E-commerce analysis and diagnosis method based on artificial intelligence

Abstract

The invention particularly relates to an e-commerce analysis and diagnosis method based on artificial intelligence. The method comprises the following steps: collecting behavior data; analyzing behaviors; comprehensively judging; and intention diagnosis evaluation. According to the method, a multi-dimensional analysis system is constructed, and a foundation is laid for accurate analysis by comprehensively collecting behavior data of a user in a whole link from contacting commodities to browsing details, comments and the like; two core indexes of a matching coefficient and an interest coefficient are introduced, and the matching coefficient quantifies the matching degree of a user and commodity information from the angles of browsing fluency, information acquisition efficiency, key information attention and the like; the latter excavates potential interests of the user by combining three dimensions of comment interaction, historical portrait matching and picture browsing; and finally, intention evaluation coefficients are calculated and graded by integrating purchase and collection behaviors, so that the purchase intention of the user can be more accurately judged, the waste of marketing resources is avoided, the marketing activity conversion rate and the input-output ratio are improved, and a scientific basis is provided for accurate marketing of e-commerce.

Description

Technical Field

[0001] The present invention relates to the field of e-commerce technology, and in particular to an e-commerce analysis and diagnosis method based on artificial intelligence. Background Art

[0002] In the field of e-commerce, with the explosive growth of user behavior data, traditional manual analysis or simple statistical methods can no longer meet the needs of real-time and precise operations. Its pain points include inefficient data processing, insufficient depth of user insights (only staying on surface indicators such as click-through rate and add-to-cart rate, lack of exploration of deep behavioral patterns such as browsing rhythm, interest preferences, emotional tendencies, etc.), vague marketing target positioning and poor content adaptability, fragmented technical tools, and rising user retention costs due to intensified industry competition.

[0003] From the perspective of user insights, existing technologies lack the ability to deeply deconstruct user behavior. Traditional analysis focuses only on superficial indicators such as click-through rate and conversion rate, but ignores the subtle behavioral characteristics of users during browsing:

[0004] Browsing rhythm: The length of time users pause on a product detail page indicates how deeply they digest the information (e.g., a long pause may indicate a careful reading of specifications).

[0005] Focus of interest: The number of times users zoom in on specific areas of product images and the frequency of keyword hits in the comment section directly reflect their focus (e.g., users who zoom in on fabric details multiple times are more interested in the material).

[0006] Sentiment: The user's emotional polarity (positive / negative) and interactive behavior (likes, replies) towards the review content reveal their underlying attitude towards the product (e.g., frequently liking negative reviews may indicate interest in competing products).

[0007] The limitations of existing technologies are mainly reflected in: inefficient data processing, shallow user insights, extensive marketing matching, fragmented system architecture, and lagging model iteration;

[0008] Therefore, there is an urgent need for an e-commerce analysis and diagnosis method based on artificial intelligence to address the above-mentioned problems. Summary of the Invention

[0009] The purpose of the present invention is to propose an e-commerce analysis and diagnosis method based on artificial intelligence in order to solve the above problems.

[0010] In order to achieve the above object, the present invention adopts the following technical solutions:

[0011] An e-commerce analysis and diagnosis method based on artificial intelligence, comprising:

[0012] Behavioral data collection: Obtaining user interaction behavior data corresponding to marketing information displayed on the user's mobile phone;

[0013] Behavior analysis: Analyze the user's interactive behavior information data to obtain the matching coefficient and interest coefficient;

[0014] Comprehensive judgment: The matching coefficient and interest coefficient are weighted and calculated to obtain the basic evaluation coefficient. The basic evaluation coefficient is then combined with the coefficients corresponding to the add to cart and favorite actions to obtain the intention evaluation coefficient.

[0015] Intention diagnosis assessment: Determine the user's purchase intention for the marketing product based on the intention assessment coefficient.

[0016] Preferably, the behavioral data collection specifically includes:

[0017] When the shopping app on the user's mobile phone receives a product marketing image display, it obtains the corresponding user interaction behavior information data, where the user interaction behavior information data includes:

[0018] Data information on clicking on product marketing images, data information on interaction time after entering the product page, data information on adding to cart and adding to favorites, and data information on viewing product details and comments.

[0019] Preferably, the process of obtaining the matching coefficient includes:

[0020] Obtain the pause time between each page slide when the user browses the product details page, and preset a pause time threshold for page slides. Compare the pause time between each page slide of the user with the pause time threshold for page slides. Record the pause time between page slides that is less than the pause time threshold for page slides as normal pause time, and record the pause time between page slides that is greater than the pause time threshold for page slides as abnormal pause time.

