An ai virtual counter marketing method and system based on dynamic evaluation

Abstract

The application discloses an AI virtual counter marketing method and system based on dynamic evaluation, comprising the following steps: constructing a marketing function on a service end and accessing an online business system through a standardized interface, and receiving running parameters configured through a console; when a target user enters a subscription page, triggering and popping up a virtual counter interaction interface; collecting commodity data and performing attraction score calculation of commodities, forming a to-be-replaced commodity list based on the attraction score; based on the attraction score and a real-time display state of the commodities, performing counter display effect evaluation to determine commodity batch replacement; and updating commodities displayed in the counter according to a determination result of the commodity batch replacement determination and the to-be-replaced commodity list. The application introduces innovative attraction calculation algorithms and display effect evaluation algorithms to eliminate inefficient commodities in real time, quantitatively evaluate the display effect of the entire counter, and intelligently trigger batch replacement.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of internet marketing, in particular to an AI virtual counter marketing method and system based on dynamic evaluation. BACKGROUND

[0002] With the in-depth development of internet service business, benefit mall (such as membership, points, digital service subscription, etc.) has become an important platform for enterprises to improve user stickiness and value realization. In such scenarios, how to accurately recommend suitable benefit goods to users to improve the subscription rate and user satisfaction is the core goal of operation.

[0003] At present, common marketing recommendation schemes are mostly based on list recommendation of collaborative filtering or content recommendation, or traditional customer service robots. For list recommendation, such technology cannot respond to market strategy changes and user interest drift in real time; traditional customer service robots focus on answering questions and are difficult to form deep linkage with product display and user real-time behavior data, and cannot realize dynamic and accurate product recommendation and update based on conversation content. The existing scheme only focuses on the performance of a single product (such as CTR), lacks quantitative evaluation and optimization mechanism for the synergistic effect of the entire product combination; product display update relies on fixed period or manual operation, and cannot automatically and intelligently optimize the display content based on user real-time feedback data (click, purchase), resulting in the decay of counter display effect over time. SUMMARY

[0004] The purpose of the present application is to provide an AI virtual counter marketing method and system based on dynamic evaluation to solve the problems of static recommendation strategy and lack of overall effect evaluation mentioned in the background art.

[0005] To achieve the above purpose, the present application adopts the following technical solutions:

[0006] According to one aspect of the present application, an AI virtual counter marketing method based on dynamic evaluation is provided, which comprises: constructing a marketing function on a server and accessing an online business system through a standardized interface, receiving running parameters configured through a console;

[0007] When a target user enters a subscription page, a virtual counter interaction interface is triggered and popped up;

[0008] Collecting product data and calculating the attraction score of the product, and forming a list of products to be replaced based on the attraction score;

[0009] Based on the attraction score and the real-time display state of the product, the display effect of the counter is evaluated to determine the replacement of the product batch;

[0010] According to the determination result of the commodity batch replacement determination and the replacement commodity list, the commodities displayed in the counter are updated.

[0011] According to the foregoing scheme, the operation parameters include but are not limited to user configuration, counter display commodity quantity, display period, attention threshold, single product replacement threshold, and single replacement commodity quantity.

[0012] According to the foregoing scheme, the user configuration includes providing a user filtering interface, and supporting combined query and filtering according to one or more conditions of user basic attributes, ordered commodity conditions, and recent login time.

[0013] According to the foregoing scheme, the calculation formula of the attraction score is:

[0014] ;

[0015] α is the attraction score, w y represents a weight factor in the time period y, C y represents the number of commodity clicks in the time period y, B y represents the number of commodity purchases in the time period y, S y represents the number of commodity displays in the time period y, σ is a decay coefficient, and t is the current time; wherein w y and σ are preset parameters.

[0016] According to the foregoing scheme, the attraction score of the commodity is compared with a preset single product replacement threshold, if the attraction score is lower than the single product replacement threshold, the commodity is marked as a replacement commodity, and the attraction score is sorted to form the replacement commodity list.

