performance matrix at the product level of the preferred embodiment of the present application is shown.

[0194] Analysis of the characteristics of the five modules:

[0195] Q1 area: continue to maintain (excellent performance and high double importance)

[0196] These features have a significant impact on consumers' purchase decisions and satisfaction, both pre-purchase and post-purchase. These features are either highly focused on pre-purchase or have a significant impact on satisfaction post-purchase, or both. Regardless of the stage, their importance indicates their criticality to the overall experience. Even if they underperform in a stage, businesses should consider these features as "continue to maintain" priorities because they influence consumers' core needs and long-term loyalty. Businesses should ensure that these features' advantages are not lost, or it could significantly impact consumer satisfaction and even lead to customer churn.

[0197] For example: "Nutritional content" in plant milk is a highly focused core attribute for consumers and also receives significant positive feedback post-purchase. Businesses should ensure that the advantage of this attribute is maintained and avoid any decline in quality to maintain consumer loyalty.

[0198] Q2 Zone: Prioritize Improvement (Underperforming and High Dual Importance)

[0199] Attributes in this zone are critical to consumers' purchase decisions and satisfaction, indicating high importance both pre-purchase and post-purchase. However, these features do not meet consumers' expectations. Businesses should prioritize these attributes for improvement to enhance their performance and meet consumers' needs to avoid impacting overall satisfaction.

[0200] For example: "Taste" in plant milk may be a highly focused attribute for consumers pre-purchase, but if the taste does not meet expectations, businesses should prioritize improvement to enhance the overall consumer experience.

[0201] Q3 Zone: Low Priority (Underperforming and High Dual Importance)

[0202] Attributes in this zone have some impact on consumers, but their pre-purchase or post-purchase importance is only one high, and they underperform. These attributes have high importance but do not meet expectations. Therefore, compared to Q2, businesses can prioritize these attributes for optimization as secondary priorities. In the case of limited resources, businesses can slightly delay optimization of these attributes and prioritize addressing critical issues in the Q2 zone.

[0203] For example: "Packaging design" in plant milk may attract higher attention pre-purchase, but its actual effectiveness does not meet consumer expectations, and businesses can improve it if resources allow.

[0204] Q4 Zone: Irrelevant (Underperforming and Low Dual Importance)

[0205] This area has less impact on consumer's purchase decision and satisfaction, and the performance is also not good. Merchants should consider these features as low-priority attributes and can be ignored when resources are limited. Because these features are neither highly focused by consumers nor bring obvious benefits to consumers, merchants do not need to invest too much resources.

[0206] For example: The "product appearance" of plant milk may not be widely focused by consumers, and it has no significant impact on the product experience. Therefore, merchants can choose to temporarily optimize it.

[0207] Q5 area: appropriately reduce (good performance but low double importance)

[0208] The attributes in this area perform well and receive positive feedback from consumers in actual use. However, consumers pay less attention to these features, indicating that they have less impact on purchase decisions. Merchants can consider reducing further investment in these attributes, especially in the case of limited resources, focusing more resources on key features that have a greater impact.

[0209] For example: The "shelf life" of plant milk may perform well after purchase, but consumers pay less attention to this attribute before purchase. Merchants can appropriately reduce investment in this feature and optimize features that have a greater impact on purchase decisions.

[0210] Comparative analysis

[0211] Unlike traditional importance-performance analysis (IPA) methods, the present invention introduces a comprehensive pre-purchase and post-purchase perspective, proposing a double importance-performance analysis. Compared with existing research, the method of the present invention is more comprehensive and accurate, considering both consumer pre-purchase concerns and post-purchase satisfaction feedback, thus more accurately identifying key needs and optimization directions. Through the double importance matrix, we can not only evaluate the impact of each product feature on consumers, but also cross the limitations of a single dimension, comprehensively understanding the performance of the product at different stages.

[0212] Further, the preferred embodiment of the present invention divides product-level needs into five areas, helping merchants more scientifically identify which features should be prioritized for optimization, which should remain in the existing state, and which can reduce resource investment. This detailed division provides clear guidance for merchant decision-making, especially in the case of limited resources, helping merchants make more efficient optimization and resource allocation.