[0021] After counting the number of all normal pauses and the number of pauses between all page slides, divide the number of all normal pauses by the number of pauses between all page slides to get the normal browsing ratio;

[0022] All normal pause times are accumulated to obtain the total normal pause time;

[0023] Accumulate all abnormal pause times to get the total abnormal pause time

[0024] Obtain the time when the user clicks on the product and the time when the user exits the product browsing process, and record the difference between the two times as the total browsing time;

[0025] Divide the total normal pause time by the total browsing time to get the effective time ratio;

[0026] Subtract the total abnormal pause time from the total browsing time to get the effective browsing time;

[0027] Determine all keywords for the product, including specifications, after-sales policies, and related recommendations;

[0028] Obtain the time when a keyword appears and disappears on the phone screen when the user is browsing product details; subtract the time when the keyword appears from the time when the keyword disappears to obtain the duration of the keyword appearance;

[0029] Obtain the appearance time of each keyword in turn, and accumulate them to obtain the total appearance time of the keyword; divide the total appearance time of the keyword by the effective browsing time to obtain the keyword ratio;

[0030] The matching coefficient is obtained by weighting the normal browsing ratio, effective time ratio, and keyword ratio.

[0031] Preferably, the process of obtaining the interest coefficient includes:

[0032] After analyzing the interactive data information in the comment area, the comment interaction value is obtained;

[0033] The profile matching value is obtained by analyzing the data information of related similar products browsed by users within a preset time period;

[0034] Image values ​​are obtained by analyzing users' browsing operations on the details page and comment area of ​​marketing products.

[0035] Preferably, the process of obtaining the comment interaction value includes:

[0036] Obtain the time when the user opens the comment area and the time when the user exits the comment area, and record the time difference between the two times as the comment stay time;

[0037] Get the number of comments clicked by users and the corresponding interactive behaviors for each comment, including clicks on "View more comments", "Reply", and "Like";

[0038] Accumulate the interactive behaviors corresponding to each comment, get the total number of interactions, and divide it by the comment dwell time to get the interaction rate;

[0039] Through natural language processing technology, the comments viewed by users are matched with the preset keyword library and the number of hits is counted;

[0040] Match the content of each comment viewed by the user with the preset keyword library in turn to obtain the number of keyword hits corresponding to each comment;

[0041] The sentiment analysis model in natural language processing is used to evaluate the sentiment of the comments viewed by users. The score range is -1 to 1, where -1 is negative and 1 is positive.

[0042] Arrange the number of hits for each keyword in descending order according to the numerical value, and extract the three keywords with the largest number of hits;

[0043] Multiply the three largest keyword hit times by the corresponding comment sentiment score to get the sentiment score of the three largest keyword hit times;

[0044] The three obtained sentiment scores are averaged. If the obtained value is greater than 0, it is marked as the average sentiment score;

[0045] The comment interaction value is obtained by comprehensively calculating the interaction rate and the sentiment score.

[0046] Preferably, the process of obtaining the portrait matching value includes:

[0047] Obtain the user's searches or favorites for similar products and their corresponding purchase prices within a preset historical time period;

[0048] Obtain the details page of each similar product purchased in turn, extract the corresponding product keywords from them, and unify the keywords of all similar products purchased to obtain a keyword library of similar products;

[0049] Match each keyword of the marketing product with the keyword library of similar products, count the number of successful matches, and record it as the success value;

[0050] Divide the total number of keywords for the marketing product by the total number of keywords in the keyword library for similar products to get the overlap degree;

[0051] Calculate the average price of all similar products by averaging the prices;

[0052] Calculate the difference between the price of the marketing product and the average price of similar products to obtain the price difference; preset the allowable floating range of the price difference, match the price difference with the allowable floating range of the price difference, and if the price difference is within the allowable floating range of the price difference, record it as a reasonable price difference;

[0053] The portrait matching value is obtained by comprehensively processing the success value, overlap degree and reasonable price difference.

[0054] Preferably, the process of acquiring the image value includes:

[0055] Monitor changes in the properties of the image's parent container. If the size exceeds a preset threshold, it is considered an image enlargement operation.