[0017] According to the foregoing scheme, the counter display effect evaluation includes calculating a counter display effect score:

[0018] ;

[0019] F is the counter display effect score, n is the total number of commodities in the counter, is the attraction score of commodity i at time t, is a binary function, indicating whether commodity i is displayed at time t.

[0020] According to the foregoing scheme, the counter display effect score is compared with a preset attention threshold, if the counter display effect score is less than the preset attention threshold, the commodity batch replacement is triggered.

[0021] According to the foregoing scheme, the updating of the commodities displayed in the counter further includes performing forced refreshing of the counter commodities at a preset display period.

[0022] Based on the foregoing scheme, the virtual counter interaction interface comprises a virtual cashier and a dynamic commodity display counter; the virtual cashier performs multi-round dialogue interaction with the user through natural language processing technology, identifies the user's intention and recommends commodities.

[0023] According to another aspect of the present application, an AI virtual counter marketing system based on dynamic evaluation is provided, which comprises: a configuration management module, a virtual counter interaction module, an attraction calculation module, a display effect evaluation module, and a commodity updating module.

[0024] The configuration management module is configured for an operator to configure operation parameters.

[0025] The virtual counter interaction module is configured to dynamically display commodities, perform multi-round dialogue interaction with the user, identify the user's intention and recommend commodities.

[0026] The attraction calculation module is configured to calculate the attraction score of commodities according to commodity data and generate a list of commodities to be replaced.

[0027] The display effect evaluation module is configured to calculate the display effect score of the counter and make a judgment on batch replacement of commodities.

[0028] The commodity updating module is configured to perform replacement and updating operations of commodities according to the judgment result of the judgment on batch replacement of commodities and the list of commodities to be replaced, or according to a preset display period.

[0029] The above technical solution has at least the following advantages and positive effects compared with the prior art: by accurately positioning the triggering time of the virtual counter to the most intense ordering or renewal page of the user's intention, intervention is performed at the best decision point, and the user's demand is deeply mined in combination with multi-round interaction, which significantly improves the accuracy and success rate of marketing conversion; innovative attraction calculation algorithm (a) and display effect evaluation algorithm (F) are introduced to eliminate inefficient commodities in real time, quantify the display effect of the entire counter and intelligently trigger batch replacement, and ensure that the counter always maintains the best attraction; the combination of the virtual cashier and the dynamic counter provides a personified and interactive shopping experience, which significantly enhances the user's service experience and satisfaction.

[0030] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. BRIEF DESCRIPTION OF DRAWINGS

[0031] The accompanying drawings, which are incorporated herein and constitute part of the specification, illustrate embodiments consistent with the application and, together with the description, further serve to explain the principles of the application. It is to be understood that the drawings are only schematic, and that they do not necessarily correspond to the actual size, or proportions of the embodiments. The same reference numbers in different drawings indicate the same or similar elements.

[0032] Figure 1 A schematic diagram of an AI virtual counter marketing method based on dynamic evaluation of the present application;

[0033] Figure 2 A schematic diagram of a commodity batch replacement process of the present application. DETAILED DESCRIPTION

[0034] In order to more clearly illustrate the objectives, technical solutions and advantages of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The example embodiments can be implemented in various forms, and should not be understood as being limited to the examples described herein; on the contrary, the purpose of providing these embodiments is to enable the ideas of the example embodiments to be fully conveyed to those skilled in the art.

[0035] In addition, the described features, structures or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, many specific details are provided to give a sufficient understanding of the embodiments of the present application. However, those skilled in the art will realize that the technical solutions of the present application can be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. can be used. In other cases, well-known methods, devices, implementations or operations are not shown or described in detail to avoid obscuring the aspects of the present application.

[0036] The block diagrams shown in the drawings are only functional entities, and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0037] The flowcharts shown in the drawings are only exemplary illustrations, and do not necessarily include all contents and operations / steps, nor do they necessarily have to be executed in the order described. For example, some operations / steps can be further divided, and some operations / steps can be combined or partially combined, so the actual execution order can be changed according to the actual situation.