[0213] The following table compares the method of the present application with the existing method. The existing IPA method mainly relies on post-importance and performance to measure demand priority, while the present application provides more detailed and comprehensive demand analysis from both pre- and post-purchase dimensions and the specific performance of the product (pre-importance & post-importance & performance). In addition, the preferred embodiment of the present application innovatively proposes Figure 6 The visualization scheme in the present application displays three dimensions in a two-dimensional graph, which can more intuitively display more information. Compared with a three-dimensional scatter plot, this two-dimensional graph not only facilitates observation, but also helps merchants quickly understand and make decisions.

[0214] Table 3 Differences between the method of the present application and the existing method

[0215]

[0216] 5. Result presentation and report generation:

[0217] The system visually presents the analysis results to the merchant through visualization tools, including the double importance matrix at the market level and the double importance-performance matrix at the product level. These matrices not only help the merchant accurately identify key consumer needs, but also provide clear guidance for product optimization and decision-making, helping the merchant make more intelligent strategic choices in the fierce market competition.

[0218] The technical solution of the present application processes the user generated content (UGC) of the e-commerce platform through a large language model, and combines the double importance matrix and the double importance-performance matrix at the market level and the product level to fully reveal the behavior differences of consumers in purchase decision-making and use feedback. This innovative demand analysis method can provide more accurate market insights and product optimization suggestions for merchants, and has significant theoretical value and practical significance.

[0219] The important innovation work and technical points of the present application include:

[0220] 1. Demand analysis method based on large language model

[0221] The present application proposes an innovative demand analysis method for e-commerce based on a large language model. Through sentiment analysis and attribute extraction of user generated content (UGC) on e-commerce platforms, comprehensive demand analysis is conducted from two dimensions of pre-purchase user attention and post-purchase customer satisfaction, fully capturing changes in consumer demand.

[0222] 2. Generation of double importance matrix

[0223] The present invention proposes a new method to generate a dual importance matrix. By calculating the pre-purchase user attention and post-purchase customer satisfaction, this method helps merchants identify the relative importance of product attributes at the market level, providing data-based demand priority ranking for merchants.

[0224] 3. Construction of dual importance-performance matrix

[0225] The present invention further combines the actual performance of the product and proposes a dual importance-performance matrix. This matrix helps merchants analyze the relationship between the actual performance of each product attribute and its importance in the market, thereby providing scientific product optimization decision support for merchants.

[0226] 4. Demand analysis framework with a new perspective

[0227] The present invention conducts demand analysis based on the dual perspectives before and after purchase, solving the problem of traditional methods focusing only on post-purchase satisfaction. This framework not only focuses on the consumer's attention before making a purchase decision, but also analyzes the impact of product performance after actual use on consumer satisfaction, thereby providing a complete demand analysis process and helping merchants achieve more accurate product optimization.

[0228] Compared with existing e-commerce demand analysis technologies based on traditional data analysis methods, the present invention has significant advantages. Traditional technologies rely more on simple keyword extraction or sentiment analysis, usually only focusing on post-purchase user feedback, ignoring the consumer's attention points and decision-making process before purchase. However, through the application of large language models, the present invention combines the demand perspectives before and after purchase, comprehensively analyzes user attention and satisfaction, and can more accurately capture the real needs and emotional changes of consumers in the purchase decision-making process.

[0229] In addition, the present invention innovatively proposes the construction method of dual importance matrix and dual importance-performance matrix, combining the behavior characteristics of users before and after purchase, not only helping merchants identify the most critical demand characteristics in the market, but also analyzing the relationship between the actual performance of product attributes and their importance. This method provides more detailed demand priority ranking and can provide more accurate decision support for merchants in product optimization, resource allocation, and market promotion. Therefore, the present invention has higher accuracy and comprehensiveness than traditional methods, and can bring greater market competitiveness to merchants.

[0230] The present invention also provides a storage medium for storing a computer program, which is executed to perform at least the method described above.

[0231] The embodiment of the present application further provides a control device, comprising a processor and a storage medium for storing a computer program; wherein the processor is used to execute the computer program and at least execute the method as described above.

[0232] The embodiment of the present application further provides a processor, which executes a computer program and at least executes the method as described above.

[0233] The storage medium can be implemented by any type of nonvolatile storage device, or a combination thereof. The nonvolatile storage device can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a ferromagnetic random access memory (FRAM), a flash memory, a magnetic surface storage device, an optical disc, or a compact disc read-only memory (CD-ROM). The magnetic surface storage device can be a disk storage device or a tape storage device. The storage medium described in the embodiment of the present application is intended to include, but is not limited to, these and any other suitable types of storage devices.

