Cross-border E-commerce User Portrait Analysis Method and System

Abstract

This application is applicable to the technical field of user portrait analysis, and particularly relates to a cross-border e-commerce user portrait analysis method and system. The method includes: when it is detected that a user clicks on multiple products, obtaining the corresponding behavior information; when a favorite operation is triggered, marking the corresponding product as the first product, and when the user continues to click on the second product, analyzing according to the behavior information to obtain a first attention feature chain and a second attention feature chain; analyzing the first attention feature chain and the second attention feature chain to obtain a product attention sequence; analyzing according to the product attention sequence to obtain product push information. The cross-border e-commerce user portrait analysis method and system provided by this application can effectively recommend products according to the specific attribute preference weights of users for products, reducing the selection cost of users, increasing the conversion rate of the platform, and improving user satisfaction.

Description

Technical Field

[0001] This application belongs to the technical field of user portrait analysis, and particularly relates to a cross-border e-commerce user portrait analysis method and system. Background Art

[0002] User portrait analysis refers to a systematic method of collecting, integrating, and analyzing various types of user data (such as behavior, preferences, needs, background, etc.) to construct a virtual typical user model (i.e., user portrait), so as to deeply understand the characteristics of the target user group and guide work such as product design and marketing strategy formulation.

[0003] In related technologies, in the personalized recommendation system of the current e-commerce platform, the construction of user portraits mainly relies on the statistical analysis of historical behavior data and the matching algorithm of product tags. Usually, by collecting explicit behavior data such as user clicks, favorites, and purchase records, and combining static attributes such as product categories, price ranges, and basic function parameters, a product recommendation system is established to achieve product recommendation. The traditional recommendation system cannot effectively recommend products according to the preference weights of specific attributes of products by users. For example, a user's demand for a pair of headphones is first for high battery life function and then for headphones priced at 200 yuan. The user has favorited high-battery-life headphones priced at 300 yuan, but has not favorited headphones priced at 200 yuan without high battery life function. The traditional recommendation system will only recommend based on the favorited headphones priced at 300 yuan with high battery life function, ignoring the price demand of the user under the high battery life function, resulting in a deviation between the recommended result of the product and the true needs of the user, ultimately leading to an increase in the user's selection cost, a decrease in the platform conversion rate, and a decrease in user experience satisfaction. Summary of the Invention

[0004] The embodiments of this application provide a cross-border e-commerce user portrait analysis method and system, which can improve the problem that products cannot be effectively recommended according to the preference weights of specific attributes of products by users, resulting in an increase in the user's selection cost, a decrease in the platform conversion rate, and a decrease in user experience satisfaction.

[0005] In a first aspect, the embodiments of this application provide a cross-border e-commerce user portrait analysis method, including:

[0006] When it is detected that a user clicks on multiple products, obtain the corresponding behavior information; wherein, the multiple products refer to products of the same category, and the behavior information includes browsing behavior information and interaction behavior information. The browsing behavior information is used to reflect the browsing data of the user when browsing the product homepage; the interaction behavior information is used to reflect the interaction data of the user when interacting on the product homepage;

[0007] When the collection operation is triggered, the corresponding product is marked as the first product. After the user clicks on the second product, the first attention feature chain and the second attention feature chain are obtained by analyzing the behavior information. Among them, the collection operation is used to reflect that the user collects the product or adds it to the shopping cart. The first attention feature chain is used to reflect the degree of attention of the user to different features of the first product. The second attention feature chain is used to reflect the degree of attention of the user to different features of the second product. The second product is used to reflect the product for which the collection operation has not been performed.

[0008] Analyze the first attention feature chain and the second attention feature chain to obtain a product attention sequence. Among them, the product attention sequence is used to reflect the order of the user's needs for different features of a certain category of products. The need refers to the degree of attention of the user to the features.

[0009] Analyze according to the product attention sequence to obtain product push information. Among them, the product push information is used to reflect the products recommended for the user.

[0010] In the technical solutions described above in the embodiments of the present application, at least the following technical effects are achieved:

[0011] The cross-border e-commerce user portrait analysis method provided by the embodiments of the present application first obtains the behavior information corresponding to when the user enters and clicks on products with the same category, including browsing behavior information for reflecting the data generated when the user browses the first product and interaction behavior information for reflecting the data generated when the user interacts with the first product. When the collection operation is triggered, the corresponding product is marked as the first product. After the user clicks on the second product for reflecting the product for which the collection operation has not been performed, the first attention feature chain for reflecting the different features of the first product and the second attention feature chain for reflecting the degree of attention of the user to the different features of the second product are obtained by analyzing the behavior information. Then, the product attention sequence of the user's preference for the product features of the product category is obtained by analyzing the first attention feature chain and the second attention feature chain. Finally, the product push information for reflecting the products recommended for the user is obtained by analyzing the product attention sequence. This method can effectively compare and analyze the degree of attention to different features of the first product in the first state and the degree of attention to different features of the second product in the second state, obtain the priority order of the user's preference for the features of this category of products, and then recommend products of this category to the user according to the priority order through the cross-border e-commerce user portrait analysis system, which can reduce the user's selection cost, improve the conversion rate of the platform, and improve the user's satisfaction.

[0012] In a possible implementation manner of the first aspect, the analyzing the behavior information to obtain the first attention feature chain and the second attention feature chain includes:

[0013] Confirm the behavior information of the first product as the first behavior information; wherein, the first behavior information is used to reflect the behavior information of the first product.

[0014] Integrate the behavior information of multiple second products to obtain a second behavior information set; wherein, the second behavior information set is used to reflect the set of behavior information of multiple second products.

[0015] Analyze according to the first behavior information to obtain a first attention feature chain.

[0016] Perform multi-feature classification according to the second behavior information set to obtain a second attention feature chain..

[0017] In a possible implementation manner of the first aspect, the analyzing according to the first behavior information to obtain a first attention feature chain includes:

[0018] Analyze according to the browsing behavior information of the first behavior information to obtain a browsing path, the browsing time corresponding to each node, and the information entropy; wherein, the browsing path is used to reflect the browsing order of the user for the product display page, the display page includes a picture display page, a parameter display page, and a comment display page, the browsing time is used to reflect the stay time of the user on the product display page; the information entropy is used to reflect the information richness of the product display page, and the node is used to reflect different browsing contents of the browsing path.

[0019] Analyze the interaction behavior information of the first behavior information and the browsing path to obtain the sliding speed corresponding to each node; wherein, the sliding speed is used to reflect the speed at which the user slides the product display page.

[0020] Analyze according to the browsing time, the sliding speed, and the information entropy corresponding to each node to obtain a dimension attention priority sequence; wherein, the dimension attention priority sequence is used to reflect the priority attention situation of the user for different display pages of the product.

[0021] Process the browsing path according to the dimension attention priority sequence to obtain a first attention feature chain.

[0022] In a possible implementation manner of the first aspect, the performing multi-feature classification according to the second behavior information set to obtain a second attention feature chain includes:

[0023] Collect features for the multiple products corresponding to the second behavior information set to obtain a feature set; wherein, the feature set is used to reflect the feature set of the multiple products in the second behavior information set.

[0024] Perform duplicate feature partitioning on the feature set to obtain multiple duplicate features and multiple non-duplicate features; wherein, the duplicate features are used to reflect the same features among multiple commodities, and the non-duplicate features are used to reflect the different features among multiple commodities;

[0025] Perform occurrence frequency analysis on multiple non-duplicate features to obtain multiple required features; wherein, the required features are used to reflect the features whose occurrence frequency of the non-duplicate features is greater than or equal to a preset occurrence frequency;

[0026] Identify multiple duplicate features and multiple required features as multiple necessary features; wherein, the necessary features are used to reflect the features that must be possessed by users when browsing the same type of commodities;

[0027] Arrange multiple necessary features according to the dimension attention priority sequence to obtain a second attention feature chain.

[0028] In a possible implementation manner of the first aspect, the analysis based on the browsing time, the sliding speed, and the information entropy corresponding to each node to obtain a dimension attention priority sequence includes:

[0029] Obtain a reference sliding speed; wherein, the reference sliding speed is used to reflect the speed at which a user slides the commodity display page during normal browsing;

[0030] Process the reference sliding speed and the sliding speed of each node to obtain a first weight for each node; wherein, the first weight is used to reflect the degree of influence of the sliding speed on the attention to the node;

[0031] Process the browsing time corresponding to each node to obtain a second weight corresponding to each node; wherein, the second weight is used to reflect the degree of influence of the browsing time on the attention to the node;

[0032] Analyze based on the first weight, the second weight, and the information entropy to obtain a dimension attention priority sequence.