[0056] Obtain the marketing product area corresponding to the image that the user zoomed in on, and extract the zoom dwell time of the zoomed-in area;

[0057] Divide the overall appearance of the marketing product into multiple areas, and distinguish each area based on its own characteristics;

[0058] Obtain the image displayed on the user's mobile phone screen, extract the corresponding marketing product features from it, and identify the marketing product area corresponding to the image;

[0059] Obtain the marketing product area corresponding to each time the user zooms in on the image, and count the number of times each area of ​​the marketing product is zoomed in; extract the two largest zooming times and the image zooming dwell time corresponding to the two largest zooming times;

[0060] Multiply the two largest amplification times by their corresponding amplification dwell times to obtain the quantitative value;

[0061] The two quantized values ​​are respectively used as the major semi-axis and the minor semi-axis of the ellipse to construct an ellipse model, and the area of ​​the ellipse model is recorded as the image value.

[0062] Preferably, the weighted calculation of the matching coefficient and the interest coefficient to obtain the basic evaluation coefficient includes:

[0063] After normalizing the matching coefficient and the interest coefficient, a circle is constructed using the matching coefficient as the radius of the circle and the interest coefficient as the height of the circle. A cone model is established and the volume of the cone model is calculated to obtain the basic evaluation coefficient.

[0064] If the user adds the promotional product to the shopping cart, the action value of the add-to-cart action is recorded as 1, otherwise it is -1;

[0065] Similarly, if the user adds the marketing product to the collection, the action value of the collection action is recorded as 1, otherwise it is -1;

[0066] Preset weight factors corresponding to the add-to-cart and favorite actions, and multiply the action value of the add-to-cart action and the action value of the favorite action with their corresponding weight factors, and then sum them to obtain the add-to-cart and favorite coefficient;

[0067] After normalizing the basic evaluation coefficient and the purchase and collection coefficient, determine whether the purchase and collection coefficient is greater than 0:

[0068] If the purchase and collection coefficient is greater than 0, the basic evaluation coefficient and the purchase and collection coefficient are used as the two right-angled sides of a right triangle, and the remaining side is connected to form a complete right triangle. The area of ​​the right triangle is recorded as the intention evaluation coefficient;

[0069] If the purchase and collection coefficient is less than 0, the basic evaluation coefficient and the purchase and collection coefficient will be used as the two right-angled sides of the right triangle respectively. After connecting the remaining side to form a complete right-angled triangle, the center of the right-angled side corresponding to the collection coefficient will be used as the origin, and a circle will be established with half of the collection coefficient as the radius. The circle will cut the right triangle and calculate the area of ​​the remaining right triangle, which will be recorded as the intention evaluation coefficient.

[0070] Preferably, determining the user's purchase intention for the marketing product based on the intention evaluation coefficient specifically includes:

[0071] Three groups of threshold value ranges are preset, and the value range of each group of threshold values ​​corresponds to a purchase intention level. The intention evaluation coefficient is compared with the value range of the three groups of threshold values ​​to obtain the purchase intention level corresponding to the intention evaluation coefficient. The purchase intention level levels include low intention, medium intention and high intention.

[0072] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0073] 1. The present invention lays the foundation for accurate analysis by constructing a multi-dimensional analysis system and comprehensively collecting behavioral data on users from all aspects, from contact with products to browsing details and comments. It introduces two core indicators, the matching coefficient and the interest coefficient. The former quantifies the degree of matching between users and product information from the perspectives of browsing fluency, information acquisition efficiency, and attention to key information. The latter combines comment interaction, historical portrait matching, and image browsing to explore users' potential interests. Finally, the intention evaluation coefficient is calculated based on the purchase and collection behaviors and the grades are divided. This can more accurately judge users' purchasing intentions, avoid wasting marketing resources, improve the conversion rate and input-output ratio of marketing activities, and provide a scientific basis for e-commerce precision marketing.

[0074] 2. The present invention collects user behavior data in real time by utilizing SDK and embedded point technology, and timely reflects changes in user behavior and market feedback based on dynamically calculated matching coefficients, interest coefficients, and intention evaluation coefficients. Once user behavior data is updated, the system immediately recalculates the correlation coefficient and updates the purchase intention evaluation results. E-commerce companies can quickly adjust their marketing strategies accordingly, such as optimizing product information display and improving the design of comment areas for low-intention groups; increasing resource investment in high-intention users and pushing personalized promotions; in addition, it can monitor the differences in effects across different marketing channels and time periods and optimize resource allocation. Through dynamic evaluation and real-time strategy adjustment, companies can better adapt to market changes, enhance user stickiness, and improve market competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Further details, features and advantages of the present application are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which:

[0076] Figure 1 is a flow chart of the present invention; DETAILED DESCRIPTION

[0077] Several embodiments of the present application will be described in more detail below with reference to the accompanying drawings so that those skilled in the art can implement the present application. The present application can be embodied in many different forms and for many different purposes and should not be limited to the embodiments described herein. These embodiments are provided to make the present application comprehensive and complete and to fully convey the scope of the present application to those skilled in the art. The embodiments do not limit the present application.