[0234] In several embodiments provided by the present application, it should be understood that the disclosed system and method can be implemented in other manners. The described device embodiments are merely schematic, for example, the division of the units is only a logical function division, and there can be another division manner in actual implementation, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed coupling, or direct coupling or communication connection between the components can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or in other forms.

[0235] The units described as separated components can or can not be physically separated, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units; some or all of the units can be selected according to actual needs to achieve the purposes of the embodiments.

[0236] In addition, each function unit in each embodiment of the present application can be integrated in one processing unit, or each unit can be separately as a unit, or two or more units can be integrated in one unit; the integrated unit can be realized in the form of hardware or in the form of hardware plus software function unit.

[0237] Those skilled in the art can understand that all or part of the steps of the above-mentioned method embodiments can be completed by program instruction related hardware, and the foregoing program can be stored in a computer readable storage medium, and the program executes the steps of the above-mentioned method embodiments when executed; and the foregoing storage medium includes mobile storage equipment, read-only memory (ROM), random access memory (RAM), magnetic disc or optical disc and various storage program codes.

[0238] Alternatively, the integrated unit of the present application can be stored in a computer readable storage medium if it is realized in the form of a software function module and sold or used as an independent product. Based on this understanding, the technical solutions of the embodiments of the present application can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the embodiments of the present application. The foregoing storage medium includes mobile storage equipment, ROM, RAM, magnetic disc or optical disc and various storage program codes.

[0239] The methods disclosed in the several method embodiments of the present application can be combined arbitrarily without conflict to obtain new method embodiments.

[0240] The features disclosed in the several product embodiments of the present application can be combined arbitrarily without conflict to obtain new product embodiments.

[0241] The features disclosed in the several method or device embodiments of the present application can be combined arbitrarily without conflict to obtain new method or device embodiments.

[0242] The above is a further detailed description of the present application in combination with specific preferred embodiments, and the specific implementation of the present application cannot be limited to these descriptions. For those skilled in the art, without departing from the concept of the present application, a number of equivalent substitutions or obvious modifications can be made, and the performance or use is the same, which should be regarded as belonging to the protection scope of the present application.

Claims

1. A method for e-commerce demand analysis based on a large language model, characterized in that, Includes the following steps: S1. Data Collection: Collect user-generated content (UGC) of merchants' own products and target competitors from e-commerce platforms, including product review data and user Q&A data; S2. Data Processing and Analysis: Utilize large language models to process and analyze the collected user-generated content, extract product attributes and their emotional information, and transform the extracted information into structured data; S3. Market-level dual importance matrix generation: Based on the extracted structured data, calculate the pre-purchase attention and post-purchase importance of each product attribute, and combine the attention and importance to generate a market-level dual importance matrix to evaluate the relative importance of different product attributes in consumer purchase decisions and satisfaction. S4. Product-level dual importance-performance matrix generation: Combining the market-level dual importance matrix, calculate the actual performance value of each product attribute for the merchant's own products, and generate a product-level dual importance-performance matrix to identify product features that need to be optimized. S5. Results Presentation and Report Generation: The generated dual importance matrix and dual importance-performance matrix will be presented in a visual manner, and an analysis report will be generated to provide merchants with market insights and product optimization decision support.

2. The e-commerce demand analysis method based on a large language model according to claim 1, characterized in that, In step S2, when processing and analyzing user-generated content using a large language model, the specific steps include: A list of attributes for predefined product categories; For each comment, a pre-set prompt template is used to guide the large language model to identify the attributes involved in the comment and their corresponding sentiment polarity; The identified attributes and their emotional information are transformed into structured data.

3. The e-commerce demand analysis method based on a large language model according to claim 1, characterized in that, In step S3, calculating the pre-purchase attention level for each product attribute specifically includes the following steps: Analyze user question data in the user question and answer module and count the mention frequency of each attribute; The pre-purchase attention level for each attribute is calculated based on its mention frequency.

4. The e-commerce demand analysis method based on a large language model according to claim 1, characterized in that, In step S3, calculating the post-purchase importance of each product attribute specifically includes the following steps: Based on review data, a joint analysis method is used to calculate the impact of each attribute on customer satisfaction through multiple regression analysis modeling. The post-purchase importance of each attribute is determined based on the difference in regression coefficients.