[0033] In a possible implementation manner of the first aspect, the analysis based on the first weight, the second weight, and the information entropy to obtain a dimension attention priority sequence includes:

[0034] Perform dimension classification according to the information entropy corresponding to each node to obtain a first dimension information entropy and a second dimension information entropy; wherein, the first dimension information entropy is used to reflect the information complexity in terms of picture vision on the display page, and the second dimension information entropy is used to reflect the information complexity in terms of text language on the display page;

[0035] Process according to the first - dimension information entropy, the second weight corresponding to the corresponding node, and the first weight to obtain the first - dimension attention ratio; wherein, the first - dimension attention ratio is used to reflect the user's attention degree in the first dimension; the first dimension is used to reflect the picture display page of the display page.

[0036] Process and compare according to the second - dimension information entropy, browsing time, and the first weight corresponding to each node to obtain the second - dimension attention sequence; wherein, the second - dimension attention sequence is used to reflect the user's attention order in the second dimension, and the second dimension is used to reflect the parameter display page and the comment display page of the display page.

[0037] Confirm the first - dimension attention ratio and the second - dimension attention sequence as the dimension attention priority sequence.

[0038] In a possible implementation manner of the first aspect, the process of processing and comparing according to the second - dimension information entropy, browsing time, and the first weight corresponding to each node to obtain the second - dimension attention sequence includes:

[0039] Compare the browsing time corresponding to each node with the second - dimension information entropy to obtain the attention ratio; wherein, the attention ratio is used to reflect the user's attention distribution status of each node in the second dimension.

[0040] Process the attention ratio corresponding to each node with the first weight to obtain the attention degree; wherein, the attention degree is used to reflect the user's attention degree of each node in the second dimension.

[0041] Sort each node in the second dimension in descending order of the attention degree to obtain the second - dimension attention sequence.

[0042] In a possible implementation manner of the first aspect, the analysis of the first attention feature chain and the second attention feature chain to obtain the product attention sequence includes:

[0043] Compare the first - dimension attention ratio in the first attention feature chain with the first - dimension attention ratio in the second attention feature chain to obtain the first - dimension change status; wherein, the first - dimension change status is used to reflect the user's attention change in the first dimension.

[0044] Perform corresponding - dimension feature comparison on the second - dimension attention sequence in the first attention feature chain and the second - dimension attention sequence in the second attention feature chain to obtain the second - dimension difference feature; wherein, the second - dimension difference feature is used to reflect the difference feature of the corresponding node in the user's second - dimension attention sequence.

[0045] Compare the first - dimension change situation with the preset change situation to obtain the first - dimension attention feature, where the first - dimension attention feature is used to reflect the user's attention feature in the first dimension, and the preset change situation is used to reflect the preset change situation value.

[0046] Confirm the first - dimension attention feature and the second - dimension difference feature as the product attention sequence.

[0047] In a possible implementation manner of the first aspect, the step of comparing the first - dimension change situation with the preset change situation to obtain the first - dimension attention feature includes:

[0048] When the first - dimension change situation is greater than or equal to the preset change situation, confirm the first - dimension feature in the first attention feature chain as the first - dimension attention feature.

[0049] When the first - dimension change situation is less than the preset change situation, confirm the first - dimension feature in the second attention feature chain as the first - dimension attention feature.

[0050] In a second aspect, an embodiment of the present application provides a cross - border e - commerce user portrait analysis system, including:

[0051] An acquisition module, configured to obtain corresponding behavior information when detecting that a user clicks on multiple products. The multiple products refer to products of the same category. The behavior information includes browsing behavior information and interaction behavior information. The browsing behavior information is used to reflect the browsing data of the user when browsing the product homepage. The interaction behavior information is used to reflect the interaction data of the user when interacting on the product homepage.

[0052] A first analysis module, configured to mark the corresponding product as the first product when the favorite operation is triggered. When the user then clicks on a second product, analyze according to the behavior information to obtain a first attention feature chain and a second attention feature chain. The favorite operation is used to reflect that the user favorites the product or adds it to the shopping cart. The first attention feature chain is used to reflect the degree of attention of the user to different features of the first product. The second attention feature chain is used to reflect the degree of attention of the user to different features of the second product. The second product is used to reflect the product for which the favorite operation has not been performed.

[0053] A second analysis module, configured to analyze the first attention feature chain and the second attention feature chain to obtain a product attention sequence. The product attention sequence is used to reflect the order of the user's demands for different features of a certain category of products. The demand refers to the degree of attention of the user to the features.

[0054] A push module, configured to analyze according to the product follow-up sequence to obtain product push information, where the product push information is used to reflect the products recommended for the user.

[0055] In a third aspect, an embodiment of the present application provides a cross-border e-commerce user portrait analysis device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the method according to any one of the above first aspects.

[0056] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, it implements the method according to any one of the above first aspects.

[0057] In a fifth aspect, an embodiment of the present application provides a computer program, and when the computer program runs on a cross-border e-commerce user portrait analysis device, it causes the cross-border e-commerce user portrait analysis device to execute the cross-border e-commerce user portrait analysis method according to any one of the above first aspects.

[0058] It can be understood that the beneficial effects of the above second to fifth aspects can be referred to the relevant descriptions in the above first aspect, and will not be elaborated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0060] Figure 1 is a schematic flowchart of a cross-border e-commerce user portrait analysis method provided by an embodiment of the present application;

[0061] Figure 2 is a schematic implementation flowchart of a cross-border e-commerce user portrait analysis method provided by an embodiment of the present application;

[0062] Figure 3 is a schematic structural diagram of a cross-border e-commerce user portrait analysis system provided by an embodiment of the present application;

[0063] Figure 4 is a schematic structural diagram of a cross-border e-commerce user portrait analysis device provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0064] In the following description, for the purpose of illustration rather than limitation, specific details such as specific system architectures, technologies, etc. are presented to provide a thorough understanding of the embodiments of the present application. However, those skilled in the art should understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from obscuring the description of the present application.

[0065] It should be understood that when used in the specification and appended claims of the present application, the term "comprising" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0066] It should also be understood that the term "and / or" as used in the specification and appended claims of the present application refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0067] As used in the specification and appended claims of the present application, the term "if" can be interpreted as "when", "once", "in response to determining", or "in response to detecting" depending on the context. Similarly, the phrase "if determined" or "if [the described condition or event] is detected" can be interpreted as meaning "once determined", "in response to determining", "once [the described condition or event] is detected", or "in response to detecting [the described condition or event]" depending on the context.

[0068] In addition, in the description of the specification and appended claims of the present application, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0069] The reference to "one embodiment" or "some embodiments" or the like described in the specification of the present application means that a specific feature, structure, or characteristic described in connection with that embodiment is included in one or more embodiments of the present application. Thus, statements such as "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments", etc. that appear in different places in this specification do not necessarily all refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "comprising", "including", "having", and their variants all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0070] In the related art, in the personalized recommendation system of the current e-commerce platform, the construction of user portraits mainly relies on the statistical analysis of historical behavior data and the matching algorithm of product tags. Usually, by collecting explicit behavior data such as user clicks, favorites, and purchase records, and combining static attributes such as product categories, price ranges, and basic function parameters, a product recommendation system is established to achieve product recommendations. The traditional recommendation system cannot effectively recommend products according to the preference weights of specific attributes of users for products. For example, a user's demand order for a pair of headphones is first the high battery life function and then headphones priced at 200 yuan. The user has favorited high-battery-life headphones priced at 300 yuan, but has not favorited headphones priced at 200 yuan without the high battery life function. The traditional recommendation system will only recommend based on the favorited headphones priced at 300 yuan with the high battery life function, ignoring the price requirement below the high battery life function in the user's demand order, resulting in a deviation between the recommended result of the product and the user's true demand, ultimately leading to an increase in the user's selection cost, a decrease in the platform conversion rate, and a decrease in the user experience satisfaction.