[0078] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. It will be further understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the relevant art and / or the context of this specification, and will not be interpreted in an idealized or overly formal sense unless explicitly defined as such herein.

[0079] See also Figure 1 As shown, the present invention provides a technical solution:

[0080] An e-commerce analysis and diagnosis method based on artificial intelligence, comprising:

[0081] Behavioral data collection: Obtaining user interaction behavior data corresponding to marketing information displayed on the user's mobile phone;

[0082] Specifically include:

[0083] When the shopping app on the user's mobile phone receives a product marketing image display, it obtains the corresponding user interaction behavior information data, where the user interaction behavior information data includes:

[0084] Data on clicks on product marketing images, interaction time after entering the product page, add-to-cart and favorites data, and product details and review viewing data;

[0085] Data on clicks on product marketing images: When a product marketing image is displayed on a user's mobile shopping app, the software development kit (SDK) or JavaScript tracking technology is used to accurately capture the time, coordinates, device model, and other data related to the user's click on the product marketing image.

[0086] Interaction time data after entering the product page: After a user enters a product page, a front-end timer (such as setInterval) is used to continuously record the time difference between entering the page and leaving the page as interaction time information. This information is also combined with page switching events (such as visibilitychange) to handle situations such as users switching out of the app.

[0087] Add to cart and favorites data: Bind event listeners to the "Add to cart" and "Favorites" buttons in the shopping app. When the user triggers the action, immediately obtain information such as the product ID, operation time, and user account;

[0088] Product details and review viewing data: Embed tracking code or integrate SDK into the front-end page of the shopping software to monitor user behavior in real time: On the product details page, use scroll events (such as scroll) to calculate the user's browsing depth (page scroll percentage), record the length of stay through lifecycle hooks (such as onLoad / onUnload), and bind interactive events (such as play / click) to elements such as videos and parameter switching to obtain content interaction data; set a loading monitoring event in the product buyer review area. When the page loads the review content, record the trigger time when the user views the review, the number of reviews viewed, whether all reviews are expanded, and other information;

[0089] Behavior analysis: Analyze the user's interactive behavior information data to obtain the matching coefficient and interest coefficient;

[0090] The process of obtaining the matching coefficient includes:

[0091] Obtain the pause time between each page slide when the user browses the product details page, and preset a pause time threshold for page slides. Compare the pause time between each page slide of the user with the pause time threshold for page slides. Record the pause time between page slides that is less than the pause time threshold for page slides as normal pause time, and record the pause time between page slides that is greater than the pause time threshold for page slides as abnormal pause time.

[0092] After counting the number of all normal pauses and the number of pauses between all page slides, divide the number of all normal pauses by the number of pauses between all page slides to get the normal browsing ratio;

[0093] The normal browsing ratio reflects the proportion of pauses that conform to the user's normal rhythm during browsing. The higher the normal browsing ratio, the smoother and more natural the user's browsing behavior. For example, if the total number of pauses is 10 and the number of normal pauses is 8, the normal browsing ratio is 0.8.

[0094] All normal pause times are accumulated to obtain the total normal pause time;

[0095] Accumulate all abnormal pause times to get the total abnormal pause time

[0096] Obtain the time when the user clicks on the product and the time when the user exits the product browsing process, and record the difference between the two times as the total browsing time;

[0097] Divide the total normal pause time by the total browsing time to get the effective time ratio;

[0098] Subtract the total abnormal pause time from the total browsing time to get the effective browsing time;

[0099] Determine all keywords for the product, including specifications, after-sales policies, and related recommendations;

[0100] Obtain the time when a keyword appears and disappears on the phone screen when the user is browsing product details; subtract the time when the keyword appears from the time when the keyword disappears to obtain the duration of the keyword appearance;

[0101] Obtain the appearance time of each keyword in turn, and accumulate them to obtain the total appearance time of the keyword; divide the total appearance time of the keyword by the effective browsing time to obtain the keyword ratio;

[0102] The keyword ratio reflects the user's attention to the key information of the product. The higher the keyword ratio, the more the user pays attention to the core information of the product.