5. The e-commerce demand analysis method based on a large language model according to claim 1, characterized in that, In step S3, generating the dual importance matrix at the market level specifically includes the following steps: Normalize the level of attention before purchase and the level of importance after purchase; Plot a double importance matrix with post-purchase importance on the horizontal axis and pre-purchase attention on the vertical axis; The matrix is ​​divided into four quadrants based on the mean, corresponding to core features, features of interest, low-value features, and potential optimization features, respectively.

6. The e-commerce demand analysis method based on a large language model according to claim 1, characterized in that, In step S4, calculating the actual performance value of each product attribute specifically includes the following steps: Based on the comment data, the percentage of positive sentiment for each attribute is calculated, which is used as the actual performance value of that attribute.

7. The e-commerce demand analysis method based on a large language model according to claim 1, characterized in that, In step S4, generating the dual importance-performance matrix at the product level specifically includes the following steps: The mean values ​​of the pre-purchase importance and post-purchase importance of each attribute are determined as the benchmark for evaluating the relative importance of the attributes; a conditional judgment mechanism is used to determine the weight adjustment factor of each attribute, which is dynamically adjusted based on the comparison results of the pre-purchase importance and post-purchase importance of the attribute with their respective mean values; based on the weight adjustment factor, pre-purchase importance, and post-purchase importance of each attribute, a weighted calculation is performed to obtain the dual importance of each attribute. Using the dual importance of attributes as the vertical axis and actual performance values ​​as the horizontal axis, a dual importance-performance matrix is ​​plotted at the product level. This matrix is ​​divided into five modules, each with the following characteristics: The attributes contained in the first module have high dual importance and high actual performance values, indicating that these attributes have a significant impact on consumer decisions and perform well, and should be maintained. The attributes contained in the second module have high dual importance but low actual performance values, indicating that these attributes have a significant impact on consumer decisions but perform poorly and need to be improved first. The attributes included in the third module have high dual importance but low actual performance values, indicating that these attributes have some impact on consumers but their performance has not met expectations, and can be kept as low priority for optimization. The attributes contained in the fourth module have low dual importance and low actual performance values, indicating that these attributes have little impact on consumer decisions and perform only moderately, and can be kept irrelevant. The attributes included in the fifth module have low dual importance but high actual performance values, indicating that although these attributes have little impact on consumer decisions, they perform well and investment can be appropriately reduced.

8. The e-commerce demand analysis method based on a large language model according to claim 7, characterized in that, In step S4, the dual importance of each attribute is calculated by weighting the pre-purchase attention, post-purchase importance, and a weight adjustment factor adjusted based on conditional judgment. The conditional judgment mechanism is as follows: when both the pre-purchase importance and post-purchase importance of an attribute are higher than their average, a higher weight adjustment factor is assigned; when either importance is higher than its average, a medium weight adjustment factor is assigned; and when both are lower than their average, a lower weight adjustment factor is assigned.

9. The e-commerce demand analysis method based on a large language model according to claim 1, characterized in that, In step S5, the presentation of results and generation of reports specifically include the following steps: Visualization tools are used to display the dual importance matrix at the market level and the dual importance-performance matrix at the product level; Generate analytical reports to provide market insights and support for product optimization decisions.

10. An e-commerce demand analysis system based on a large language model, characterized in that, Includes the following steps: Data collection module: Collects user-generated content (UGC) of merchants' own products and target competitors from e-commerce platforms, including product review data and user Q&A data; Data processing and analysis module: Utilizes large language models to process and analyze the collected user-generated content, extracting product attributes and their emotional information, and transforming the extracted information into structured data; Market-level dual importance matrix generation module: Based on the extracted structured data, it calculates the pre-purchase attention and post-purchase importance of each product attribute, and combines the attention and importance to generate a market-level dual importance matrix, which is used to evaluate the relative importance of different product attributes in consumer purchase decisions and satisfaction. Product-level dual importance-performance matrix generation module: Combining the market-level dual importance matrix, it calculates the actual performance value of each product attribute for the merchant's own products and generates a product-level dual importance-performance matrix to identify product features that need to be optimized. Results Presentation and Report Generation Module: The generated dual importance matrix and dual importance-performance matrix are presented in a visual manner, and analysis reports are generated to provide merchants with market insights and product optimization decision support.

Images