[0071] To solve the above problems, the embodiments of the present application provide a cross-border e-commerce user portrait analysis method and system. In this method, when it is detected that a user enters and clicks on multiple products with the same category, behavior information is obtained, which includes browsing behavior information for reflecting the data generated when the user browses the first product and interaction behavior information for reflecting the data generated when the user interacts with the first product. When the favorite operation is triggered, the corresponding product is marked as the first product. After the user continues to click on the second product for reflecting the product without the favorite operation, analysis is performed according to the behavior information to obtain a first attention feature chain for reflecting the different features of the user for the first product and a second attention feature chain for reflecting the attention degree of the user for the different features of the second product. Then, analysis is performed according to the first attention feature chain and the second attention feature chain to obtain a product attention sequence for reflecting the product feature preferences of the user for the product category. Finally, analysis is performed according to the product attention sequence to obtain product push information for reflecting the products recommended for the user. This method can effectively perform a comparative analysis according to the attention degrees of the different features of the first product in the first state and the different features of the second product in the second state, obtain the priority order of the feature preferences of the user for the products in this category, and then recommend products in this category to the user according to the priority order through the cross-border e-commerce user portrait analysis system, which can reduce the user's selection cost, improve the conversion rate of the platform, and improve the user's satisfaction.

[0072] The cross-border e-commerce user portrait analysis method provided by the embodiments of the present application can be applied to a cross-border e-commerce user portrait analysis device. At this time, the cross-border e-commerce user portrait analysis device is the execution subject of the cross-border e-commerce user portrait analysis method provided by the embodiments of the present application. The embodiments of the present application do not impose any restrictions on the specific type of the cross-border e-commerce user portrait analysis device.

[0073] For example, the cross-border e-commerce user portrait analysis device can be a station (STAION, ST) in a WLAN, a mobile phone, a tablet computer, a laptop computer, an ultra-mobile personal computer (UMPC), a netbook, a desktop computer, a smart large screen, a computing device, or other processing devices connected to a wireless modem, a computer, a laptop computer, a handheld computing device. For example, a mobile terminal in a 5G network or a mobile terminal in a future evolved Public Land Mobile Network (PLMN), etc.

[0074] To better understand the cross-border e-commerce user portrait analysis method provided in the embodiments of the present application, the following provides an exemplary introduction to the specific implementation process of the cross-border e-commerce user portrait analysis method provided in the embodiments of the present application.

[0075] Figure 1 and Figure 2 shows a schematic flowchart of the cross-border e-commerce user portrait analysis method provided in the embodiments of the present application. Please refer to Figure 1 and Figure 2 , the cross-border e-commerce user portrait analysis method includes:

[0076] S100, when it is detected that the user clicks on multiple products, obtain the corresponding behavior information; wherein, the multiple products refer to products of the same category, and the behavior information includes browsing behavior information and interaction behavior information. The browsing behavior information is used to reflect the browsing data of the user when browsing the product homepage; the interaction behavior information is used to reflect the interaction data of the user when interacting on the product homepage.

[0077] It can be understood that after the user enters a product list or page of the same or similar category by searching for keywords, clicks on one of the products to enter the display page of the product, and until the user exits the browsing of the product, a browsing of the product is completed, which is the behavior information corresponding to the product. The browsing data may include data such as the browsing time and browsing content generated by the user when browsing the product. The interaction data may include data such as the sliding speed and favorite operation generated by the user when performing browsing operations on the product page.

[0078] Exemplarily, browsing behavior information can be collected through front-end data embedding, Cookies tracking, server logs, and SDKs to collect data such as page access, stay time, and path to form the required browsing behavior information. Interaction behavior information, on the other hand, relies on front-end data embedding (including full data embedding), API / crawler scraping, database storage, and back-end business data integration to record operations such as clicks, inputs, and interactions of users on the product display page to form the required interaction behavior information.

[0079] S200, when a favorite operation is triggered, mark the corresponding product as the first product. When the user then clicks on a second product, analyze the behavior information to obtain a first attention feature chain and a second attention feature chain; where the favorite operation is used to reflect that the user favorites or adds a product to the shopping cart, the first attention feature chain is used to reflect the degree of attention of the user to different features of the first product, the second attention feature chain is used to reflect the degree of attention of the user to different features of the second product, and the second product is used to reflect a product for which no favorite operation has been performed.

[0080] It can be understood that when a user browses multiple products of the same category in a product list, if a user does not favorite a certain product, it may indicate that none of the products the user browsed met their intentions. When a user favorites a product during the browsing process, if the user then continues to click on other products and starts browsing, it indicates that the product the user favorited basically meets the user's purchase intention, but still does not fully meet the user's purchase needs.

[0081] Exemplarily, by analyzing the interaction behavior information in the behavior information, when it is detected that a user browses a product and the favorite operation in the interaction behavior information corresponding to the behavior information of the product is triggered, then determine the product as the first product. Furthermore, when the user then clicks on a second product, integrate the behavior information corresponding to multiple second products into a set of behavior information used to reflect the behavior information of multiple second products, and then analyze according to the behavior information used to reflect the first product to obtain a first attention feature chain. At the same time, process according to the set of behavior information used to reflect the behavior information of multiple second products to obtain a second attention feature chain.

[0082] It is also possible to analyze the interaction behavior information. When the customer service consultation in the interaction behavior information is triggered, obtain the consultation content information, perform language text analysis based on the consultation content to obtain consultation words, and then match the consultation words with the product features to obtain a matching result. When the matching result indicates a successful match, the product is determined as the first product, and the behavior information of the product is determined as the first behavior information. Otherwise, it is the second product and the corresponding second behavior information. Then, integrate the behavior information corresponding to multiple second products into a set of behavior information used to reflect the behavior information of multiple second products, and analyze based on the behavior information used to reflect the first product to obtain the first attention feature chain. At the same time, process the set of behavior information used to reflect the behavior information of multiple second products to obtain the second attention feature chain.

[0083] In a possible implementation manner, in step S200, when the favorite operation is triggered and the user has not stopped browsing, analyze based on the behavior information to obtain the first attention feature chain and the second attention feature chain, including:

[0084] S210, confirm the behavior information of the first product as the first behavior information; where the first behavior information is used to reflect the behavior information of the first product.

[0085] It can be understood that the behavior information corresponding to multiple products is divided by whether the trigger operation in the behavior information corresponding to the product is triggered. That is, when the user browses the first product and performs a favorite operation on the first product, the behavior information corresponding to the first product information is determined as the first behavior information.

[0086] S220, integrate the behavior information of multiple second products to obtain a second set of behavior information; where the second set of behavior information is used to reflect the set of behavior information of multiple second products.

[0087] It can be understood that since multiple second products will be generated during the user's browsing process. Therefore, the behavior information corresponding to multiple second products can be integrated into a second set of behavior information.

[0088] S230, analyze based on the first behavior information to obtain the first attention feature chain.

[0089] Exemplarily, by analyzing the browsing behavior information of the first line of information, the browsing order of the user for the picture display page, parameter display page, and comment display page of the commodity can be obtained, as well as the residence time corresponding to each of the picture display page, parameter display page, and comment display page of the commodity and the information richness corresponding to each of the picture display page, parameter display page, and comment display page of the commodity. Then, by analyzing the interaction behavior information in the first line of information and the browsing order of the user for the picture display page, parameter display page, and comment display page of the commodity, the speed at which the user slides each display page corresponding to each node of the browsing order can be obtained. Then, by analyzing the residence time of the user corresponding to each node on each display page, the speed at which the user slides each display page, and the information richness of each display page, the priority attention situation of the user for different display pages of the commodity can be obtained. Finally, by processing the priority attention situation of the user for different display pages of the commodity and the browsing order of the user for each display page, the first attention feature chain is obtained.

[0090] It is also possible to analyze the interaction behavior information in the first line of information to obtain the consultation information for the user to communicate with the customer service. Then, by processing the feature words in the consultation information and the commodity features, multiple identical features can be obtained. Then, by sorting the multiple identical features according to the appearance order, the first attention feature chain is obtained.

[0091] In a possible implementation manner, in step S230, analyzing according to the first line of information to obtain the first attention feature chain includes:

[0092] S231, analyzing according to the browsing behavior information of the first line of information to obtain the browsing path, the browsing time corresponding to each node, and the information entropy; wherein, the browsing path is used to reflect the browsing order of the user for the display page of the commodity, the display page includes a picture display page, a parameter display page, and a comment display page, the browsing time is used to reflect the residence time of the user on the display page of the commodity; the information entropy is used to reflect the information richness of the display page of the commodity, and the node is used to reflect different browsing contents of the browsing path.