[0103] The matching coefficient is obtained by weighting the normal browsing ratio, effective time ratio, and keyword ratio;

[0104] After presetting the weight factors of normal browsing ratio, effective time ratio, and keyword ratio, the normal browsing ratio, effective time ratio, and keyword ratio are multiplied by their corresponding weight factors to obtain the matching coefficient;

[0105] The process of obtaining the interest coefficient includes:

[0106] After analyzing the interactive data information in the comment area, the comment interaction value is obtained;

[0107] The profile matching value is obtained by analyzing the data information of related similar products browsed by users within a preset time period;

[0108] Image values ​​are obtained by analyzing users' browsing operations on the details page and comment area of ​​marketing products;

[0109] The interest coefficient is obtained by weighting the comment interaction value, portrait matching value, and image value; the weight factors of the interaction value, portrait matching value, and image value are preset, and the interaction value, portrait matching value, image value, and their corresponding weight factors are multiplied and the sum is used to obtain the interest coefficient;

[0110] The process of obtaining comment interaction value includes:

[0111] Obtain the time when the user opens the comment area and the time when the user exits the comment area, and record the time difference between the two times as the comment stay time;

[0112] Get the number of comments clicked by users and the corresponding interactive behaviors for each comment, including clicks on "View more comments", "Reply", and "Like";

[0113] Accumulate the interactive behaviors corresponding to each comment, get the total number of interactions, and divide it by the comment dwell time to get the interaction rate;

[0114] The interaction rate reflects the frequency of user interaction with the comment section content within a unit of time. The higher the interaction rate, the more engaged the user is in the comment section and the more likely they are to be interested in or concerned about the product.

[0115] Using natural language processing technology, we match user reviews with a pre-set keyword library (e.g., words related to a product's core selling points and common concerns) and count the number of hits.

[0116] Assume that there are words such as "durable" and "battery life" in the keyword library, and "durable" appears in the comments viewed by the user, the hit count is 1;

[0117] Match the content of each comment viewed by the user with the preset keyword library in turn to obtain the number of keyword hits corresponding to each comment;

[0118] Use sentiment analysis models in natural language processing (such as the BERT-based sentiment classification model) to assign sentiment scores to comments viewed by users, with a score range of -1 to 1, where -1 represents negative and 1 represents positive.

[0119] Arrange the number of hits for each keyword in descending order according to the numerical value, and extract the three keywords with the largest number of hits;

[0120] Multiply the three largest keyword hit times by the corresponding comment sentiment score to get the sentiment score of the three largest keyword hit times;

[0121] The three obtained sentiment scores are averaged. If the obtained value is greater than 0, it is marked as the average sentiment score;

[0122] A sentiment score greater than 0 indicates that the user's overall sentiment attitude towards the key information in the review is positive;

[0123] The interaction rate and sentiment score are calculated together to get the comment interaction value;

[0124] The interaction rate and emotional score are recorded as SI and EM respectively and then inserted into the formula: Get the comment interaction value ζ; where EM′ is the reference sentiment average; a1 and a2 are the weight factors corresponding to the interaction rate and sentiment average respectively;

[0125] The process of obtaining the portrait matching value includes:

[0126] Obtain the user's searches or favorites for similar products and their corresponding purchase prices within a preset historical time period;

[0127] In a preset historical time range (such as the past 30 days or 90 days), obtain the information of similar products that users searched, collected, or purchased, and their corresponding purchase prices. This step uses the user behavior database of the e-commerce platform to filter out similar products based on product classification tags (such as "laptops" and "sports shoes"). For example, a user purchased three laptops of different brands in the past 30 days, and the names, purchase prices, and other information of these three products are recorded. This data provides a basis for subsequent analysis and reflects the user's historical consumption preferences;

[0128] Obtain the details page of each similar product purchased in turn, extract the corresponding product keywords from them, and unify the keywords of all similar products purchased to obtain a keyword library of similar products;

[0129] The system sequentially visits the detail pages of similar products purchased by the user and uses natural language processing techniques (such as text extraction and keyword extraction algorithms) to extract keywords from fields such as product titles, descriptions, and specifications. For example, it extracts keywords such as "Core i7," "16GB RAM," and "512GB SSD" from the detail page of a laptop computer. Keywords extracted from all similar products are then deduplicated and merged to form a keyword library for similar products. This keyword library represents the core characteristics of the user's historically purchased products.

[0130] Match each keyword of the marketing product with the keyword library of similar products, count the number of successful matches, and record it as the success value;

[0131] Divide the total number of keywords for the marketing product by the total number of keywords in the keyword library for similar products to get the overlap degree;

[0132] Divide the total number of keywords for the marketing product by the total number of keywords in the keyword library for similar products to get the overlap. Assuming the marketing product has 5 keywords and the keyword library for similar products has 20 keywords, the overlap is 0.25. The overlap reflects the overall correlation between the marketing product keywords and the user's historical preference keywords.

[0133] Calculate the average price of all similar products by averaging the prices;

[0134] Calculate the difference between the price of the marketing product and the average price of similar products to obtain the price difference; preset the allowable floating range of the price difference, match the price difference with the allowable floating range of the price difference, and if the price difference is within the allowable floating range of the price difference, record it as a reasonable price difference;

[0135] The portrait matching value is obtained by comprehensively processing the success value, overlap degree, and reasonable price difference;

[0136] Preset the weight factors of success value, overlap, and reasonable price difference, and multiply the success value, overlap, and reasonable price difference with their corresponding weight factors to obtain the portrait matching value;

[0137] The process of obtaining image values ​​includes:

[0138] Monitor changes in the width / height attributes of the image's parent container. If the size exceeds a preset threshold (e.g., more than 50% of the screen's visible area), it is considered an image enlargement operation.

[0139] Obtain the marketing product area corresponding to the image that the user zoomed in on, and extract the zoom dwell time of the zoomed-in area;

[0140] The moment when the image displayed on the user's mobile phone screen is enlarged is recorded as the first time, and the time when the image displayed on the screen changes is recorded as the second time. The time difference between the first time and the second time is calculated to obtain the enlargement dwell time; the unit of the enlargement dwell time is seconds;

[0141] Divide the overall appearance of the marketing product into multiple areas, and distinguish each area based on its own characteristics;

[0142] Obtain the image displayed on the user's mobile phone screen, extract the corresponding marketing product features from it, and identify the marketing product area corresponding to the image;

[0143] Obtain the marketing product area corresponding to each time the user zooms in on the image, and count the number of times each area of ​​the marketing product is zoomed in; extract the two largest zooming times and the image zooming dwell time corresponding to the two largest zooming times;

[0144] Multiply the two largest amplification times by their corresponding amplification dwell times to obtain the quantitative value;

[0145] The two quantized values ​​are used as the major and minor axes of the ellipse respectively to construct an ellipse model, and the area of ​​the ellipse model is recorded as the image value;

[0146] Comprehensive judgment: The matching coefficient and interest coefficient are weighted and calculated to obtain the basic evaluation coefficient. The basic evaluation coefficient is then combined with the coefficients corresponding to the add to cart and favorite actions to obtain the intention evaluation coefficient.

[0147] After normalizing the matching coefficient and the interest coefficient, a circle is constructed using the matching coefficient as the radius of the circle and the interest coefficient as the height of the circle. A cone model is established and the volume of the cone model is calculated to obtain the basic evaluation coefficient.

[0148] If the user adds the promotional product to the shopping cart, the action value of the add-to-cart action is recorded as 1, otherwise it is -1;

[0149] Similarly, if the user adds the marketing product to the collection, the action value of the collection action is recorded as 1, otherwise it is -1;

[0150] Preset weight factors corresponding to the add-to-cart and favorite actions, and multiply the action value of the add-to-cart action and the action value of the favorite action with their corresponding weight factors, and then sum them to obtain the add-to-cart and favorite coefficient;

[0151] After normalizing the basic evaluation coefficient and the purchase and collection coefficient, determine whether the purchase and collection coefficient is greater than 0:

[0152] If the purchase and collection coefficient is greater than 0, the basic evaluation coefficient and the purchase and collection coefficient are used as the two right-angled sides of a right triangle, and the remaining side is connected to form a complete right triangle. The area of ​​the right triangle is recorded as the intention evaluation coefficient;

[0153] If the purchase and collection coefficient is less than 0, the basic evaluation coefficient and the purchase and collection coefficient are used as the two right-angled sides of a right triangle respectively. After connecting the remaining side to form a complete right triangle, a circle is created with the center of the right-angled side corresponding to the collection coefficient as the origin and half of the collection coefficient as the radius. The area of ​​the remaining right triangle is calculated and recorded as the intention evaluation coefficient.

[0154] Intention diagnosis assessment: Determine the user's purchase intention for the marketing product based on the intention assessment coefficient;

[0155] Three groups of threshold value ranges are preset, and the value range of each group of threshold values ​​corresponds to a purchase intention level. The intention evaluation coefficient is compared with the value range of the three groups of threshold values ​​to obtain the purchase intention level corresponding to the intention evaluation coefficient. The purchase intention level levels include low intention, medium intention and high intention.