[0093] It can be understood that a node refers to each different section of the commodity display page when the user browses a commodity. For example, when the user browses headphones, the user first clicks on the picture section of the headphone commodity to observe, then slides to the comment section to browse, and finally browses in the parameter display section of the headphone commodity. Then the node represents the different section contents browsed by the user.

[0094] Exemplarily, browsing data marked as the first behavior information is extracted from the behavior database, and data cleaning is performed to eliminate noise and outliers. Sequence analysis techniques or visualization tools (such as the SunburstPartition graph of D3.js) are used to mine the cleaned data to identify the mainstream browsing paths of users. For example, the typical trajectory from the product detail page to the collection is identified, and then the residence time of each path node is calculated through timestamp association. Finally, the information entropy of the display pages corresponding to different nodes of the browsing path is calculated through a formula.

[0095] S232, analyze the interaction behavior information corresponding to the first behavior information and the browsing path to obtain the sliding speed corresponding to each node; wherein, the sliding speed is used to reflect the speed at which the user slides the product display page.

[0096] Exemplarily, extract the user's interaction behavior data (such as click and slide operations) and browsing path trajectory from the first behavior information. For each path node, by listening to the user's touch event, record the starting position coordinates, ending position coordinates and corresponding timestamps of the slide operation, then calculate the displacement distance (the coordinate difference in the horizontal or vertical direction) and time difference, and finally, through the ratio formula of displacement to time difference (speed = displacement / time), calculate the sliding speed node by node, so as to obtain the sliding speed corresponding to different nodes of the user.

[0097] S233, analyze according to the browsing time, sliding speed and information entropy corresponding to each node to obtain the dimension attention priority sequence; wherein, the dimension attention priority sequence is used to reflect the user's priority attention to different display pages of the product.

[0098] Exemplarily, it is possible to first obtain the speed at which the user slides the product display page during normal browsing, then process according to the reference sliding speed and the sliding speed of each node to obtain the initial degree of interest of the user in the content of the display page corresponding to each node of the browsing path, and then process according to the browsing time corresponding to each node to obtain the progressive degree of interest of the user in the content of the display page corresponding to each node of the browsing path. Finally, analyze through the progressive degree of interest of the user in the content of the display page corresponding to each node of the browsing path, the initial degree of interest of the user in the content of the display page corresponding to each node of the browsing path, and the information entropy to obtain the dimension attention priority sequence.

[0099] It is also possible to analyze the browsing time, sliding speed, and information entropy corresponding to each node through the first analysis model to obtain the dimension attention priority sequence, that is, input the browsing time, sliding speed, and information entropy corresponding to each node into the first analysis model, and the first analysis model then outputs the corresponding dimension attention priority sequence. The training process of the first analysis model can use the data obtained by processing the browsing time, sliding speed, and information entropy corresponding to each node and the dimension attention priority sequence as the training data set of the first analysis model, and then input the training data set of the first analysis model into the first analysis model for training and learning to finally obtain the first analysis model.

[0100] In a possible implementation manner, in step S233, analyzing according to the browsing time, sliding speed, and information entropy corresponding to each node to obtain the dimension attention priority sequence includes:

[0101] S2331, obtain the reference sliding speed; wherein, the reference sliding speed is used to reflect the speed at which the user slides the product display page during normal browsing.

[0102] It can be understood that the reference sliding speed is the sliding speed of this user set in advance.

[0103] Exemplarily, the reference sliding speed can be manually input by a person, and the reference sliding speed can also be directly obtained through the sliding database. The sliding database refers to a database that contains the sliding speeds of users when browsing products. These data can be obtained through means such as laboratory experiments, on-site measurements and monitoring, and past experience. After obtaining the data, the collected data is sorted, classified, and archived, useful information and rules are extracted, and then the relevant data is saved to the database to form a sliding database.

[0104] S2332, process the reference sliding speed and the sliding speed of each node to obtain the first weight of each node; wherein, the first weight is used to reflect the degree of influence of the sliding speed on the attention of the node.

[0105] It can be understood that the sliding speed of each node refers to the sliding speed of the user at each node of the browsing path, that is, the sliding speed of the user on different display pages of the product. When the user is interested in a display page in the product display page, the sliding speed will be less than or equal to the reference sliding speed, and when the user is interested in another display page in the product display page, the sliding speed will be greater than the reference sliding speed, and the degree of interest of the user in the product display page is negatively correlated with the sliding speed, that is, the faster the sliding speed of the user on a product display page, the less interested the user is in the product display page. Each node is each display page of the product.

[0106] Exemplarily, the first weight can be represented by calculating the relative speed difference between the reference sliding speed and the sliding speeds of each node, that is, the first weight = (reference sliding speed - sliding speed) ÷ reference sliding speed. If there are three nodes, and the sliding speed corresponding to the first node is 400 px / s, the sliding speed corresponding to the second node is 500 px / s, the sliding speed corresponding to the third node is 700 px / s, and the reference sliding speed is 600 px / s (where px is the pixel unit of the page height), then the first weight corresponding to the first node is 0.33 [(600 - 400) ÷ 600], the first weight corresponding to the second node is 0.17 [(600 - 500) ÷ 600], and the first weight corresponding to the third node is -0.17 [(600 - 700) ÷ 600], and so on. It can be understood that the sign of the first weight indicates whether the user is interested in the product display page. When the first weight is negative, it reflects that the user is not interested in the product display page. When the first weight is positive, it reflects that the user is interested in the product display page, and the absolute value of the first weight reflects the user's initial degree of interest in the product display page.

[0107] S2333. Process according to the browsing time corresponding to each node to obtain the second weight corresponding to each node; wherein, the second weight is used to reflect the influence degree of the browsing time on the attention of the node.

[0108] It can be understood that the second weight is the value of the browsing time corresponding to each node accounting for the total browsing time. The total browsing time refers to the duration from when the user starts browsing a product to when the user finishes browsing the product.

[0109] Exemplarily, if the user browses three product display pages, and the browsing time of the first display page is 30 s, the browsing time of the second display page is 40 s, and the browsing time of the third display page is 30 s, then the second weight corresponding to the first display page, that is, the first node, is 0.3 [30 ÷ (30 + 40 + 30)], the second weight corresponding to the second display page, that is, the second node, is 0.4 [40 ÷ (30 + 40 + 30)], and the second weight corresponding to the third display page, that is, the third node, is 0.3 [30 ÷ (30 + 40 + 30)], and so on.

[0110] S2334. Analyze according to the first weight, the second weight, and the information entropy to obtain the dimension attention priority sequence.

[0111] Exemplarily, by processing the information entropy corresponding to each node, the information complexity in the visual aspect of the pictures on the display page and the information complexity in the text language of the display page can be obtained. Then, based on the information complexity in the visual aspect of the pictures on the display page, the corresponding second weight, and the corresponding first weight, processing is performed to obtain the user's attention degree in the visual aspect of the pictures on the display page. At the same time, based on the information complexity in the text language of the display page, the corresponding browsing time, and the first weight, processing is performed to obtain the attention order of the user between the parameter display page and the comment display page of the display page.

[0112] It is also possible to analyze the first weight, the second weight, and the information entropy through a second analysis model to obtain a dimension attention priority sequence, that is, input the first weight, the second weight, and the information entropy into the second analysis model, and the second analysis model then outputs the corresponding dimension attention priority sequence. The training process of the second analysis model can use the data obtained by processing the first weight, the second weight, and the information entropy with the dimension attention priority sequence as the training data set of the second analysis model, and then input the training data set of the second analysis model into the second analysis model for training and learning to finally obtain the second analysis model.

[0113] With such a setting, by analyzing the user's behavior pattern when browsing the product display page through multi-dimensional data analysis, the user's interest points can be captured more accurately, the user's reference sliding speed can be analyzed, and by comparing the actual sliding speed with the reference sliding speed and combining the user's stay time at each node, the first weight and the second weight are calculated. Then, combined with the concept of information entropy, a comprehensive analysis of these two weights is performed to obtain a dimension attention priority sequence. The dimension attention priority sequence can help the e-commerce platform better understand the user's needs, and can also be used to optimize the product recommendation algorithm, improve the user experience, and thus increase the conversion rate and user satisfaction.

[0114] In a possible implementation manner, in step S2334, analyzing based on the first weight, the second weight, and the information entropy to obtain a dimension attention priority sequence includes:

[0115] S23341, performing dimension classification on the information entropy corresponding to each node to obtain a first-dimensional information entropy and a second-dimensional information entropy; wherein, the first-dimensional information entropy is used to reflect the information complexity in the visual aspect of the pictures on the display page, and the second-dimensional information entropy is used to reflect the information complexity in the text language of the display page.