[0156] The above formulas are obtained by collecting a large amount of data and performing software simulation, and a formula close to the actual value is selected. The influencing weight factors and specific coefficient values ​​in the formula are set by technical personnel in this field according to actual conditions, and can be adjusted and modified later.

[0157] The above description of the embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An e-commerce analysis and diagnosis method based on artificial intelligence, characterized in that: include: Behavioral data collection: Obtaining user interaction behavior data corresponding to marketing information displayed on the user's mobile phone; Behavior analysis: Analyze the user's interactive behavior information data to obtain the matching coefficient and interest coefficient; Comprehensive judgment: The matching coefficient and interest coefficient are weighted and calculated to obtain the basic evaluation coefficient. The basic evaluation coefficient is then combined with the coefficients corresponding to the add to cart and favorite actions to obtain the intention evaluation coefficient. Intention diagnosis assessment: Determine the user's purchase intention for the marketing product based on the intention assessment coefficient.

2. The e-commerce analysis and diagnosis method based on artificial intelligence according to claim 1, characterized in that: Behavioral data collection, including: When the shopping app on the user's mobile phone receives a product marketing image display, it obtains the corresponding user interaction behavior information data, where the user interaction behavior information data includes: Data information on clicking on product marketing images, data information on interaction time after entering the product page, data information on adding to cart and adding to favorites, and data information on viewing product details and comments.

3. The e-commerce analysis and diagnosis method based on artificial intelligence according to claim 2, characterized in that: The process of obtaining the matching coefficient includes: Obtain the pause time between each page slide when the user browses the product details page, and preset a pause time threshold for page slides. Compare the pause time between each page slide of the user with the pause time threshold for page slides. Record the pause time between page slides that is less than the pause time threshold for page slides as normal pause time, and record the pause time between page slides that is greater than the pause time threshold for page slides as abnormal pause time. After counting the number of all normal pauses and the number of pauses between all page slides, divide the number of all normal pauses by the number of pauses between all page slides to get the normal browsing ratio; All normal pause times are accumulated to obtain the total normal pause time; Accumulate all abnormal pause times to get the total abnormal pause time Obtain the time when the user clicks on the product and the time when the user exits the product browsing process, and record the difference between the two times as the total browsing time; Divide the total normal pause time by the total browsing time to get the effective time ratio; Subtract the total abnormal pause time from the total browsing time to get the effective browsing time; Determine all keywords for the product, including specifications, after-sales policies, and related recommendations; Obtain the time when a keyword appears and disappears on the phone screen when the user is browsing product details; subtract the time when the keyword appears from the time when the keyword disappears to obtain the duration of the keyword appearance; Obtain the appearance time of each keyword in turn, and accumulate them to obtain the total appearance time of the keyword; divide the total appearance time of the keyword by the effective browsing time to obtain the keyword ratio; The matching coefficient is obtained by weighting the normal browsing ratio, effective time ratio, and keyword ratio.

4. The e-commerce analysis and diagnosis method based on artificial intelligence according to claim 3, characterized in that: The process of obtaining the interest coefficient includes: After analyzing the interactive data information in the comment area, the comment interaction value is obtained; The profile matching value is obtained by analyzing the data information of related similar products browsed by users within a preset time period; Image values ​​are obtained by analyzing users' browsing operations on the details page and comment area of ​​marketing products; The interest coefficient is obtained by weighting the comment interaction value, portrait matching value, and image value.

5. The e-commerce analysis and diagnosis method based on artificial intelligence according to claim 4, characterized in that: The process of obtaining comment interaction value includes: Obtain the time when the user opens the comment area and the time when the user exits the comment area, and record the time difference between the two times as the comment stay time; Obtain the number of comments clicked by users, as well as the corresponding interactive behaviors for each comment, including clicks on "View More Comments," "Reply," and "Like"; Accumulate the interactive behaviors corresponding to each comment, get the total number of interactions, and divide it by the comment dwell time to get the interaction rate; Through natural language processing technology, the comments viewed by users are matched with the preset keyword library and the number of hits is counted; Match the content of each comment viewed by the user with the preset keyword library in turn to obtain the number of keyword hits corresponding to each comment; The sentiment analysis model in natural language processing is used to evaluate the sentiment of the comments viewed by users. The score range is -1 to 1, where -1 is negative and 1 is positive. Arrange the number of hits for each keyword in descending order according to the numerical value, and extract the three keywords with the largest number of hits; Multiply the three largest keyword hit times by the corresponding comment sentiment score to get the sentiment score of the three largest keyword hit times; The three obtained sentiment scores are averaged. If the obtained value is greater than 0, it is marked as the average sentiment score; The comment interaction value is obtained by comprehensively calculating the interaction rate and the sentiment score.