[0116] It can be understood that dimension classification means classifying the information entropy corresponding to each node according to the visual dimension of the pictures and the text language dimension.

[0117] Exemplarily, classification can be performed based on the attributes corresponding to each node, or can also be performed according to the information entropy type corresponding to each node. The attributes corresponding to each node are the display contents of the product display page. For example, the picture display of the product in the product display page belongs to the picture visual dimension, while the parameter introduction and comment feedback of the product in the product display page belong to the text language dimension, and so on.

[0118] S23342. Process according to the first-dimensional information entropy, the second weight corresponding to the corresponding node, and the first weight to obtain the first-dimensional attention ratio; wherein, the first-dimensional attention ratio is used to reflect the user's attention degree in the first dimension; the first dimension is used to reflect the picture display page of the display page.

[0119] It can be understood that the first-dimensional attention ratio is calculated by the formula: where, is the first-dimensional attention ratio, is the first-dimensional information entropy, is the second weight, is the first weight.

[0120] Exemplarily, if , , , then the first-dimensional attention ratio is 6.63 (6 + 0.3 + ×0.33), and so on.

[0121] S23343. Process and compare according to the second-dimensional information entropy corresponding to each node, the browsing time, and the first weight to obtain the second-dimensional attention sequence; wherein, the second-dimensional attention sequence is used to reflect the user's attention order in the second dimension, and the second dimension is used to reflect the parameter display page and the comment display page of the display page.

[0122] Exemplarily, the browsing time corresponding to each node can be processed with the second-dimensional information entropy to obtain the attention distribution status of the user for each node in the second dimension, and then the attention distribution status of the user for each node in the second dimension is processed with the first weight to obtain the attention degree situation of the user for each node in the second dimension, and then the nodes in the second dimension are sorted according to the attention degree situation of the user to obtain the second-dimensional attention sequence.

[0123] It is also possible to normalize the information entropy of each node in the second dimension to obtain an information complexity ratio, then compare according to the browsing time corresponding to each node in the second dimension to obtain a browsing time ratio, and then analyze according to the browsing time ratio and the information complexity ratio to obtain the attention distribution status of each node in the second dimension for the user. Then, process according to the attention distribution status of each node in the second dimension for the user and the first weight to obtain the attention degree of each node in the second dimension for the user. Then, sort each node in the second dimension according to the attention degree of the user to obtain a second dimension attention sequence.

[0124] In a possible implementation manner, in step S23343, after processing and comparing according to the second dimension information entropy, browsing time, and first weight corresponding to each node, a second dimension attention sequence is obtained, including:

[0125] S233431, compare the browsing time corresponding to each node with the second dimension information entropy to obtain an attention ratio; where the attention ratio is used to reflect the attention distribution status of the user for each node in the second dimension.

[0126] It can be understood that the attention ratio = browsing time ÷ second dimension information entropy. The browsing time reflects the time spent by the user to process the corresponding second dimension information entropy. When the second dimension information entropy is relatively high, the user needs to read and analyze more carefully, so the browsing time is usually longer, which can reflect the attention distribution of the user.

[0127] Exemplarily, if the browsing time is 30s and the second dimension information entropy is 6 bits / character, then the attention ratio is 5 (30÷5), and so on.

[0128] S233432, process according to the attention ratio corresponding to each node and the first weight to obtain an attention degree; where the attention degree is used to reflect the attention degree of the user for each node in the second dimension.

[0129] It can be understood that the attention degree = attention ratio × first weight.

[0130] Exemplarily, if the attention ratio is 5 and the first weight is 0.33, then the attention degree is 1.65, and so on.

[0131] S233433, sort each node in the second dimension in descending order of the attention degree to obtain a second dimension attention sequence.

[0132] For example, if there are three nodes in the second dimension, and the attention degree corresponding to the first node is 1.65, the attention degree corresponding to the second node is 0.54, and the attention degree corresponding to the third node is 1.24, then the original order of first node → second node → third node is sorted in descending order of attention degree, and it becomes first node → third node → second node, and so on.

[0133] With this setting, the second dimension attention sequence is obtained through analysis, which can reveal the user's attention preference during the browsing process, and provide an important basis for e-commerce platforms to optimize the order of product display, improve user interaction and enhance personalized recommendation effects, thereby effectively improving user experience and platform operation efficiency.

[0134] S23344, confirm the first dimension attention ratio and the second dimension attention sequence as the dimension attention priority sequence.

[0135] It can be understood that by analyzing different dimensions, we can get the first dimension attention ratio and the second dimension attention sequence, and thus get the dimension attention priority sequence. Because when users browse products, the image visual dimension occupies most of the product information, and only when they click on the product to enter the product details will the specific parameters and comments of the product be displayed.

[0136] With this setting, by dividing information entropy according to different dimensions, it is possible to analyze the degree of user attention to various product features of a category of goods from the two dimensions of visual display and text description, and conduct more refined classification and priority sorting, thereby achieving accurate positioning of user focus.

[0137] S234, processing the browsing path according to the dimension focus priority sequence to obtain the first focus feature chain.

[0138] Exemplarily, if the browsing path is A→B→C, where the dimensional attention sequence reflects that C's attention is greater than B, and A's attention is the smallest, then the first attention feature chain is C→B→A, reflecting the user's deep attention to specific content.

[0139] With this setting, through comprehensive analysis of multi-dimensional data (browsing path, dwell time, sliding speed, information entropy), the priority of users' attention features for a category of goods is constructed, and display formats that meet their attention priorities can be recommended to different user groups (such as user push technology details page of preference parameters), thereby effectively recommending products based on users' preference weights for specific attributes of products.

[0140] S240, performing multi-feature classification according to the second behavior information set to obtain a second focus feature chain.

[0141] Exemplarily, by processing multiple products corresponding to the second behavior information set, a feature set of multiple products in the second information set can be obtained. Then, by processing according to the feature set of multiple products in the second information set, the same features and different features among multiple products can be obtained. Then, by processing according to the different features among multiple products, features with an occurrence frequency greater than or equal to a preset occurrence frequency among the different features of multiple products can be obtained. Then, by determining the same features among multiple products and the features with an occurrence frequency greater than or equal to the preset occurrence frequency among the different features of multiple products as features that must be possessed by the user when browsing products of the same category, and finally sorting according to the attention priority sequence of the features that must be possessed by the user when browsing products of the same category, a second attention feature chain can be obtained.

[0142] Alternatively, by performing feature analysis on the second behavior information, multiple features in the second behavior information can be obtained. Then, by counting the occurrence times according to the multiple features, a statistical histogram can be obtained. Then, the features with an occurrence number greater than a preset occurrence number in the statistical histogram are determined as features that must be possessed by the user when browsing products of the same category. Then, by matching the features with an occurrence number less than the preset occurrence number in the statistical histogram with a preset feature database, matching features can be obtained. Then, the matching features are confirmed as features that must be possessed by the user when browsing products of the same category. Finally, by sorting multiple mandatory features according to the attention priority sequence, a second attention feature chain can be obtained.

[0143] With such a setting, by determining whether the collection operation is triggered, the behavior information generated by the user browsing products can be divided into the first behavior information and the second behavior information set, which can better refine the user behavior and perform further analysis, and can also improve the accuracy of personalized recommendation, optimize the user experience, and thus enhance the ability to recommend products that the user is interested in.

[0144] In a possible implementation manner, in step S240, according to the second behavior information set, multi-feature classification is performed to obtain a second attention feature chain, including:

[0145] S241, collecting features according to multiple products corresponding to the second behavior information set to obtain a feature set; wherein, the feature set is used to reflect the feature set of multiple products in the second behavior information set.

[0146] Exemplarily, multiple products in the second information set can be traversed for features through a programming language (such as Python) and a data processing library (such as Pandas), and corresponding data can be extracted according to the defined feature list to generate a feature set. The feature list refers to the features of the products. The product features can be manually input by humans. The product features can also be directly obtained from a feature database. The feature database refers to a database that contains the features of different types of products. These data can be obtained through means such as laboratory experiments, on-site measurements and monitoring, and past experience. After obtaining the data, the collected data is sorted, classified, and archived, useful information and rules are extracted, and the relevant data is then saved in the database to form a feature database.