6. The e-commerce analysis and diagnosis method based on artificial intelligence according to claim 5, characterized in that: The process of obtaining the portrait matching value includes: Obtain the user's searches or favorites for similar products and their corresponding purchase prices within a preset historical time period; Obtain the details page of each similar product purchased in turn, extract the corresponding product keywords from them, and unify the keywords of all similar products purchased to obtain a keyword library of similar products; Match each keyword of the marketing product with the keyword library of similar products, count the number of successful matches, and record it as the success value; Divide the total number of keywords for the marketing product by the total number of keywords in the keyword library for similar products to get the overlap degree; Calculate the average price of all similar products by averaging the prices; Calculate the difference between the price of the marketing product and the average price of similar products to obtain the price difference; preset the allowable floating range of the price difference, match the price difference with the allowable floating range of the price difference, and if the price difference is within the allowable floating range of the price difference, record it as a reasonable price difference; The portrait matching value is obtained by comprehensively processing the success value, overlap degree and reasonable price difference.

7. The e-commerce analysis and diagnosis method based on artificial intelligence according to claim 6, characterized in that: The process of obtaining image values ​​includes: Monitor changes in the properties of the image's parent container. If the size exceeds a preset threshold, it is considered an image enlargement operation. Obtain the marketing product area corresponding to the image that the user zoomed in on, and extract the zoom dwell time of the zoomed-in area; Divide the overall appearance of the marketing product into multiple areas, and distinguish each area based on its own characteristics; Obtain the image displayed on the user's mobile phone screen, extract the corresponding marketing product features from it, and identify the marketing product area corresponding to the image; Obtain the marketing product area corresponding to each time the user zooms in on the image, and count the number of times each area of ​​the marketing product is zoomed in; extract the two largest zooming times and the image zooming dwell time corresponding to the two largest zooming times; Multiply the two largest amplification times by their corresponding amplification dwell time to obtain the quantitative value; The two quantized values ​​are respectively used as the major semi-axis and the minor semi-axis of the ellipse to construct an ellipse model, and the area of ​​the ellipse model is recorded as the image value.

8. The e-commerce analysis and diagnosis method based on artificial intelligence according to claim 7, characterized in that: The basic evaluation coefficient is obtained by weighting the matching coefficient and the interest coefficient, including: After normalizing the matching coefficient and the interest coefficient, a circle is constructed using the matching coefficient as the radius of the circle and the interest coefficient as the height of the circle. A cone model is established and the volume of the cone model is calculated to obtain the basic evaluation coefficient. If the user adds the promotional product to the shopping cart, the action value of the add-to-cart action is recorded as 1, otherwise it is -1; Similarly, if the user adds the marketing product to the collection, the action value of the collection action is recorded as 1, otherwise it is -1; Preset weight factors corresponding to the add-to-cart and favorite actions, and multiply the action value of the add-to-cart action and the action value of the favorite action with their corresponding weight factors, and then sum them to obtain the add-to-cart and favorite coefficient; After normalizing the basic evaluation coefficient and the purchase and collection coefficient, determine whether the purchase and collection coefficient is greater than 0: If the purchase and collection coefficient is greater than 0, the basic evaluation coefficient and the purchase and collection coefficient are used as the two right-angled sides of a right triangle, and the remaining side is connected to form a complete right triangle. The area of ​​the right triangle is recorded as the intention evaluation coefficient; If the purchase and collection coefficient is less than 0, the basic evaluation coefficient and the purchase and collection coefficient will be used as the two right-angled sides of the right triangle respectively. After connecting the remaining side to form a complete right-angled triangle, the center of the right-angled side corresponding to the collection coefficient will be used as the origin, and a circle will be established with half of the collection coefficient as the radius. The circle will cut the right triangle and calculate the area of ​​the remaining right triangle, which will be recorded as the intention evaluation coefficient.

9. The e-commerce analysis and diagnosis method based on artificial intelligence according to claim 8, characterized in that: Determine the user's purchase intention for the marketing product based on the intention evaluation coefficient, specifically including: Three groups of threshold value ranges are preset, and the value range of each group of threshold values ​​corresponds to a purchase intention level. The intention evaluation coefficient is compared with the value range of the three groups of threshold values ​​to obtain the purchase intention level corresponding to the intention evaluation coefficient. The purchase intention level levels include low intention, medium intention and high intention.

Images