[0147] S242. Perform duplicate feature division processing on the feature set to obtain multiple duplicate features and multiple non-duplicate features; among them, the duplicate features are used to reflect the same features among multiple products, and the non-duplicate features are used to reflect the different features among multiple products.

[0148] Exemplarily, by analyzing, the intersection operation can be performed on the feature sets of each product to find out the common features of all products, which are the duplicate features, and the difference operation can be performed on the feature sets of each product to find out the unique features of each product, which are the non-duplicate features.

[0149] S243. Perform occurrence frequency analysis on multiple non-duplicate features to obtain multiple required features; among them, the required features are used to reflect the features whose occurrence frequency of non-duplicate features is greater than or equal to the preset occurrence frequency.

[0150] It can be understood that the frequency analysis refers to the number of times a feature of a product appears in the number of products browsed. The preset occurrence frequency refers to the preset occurrence frequency of the features of the products.

[0151] Exemplarily, the preset occurrence frequency can be manually input by humans. The preset occurrence frequency can also be directly obtained from a frequency database. The frequency database refers to a database that contains the occurrence times of the features of different types of products. These data can be obtained through means such as laboratory experiments, on-site measurements and monitoring, and past experience. After obtaining the data, the collected data is sorted, classified, and archived, useful information and rules are extracted, and the relevant data is then saved in the database to form a frequency database.

[0152] S244. Confirm multiple duplicate features and multiple required features as multiple necessary features; among them, the necessary features are used to reflect the features that users must possess when browsing products of the same type.

[0153] It can be understood that during the process of browsing products, users will encounter multiple products with some repeated product features. These repeated features may be the basic attributes, function descriptions, or other common attributes of the products. Moreover, each product may also have some unique required features, and the required features reflect the individuality or special attributes of the product. By integrating and confirming these repeated features and required features, a set of multiple necessary features is formed.

[0154] S245, arrange according to the priority sequence of dimensional attention based on multiple necessary features to obtain the second attention feature chain.

[0155] It can be understood that the necessary features can be the features reflected in different product display pages. For example, the features of the parameter display page in the product display page are different from those of the comment display page.

[0156] Exemplarily, if the dimensional attention sequence shows that the priority of the parameter display page is higher than that of the comment display page, then arrange the features belonging to the parameter display page among the features of the comment display page.

[0157] With such settings, through feature collection, division, screening, and sorting, it is possible to effectively identify and utilize the key attention features of users when browsing products, thereby providing strong technical support for product recommendation, display optimization, product development, and user satisfaction improvement on the e-commerce platform.

[0158] S300, analyze the first attention feature chain and the second attention feature chain to obtain the product attention sequence; among them, the product attention sequence is used to reflect the order of the needs of users for different features of a certain category of products, and the need refers to the degree of attention of users to the features.

[0159] Exemplarily, the change in the user's attention to the first dimension can be obtained by comparing the first-dimensional attention ratio in the first attention feature chain with the first-dimensional attention ratio in the second attention feature chain. At the same time, according to the second-dimensional attention sequence in the first attention feature chain and the second-dimensional attention sequence in the second attention feature chain, the differential features of the corresponding nodes in the second-dimensional attention sequence of the user are obtained. Then, according to the change in the user's attention to the first dimension and the preset change situation, the attention features of the user in the first dimension are obtained. Finally, the attention features of the user in the first dimension and the differential features of the corresponding nodes in the second-dimensional attention sequence of the user are confirmed as the product attention sequence.

[0160] It is also possible to compare the first - dimension attention ratio in the first attention feature chain with the first - dimension attention ratio in the second attention feature chain to obtain an attention ratio. At the same time, process the second - dimension attention sequence in the first attention feature chain and the second - dimension attention sequence in the second attention feature chain to obtain the differential features of the corresponding nodes in the second - dimension attention sequence of the user. Then, compare and process the attention ratio with the value 1 to obtain the attention features of the user in the first dimension. Finally, confirm the attention features of the user in the first dimension and the differential features of the corresponding nodes in the second - dimension attention sequence of the user as the product attention sequence. Among them, the attention ratio refers to the first - dimension attention ratio in the first attention feature chain divided by the first - dimension attention ratio in the second attention feature chain.

[0161] In a possible implementation manner, in step S300, analyze the first attention feature chain and the second attention feature chain to obtain the product attention sequence, including:

[0162] S310, compare the first - dimension attention ratio in the first attention feature chain with the first - dimension attention ratio in the second attention feature chain to obtain the first - dimension change situation; where the first - dimension change situation is used to reflect the change in the user's attention to the first dimension.

[0163] It can be understood that the first - dimension change situation refers to the difference between the first - dimension attention ratio in the first attention feature chain and the first - dimension attention ratio in the second attention feature chain.

[0164] Exemplarily, if the first - dimension attention ratio in the first attention feature chain is 2.022 and the second - dimension attention ratio in the second attention feature chain is 3.05, then the first - dimension change situation is 1.028 (3.05 - 2.022), and so on.

[0165] S320, perform corresponding - dimension feature comparison on the second - dimension attention sequence in the first attention feature chain and the second - dimension attention sequence in the second attention feature chain to obtain the second - dimension differential features; where the second - dimension differential features are used to reflect the differential features of the corresponding nodes in the second - dimension attention sequence of the user.

[0166] It can be understood that the dimension - feature comparison means comparing the features at the same node position, that is, comparing the parameter display pages with each other and comparing the comment display pages with each other. The differential features refer to the differences between the features in the second - dimension attention sequence of the first attention feature chain and the features in the second - dimension attention sequence of the second attention feature chain.

[0167] Exemplarily, if the features in the second - dimension attention sequence in the first attention feature chain are high user experience - high battery life - high price, while the features in the second - dimension attention sequence in the second attention feature chain are high user experience - low battery life - low price, then the second - dimension differential feature is high user experience - high battery life - low price, and so on.

[0168] S330. Compare the first - dimension change condition with the preset change condition to obtain the first - dimension attention feature. Among them, the first - dimension attention feature is used to reflect the user's attention feature in the first dimension, and the preset change condition is used to reflect the preset change condition value.

[0169] Exemplarily, the preset change condition can be input manually by a person, or the preset change condition can be directly obtained from the change - condition database. The change - condition database refers to a database that contains the features of the first - dimension change conditions of different types of commodities. These data can be obtained through means such as laboratory experiments, on - site measurements and monitoring, and past experience. After obtaining the data, the collected data is sorted, classified, and archived, useful information and rules are extracted, and then the relevant data is saved to the database to form the change - condition database. The first - dimension attention feature under different comparison results can be obtained by comparing the first - dimension change condition with the preset change condition.

[0170] It is also possible to compare the attention ratio with the value 1 to obtain the comparison result, and then determine the first - dimension attention feature according to different comparison results. Among them, the comparison result refers to the symbol after comparison, that is, the comparison result is a positive number or a negative number. When the comparison result is negative, the first - dimension attention feature is the first - dimension feature in the second attention feature chain; when the comparison result is positive, the first - dimension attention feature is the first - dimension feature in the first attention feature chain.

[0171] In a possible implementation manner, in step S330, comparing the first - dimension change condition with the preset change condition to obtain the first - dimension attention feature includes:

[0172] S331. When the first - dimension change condition is greater than or equal to the preset change condition, confirm the first - dimension feature in the first attention feature chain as the first - dimension attention feature.

[0173] It can be understood that when the first - dimension change condition is greater than or equal to the preset change condition, it indicates that the user's degree of interest in the picture visual dimension has not changed significantly. That is, compared with the commodity whose picture is not collected after browsing, the user prefers the picture display of the commodity whose picture is collected after browsing. At this time, the first - dimension feature of the first attention feature chain is confirmed as the feature that the user is more interested in visually in the picture, that is, the first - dimension attention feature.

[0174] S332. When the change condition of the first dimension is less than the preset change condition, confirm the first-dimension feature in the second attention feature chain as the first-dimension attention feature.

[0175] It can be understood that when the change condition of the first dimension is less than the preset change condition, it indicates that the user's degree of interest in the picture visual dimension has changed significantly. That is, compared with the product for which the collection operation is performed after browsing, the user prefers the picture display of the product for which the collection operation is not performed after browsing. At this time, confirm the first-dimension feature of the second attention feature chain as the feature that the user is more interested in visually in the picture, that is, the first-dimension attention feature.

[0176] With such a setting, by comparing the change condition of the first dimension, it is possible to analyze the user's preference change for the appearance features of a certain category of products in the picture visual dimension, so as to be able to recommend products with such appearance features to the user from the picture, improve the user's click-through rate, and further improve the user's conversion rate.

[0177] S340. Confirm the first-dimension attention feature and the second-dimension difference feature as the product attention sequence.

[0178] It can be understood that by determining the attention feature of the user in the first dimension and the difference feature of the user in the second dimension, it is possible to determine what the user's attention sequence for a certain category of products is.

[0179] Exemplarily, when the user browses products in the category of headphones, the first-dimension attention feature is over-ear headphones, and the second-dimension difference feature reflects that the features preferred by the user are arranged in the order of high user experience - high battery life - low price, then it shows that the user prefers this category of products in the category of headphones.

[0180] With such a setting, by analyzing the attention features of the user in different dimensions, it is possible to accurately capture the user's attention changes and difference features, and it is also possible to construct a product attention sequence that reflects the user's true preferences, thus providing strong support data for personalized service product recommendations on the e-commerce platform.

[0181] S400. Analyze according to the product attention sequence to obtain product push information; wherein, the product push information is used to reflect the products recommended for the user.

[0182] It can be understood that by analyzing the product attention sequence, it is possible to clarify the user's degree of interest in different features of a certain category of products in a certain category, so as to, based on this product attention sequence, push the products liked by the user to the user before the user purchases the product, so that the user purchases the product.

[0183] With such settings, by comparing and analyzing the degree of attention to different features of the first product in the first state and the degree of attention to different features of the second product in the second state, the priority order of the user's preferences for the features of products in this category is obtained. Then, through the cross-border e-commerce user portrait analysis system, products in this category are recommended to the user according to the priority order, which can reduce the user's selection cost, improve the conversion rate of the platform, and enhance the user's satisfaction.

[0184] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not imply the order of execution. The order of execution of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0185] Corresponding to the cross-border e-commerce user portrait analysis method described in the above embodiments, the embodiments of the present application further provide a cross-border e-commerce user portrait analysis system. Each module of the cross-border e-commerce user portrait analysis system can implement each step of the cross-border e-commerce user portrait analysis method. Figure 3 The block diagram of the cross-border e-commerce user portrait analysis system provided by the embodiments of the present application is shown. For the sake of convenience of description, only the parts related to the embodiments of the present application are shown.

[0186] Referring to Figure 3 , the cross-border e-commerce user portrait analysis system includes:

[0187] An acquisition module, configured to acquire corresponding behavior information when it is detected that the user clicks on multiple products; wherein, the multiple products refer to products of the same category, and the behavior information includes browsing behavior information and interaction behavior information. The browsing behavior information is used to reflect the browsing data of the user when browsing the product homepage; the interaction behavior information is used to reflect the interaction data of the user when interacting on the product homepage.

[0188] A first analysis module, configured to mark the corresponding product as the first product when the collection operation is triggered. When the user continues to click on the second product, analyze according to the behavior information to obtain a first attention feature chain and a second attention feature chain; wherein, the collection operation is used to reflect that the user collects the product or adds it to the shopping cart. The first attention feature chain is used to reflect the degree of attention of the user to different features of the first product, and the second attention feature chain is used to reflect the degree of attention of the user to different features of the second product. The second product is used to reflect the product for which the collection operation has not been performed.

[0189] A second analysis module, configured to analyze the first attention feature chain and the second attention feature chain to obtain a product attention sequence; wherein, the product attention sequence is used to reflect the order of the user's needs for different features of products in a certain category. The need refers to the degree of attention of the user to the features.

[0190] A push module, configured to analyze according to a product attention sequence to obtain product push information, where the product push information is used to reflect the products recommended for a user.

[0191] It should be noted that for the content such as information interaction and execution process between the above-mentioned systems / units, since they are based on the same concept as the method embodiments of this application, for their specific functions and the technical effects brought, reference can be specifically made to the method embodiment part, and details will not be repeated here.

[0192] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the above-mentioned division of each functional unit and module is used as an example. In actual applications, the above-mentioned functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction and do not limit the protection scope of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the foregoing method embodiments, and details will not be repeated here.

[0193] The embodiment of this application further provides a cross-border e-commerce user portrait analysis device. Figure 4 FIG. 4 is a schematic structural diagram of a cross-border e-commerce user portrait analysis device 4 provided by an embodiment of this application. As Figure 4 shown, the cross-border e-commerce user portrait analysis device 4 in this embodiment includes: at least one processor 40 ( Figure 4 only one is shown in the figure), at least one memory 41 ( Figure 4 only one is shown in the figure), and a computer program 42 stored in the at least one memory 41 and executable on the at least one processor 40. When the processor 40 executes the computer program 42, the cross-border e-commerce user portrait analysis device 4 implements the steps in any of the above-mentioned cross-border e-commerce user portrait analysis method embodiments, or enables the cross-border e-commerce user portrait analysis device 4 to implement the functions of each module / unit in the above-mentioned system embodiments.

[0194] Exemplarily, the computer program 42 may be divided into one or more modules / units, which are stored in the memory 41 and executed by the processor 40 to complete the present application. The one or more modules / units may be a series of computer program instruction segments capable of performing specific functions, and these instruction segments are used to describe the execution process of the computer program 42 in the cross-border e-commerce user portrait analysis device 4.

[0195] The cross-border e-commerce user portrait analysis device 4 may be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The cross-border e-commerce user portrait analysis device 4 may include, but is not limited to, a processor 40 and a memory 41. Those skilled in the art can understand that Figure 4 merely examples of the cross-border e-commerce user portrait analysis device 4 are provided, and do not constitute a limitation on the cross-border e-commerce user portrait analysis device 4. It may include more or fewer components than shown in the figure, or combine some components, or have different components. For example, it may also include input / output devices, network access devices, a bus, etc.

[0196] The processor 40 may be a central processing unit (CPU), and the processor 40 may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0197] The memory 41 may be an internal storage unit of the cross-border e-commerce user portrait analysis device 4 in some embodiments, such as the hard disk or memory of the cross-border e-commerce user portrait analysis device 4. The memory 41 may also be an external storage device of the cross-border e-commerce user portrait analysis device 4 in other embodiments, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. equipped on the cross-border e-commerce user portrait analysis device 4. Further, the memory 41 may also include both the internal storage unit and the external storage device of the cross-border e-commerce user portrait analysis device 4. The memory 41 is used to store an operating system, application programs, a boot loader, data, and other programs, such as the program code of the computer program. The memory 41 may also be used to temporarily store data that has been output or will be output.

[0198] An embodiment of the present application also provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps in any of the above method embodiments are implemented.

[0199] An embodiment of the present application provides a computer program product, and when the computer program product runs on the cross-border e-commerce user portrait analysis device, the cross-border e-commerce user portrait analysis device is enabled to implement the steps in any of the above method embodiments.

[0200] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it may be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above method embodiments of the present application, a computer program may be used to instruct relevant hardware to complete. The computer program may be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps in the above method embodiments may be implemented. Among them, the computer program includes computer program code, and the computer program code may be in the form of source code, object code, an executable file, or some intermediate form. The computer-readable medium may at least include: any entity or device capable of carrying the computer program code to the cross-border e-commerce user portrait analysis device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk, or an optical disc, etc.

[0201] In the above embodiments, the descriptions of the various embodiments have their own focuses. For parts not detailed or recorded in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0202] Those of ordinary skill in the art will appreciate that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or in a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of this application.

[0203] In the embodiments provided in this application, it should be understood that the disclosed cross-border e-commerce user portrait analysis system and device can be implemented in other ways. For example, the cross-border e-commerce user portrait analysis system embodiments described above are merely illustrative. For example, the division of the modules or units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of devices or units can be in electrical, mechanical or other forms.

[0204] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0205] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them; although this application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for analyzing cross-border e-commerce user portraits, characterized in that: include: When it is detected that a user clicks on multiple products, corresponding behavior information is obtained; wherein the multiple products refer to products of the same category, and the behavior information includes browsing behavior information and interactive behavior information, wherein the browsing behavior information is used to reflect the browsing data of the user when browsing the product homepage; and the interactive behavior information is used to reflect the interactive data when the user interacts on the product homepage; When the collection operation is triggered, the corresponding product is marked as the first product. When the user continues to click on the second product, the behavior information is analyzed to obtain the first attention feature chain and the second attention feature chain. The collection operation is used to reflect that the user has collected or added the product to the shopping cart. The first attention feature chain is used to reflect the user's attention to different features of the first product. The second attention feature chain is used to reflect the user's attention to different features of the second product. The second product is used to reflect the product that has not been collected. Analyze the first attention feature chain and the second attention feature chain to obtain a product attention sequence; wherein the product attention sequence is used to reflect the order of user's demand for different features of a certain category of products, and the demand refers to the user's attention to the feature; The commodity push information is obtained by analyzing the commodity attention sequence; wherein the commodity push information is used to reflect the commodity recommended to the user.

2. The cross-border e-commerce user portrait analysis method according to claim 1, characterized in that: The analyzing according to the behavior information to obtain the first focus feature chain and the second focus feature chain includes: confirming the behavior information of the first commodity as first behavior information; wherein the first behavior information is used to reflect the behavior information of the first commodity; Integrate the behavior information of the plurality of second commodities to obtain a second behavior information set; wherein the second behavior information set is used to reflect the collection of the behavior information of the plurality of second commodities; Analyze the first behavior information to obtain a first focus feature chain; Multi-feature classification is performed according to the second behavior information set to obtain a second focus feature chain.

3. The cross-border e-commerce user portrait analysis method according to claim 2, characterized in that: The step of analyzing the first behavior information to obtain a first focus feature chain includes: Analyze the browsing behavior information of the first behavior information to obtain a browsing path, browsing time corresponding to each node, and information entropy; wherein the browsing path is used to reflect the browsing order of the user on the display page of the product, the display page includes a picture display page, a parameter display page, and a comment display page, and the browsing time is used to reflect the user's stay time on the display page of the product; the information entropy is used to reflect the information richness of the display page of the product, and the node is used to reflect the different browsing contents of the browsing path; Analyze the interactive behavior information of the first behavior information and the browsing path to obtain a sliding speed corresponding to each node; wherein the sliding speed is used to reflect the speed at which the user slides the product display page; Analyze the browsing time, the sliding speed and the information entropy corresponding to each node to obtain a dimension attention priority sequence; wherein the dimension attention priority sequence is used to reflect the user's priority attention to different display pages of the product; The browsing path is processed according to the dimensional focus priority sequence to obtain a first focus feature chain.

4. The cross-border e-commerce user portrait analysis method according to claim 3, characterized in that: The performing multi-feature classification according to the second behavior information set to obtain a second focus feature chain includes: Collecting features of the plurality of commodities corresponding to the second behavior information set to obtain a feature set; wherein the feature set is used to reflect the feature set of the plurality of commodities in the second behavior information set; Performing repeated feature division processing on the feature set to obtain a plurality of repeated features and a plurality of non-repeated features; wherein the repeated features are used to reflect the same features among the plurality of commodities, and the non-repeated features are used to reflect the different features among the plurality of commodities; Performing an occurrence frequency analysis on the plurality of non-repeating features to obtain a plurality of required features; wherein the required features are used to reflect features whose occurrence frequency of the non-repeating features is greater than or equal to a preset occurrence frequency; Confirming the plurality of repeated features and the plurality of required features as a plurality of necessary features; wherein the necessary features are used to reflect the features that the user must possess when browsing the same type of the products; The plurality of necessary features are arranged in the priority sequence of the dimensional focus to obtain a second focus feature chain.

5. The cross-border e-commerce user portrait analysis method according to claim 3, characterized in that: The analyzing according to the browsing time, the sliding speed and the information entropy corresponding to each node to obtain the dimension attention priority sequence includes: Obtaining a reference sliding speed; wherein the reference sliding speed is used to reflect the speed at which a user slides a product display page during normal browsing; Processing is performed according to the reference sliding speed and the sliding speed of each node to obtain a first weight of each node; wherein the first weight is used to reflect the degree of influence of the sliding speed on the attention of the node; Processing is performed according to the browsing time corresponding to each node to obtain a second weight corresponding to each node; wherein the second weight is used to reflect the influence of the browsing time on the attention of the node; An analysis is performed based on the first weight, the second weight and the information entropy to obtain a dimensional attention priority sequence.

6. The cross-border e-commerce user portrait analysis method according to claim 5, characterized in that: The analyzing according to the first weight, the second weight and the information entropy to obtain the dimension attention priority sequence includes: Dimensional classification is performed according to the information entropy corresponding to each node to obtain first-dimensional information entropy and second-dimensional information entropy; wherein the first-dimensional information entropy is used to reflect the information complexity of the display page in terms of image vision, and the second-dimensional information entropy is used to reflect the information complexity of the display page in terms of text language; Processing is performed according to the first dimension information entropy, the second weight corresponding to the corresponding node, and the first weight to obtain the first dimension attention ratio; wherein the first dimension attention ratio is used to reflect the user's attention to the first dimension; the first dimension is used to reflect the picture display page of the display page; The second dimension information entropy, browsing time and first weight corresponding to each node are processed and compared to obtain a second dimension attention sequence; wherein the second dimension attention sequence is used to reflect the user's attention order to the second dimension, and the second dimension is used to reflect the parameter display page and the comment display page of the display page; Confirm the first dimension attention ratio and the second dimension attention sequence as the dimension attention priority sequence.

7. The cross-border e-commerce user portrait analysis method according to claim 6, characterized in that: The second dimension information entropy, browsing time and first weight corresponding to each node are processed and compared to obtain the second dimension attention sequence, including: The attention ratio is obtained by comparing the browsing time corresponding to each node with the information entropy of the second dimension; wherein the attention ratio is used to reflect the user's attention allocation status at each node in the second dimension; Processing the attention proportion corresponding to each node and the first weight to obtain the attention degree; wherein the attention degree is used to reflect the user's attention to each node in the second dimension; The nodes of the second dimension are sorted in descending order of the attention degree to obtain a second dimension attention sequence.

8. The cross-border e-commerce user portrait analysis method according to claim 1, characterized in that: The analyzing the first attention feature chain and the second attention feature chain to obtain a commodity attention sequence includes: Comparing the first dimension attention ratio in the first attention feature chain with the first dimension attention ratio in the second attention feature chain, obtaining a first dimension change status; wherein the first dimension change status is used to reflect the user's attention change to the first dimension; Comparing the corresponding dimensional features of the second dimensional attention sequence in the first attention feature chain with the second dimensional attention sequence in the second attention feature chain, obtaining a second dimensional difference feature; wherein the second dimensional difference feature is used to reflect the difference feature of the corresponding node in the second dimensional attention sequence of the user; According to the comparison between the first dimension change status and the preset change status, a first dimension attention feature is obtained; wherein the first dimension attention feature is used to reflect the user's attention feature in the first dimension, and the preset change status is used to reflect a preset change status value; The first dimension attention feature and the second dimension difference feature are confirmed as a product attention sequence.

9. The cross-border e-commerce user portrait analysis method according to claim 8, characterized in that: The step of comparing the first dimension change status with a preset change status to obtain the first dimension focus feature includes: When the first dimension change condition is greater than or equal to the preset change condition, confirming the first dimension feature in the first focus feature chain as the first dimension focus feature; When the first dimensional change condition is smaller than the preset change condition, the first dimensional feature in the second focus feature chain is confirmed as the first dimensional focus feature.

10. A cross-border e-commerce user portrait analysis system, characterized in that: include: An acquisition module, configured to acquire corresponding behavior information when detecting that a user clicks on multiple products; wherein the multiple products refer to products of the same category, and the behavior information includes browsing behavior information and interactive behavior information, wherein the browsing behavior information is used to reflect the browsing data of the user when browsing the product homepage; and the interactive behavior information is used to reflect the interactive data when the user interacts on the product homepage; A first analysis module is used to mark the corresponding product as the first product when the collection operation is triggered, and when the user continues to click on the second product, analyze according to the behavior information to obtain a first attention feature chain and a second attention feature chain; wherein the collection operation is used to reflect that the user has collected or added the product to the shopping cart, the first attention feature chain is used to reflect the user's attention to different features of the first product, the second attention feature chain is used to reflect the user's attention to different features of the second product, and the second product is used to reflect the product that has not been collected; A second analysis module is used to analyze the first attention feature chain and the second attention feature chain to obtain a product attention sequence; wherein the product attention sequence is used to reflect the order of user's demand for different features of a certain category of products, and the demand refers to the user's attention to the feature; The push module is used to analyze the commodity attention sequence to obtain commodity push information; wherein the commodity push information is used to reflect the commodity recommended to the user.

Images