[0038] The application will be described in detail below with reference to specific embodiments. Embodiment 1

[0039] As shown in the figure, the embodiment provides an AI virtual counter marketing method based on dynamic evaluation, taking the "benefit mall" application scenario as an example. The specific steps of the method are as follows: Figure 1

[0040] S1: Construct a marketing function on the server and access an online business system through a standardized interface, and receive running parameters configured through a console.

[0041] In this embodiment, the marketing function is specifically a virtual counter marketing tool, which is an intelligent marketing software system and service platform specially designed for benefit malls and similar business scenarios. It seamlessly accesses existing business systems through predefined standardized interfaces and protocols to form a marketing capability platform that can be uniformly managed and executed. The core of the tool includes an integrated management console and an algorithm engine. The operation personnel can intuitively configure user targeting, display rules and algorithm parameters through the console interface to customize the interaction and recommendation strategy of the virtual counter. As the carrier entity of the method, the tool encapsulates data processing, algorithm calculation, decision-making and front-end display functions into an automated process, which is the necessary technical basis and operating environment for implementing the method. In this embodiment, a virtual counter marketing tool is constructed on the server, and a benefit mall is accessed through a predefined standardized interface.

[0042] Further, the operation personnel configures parameters, sets user configuration, basic configuration and advanced configuration through a special management console interface, thereby customizing the behavior strategy of the virtual counter. The basic configuration is the core running parameter, which is the core engine parameter for controlling the behavior of the virtual counter and the algorithm decision. The advanced configuration is used to fine-tune the interactive experience and additional functions of the front-end user.

[0043] ​For user configuration, the target user of the virtual counter can be precisely defined, and basic attribute filters such as user name, user ID, registration time, last login time, user activity level, historical order quantity, number of added shopping carts, and the like can be configured; and advanced condition filters such as "ordered goods include: [drop-down selected goods]", "non-ordered goods include: [drop-down selected goods]", "logged in within [ ] days", "registration time before / after [date]", "users who have ordered [goods name] recently", and the like. Among them, "ordered goods include: [drop-down selected goods]" is cross-selling to users who have purchased certain goods; "non-ordered goods include: [drop-down selected goods]" is promotion to users who have not purchased certain goods; "logged in within [ ] days" targets recent active users; "registration time before / after [date]" targets new and old users with different strategies; "users who have ordered [goods name] recently" targets users who have just completed an order for additional sales. A user filtering interface is provided to support combined queries and filters based on one or more of user basic attributes, ordered goods, and recent login time. Different counter content is displayed for different user groups with different characteristics to achieve precise marketing.

[0044] For basic configuration, virtual cashier image, counter display goods quantity, display period, attention threshold (F threshold), single product replacement threshold (a threshold), and single replacement goods quantity can be configured. The virtual cashier image is an AI digital person image, and the clarity can be adjusted according to the server load; the counter display goods quantity sets the number of goods displayed by the counter at a time; the display period sets a fixed time interval (such as 10 minutes, 30 minutes, or 1 hour) to force refresh the counter goods at regular intervals; the attention threshold (F threshold) sets a critical value for the counter display effect score F, and when the calculated F value is lower than the threshold, the goods are replaced in batches; the single product replacement threshold (a threshold) sets a critical value for the product attractiveness score a, and when the a value of a certain product is lower than the threshold, the product is marked as "to be replaced"; and the single replacement goods quantity determines the number of goods replaced at a time when the goods are replaced in batches.

[0045] For advanced configuration, voice conversion and mode switching can be configured; voice conversion controls whether to convert the text reply of the virtual cashier into voice and broadcast to the user; mode switching can switch the console color theme, adjust the font size and contrast of the front-end page, and configure the supported language list.

[0046] S2: When the target user enters the order page, the virtual counter interaction interface is triggered and popped up.

[0047] In this embodiment, when the user enters the subscription page, a smart "virtual counter interaction interface" is automatically triggered and popped up, which combines "anthropomorphic dialogue" (virtual clerk) and "visual product display" (dynamic product display counter) to provide precise marketing services for users in the best way at the best time, thereby greatly improving user experience and marketing efficiency.

[0048] For the subscription page, including the first subscription page and the renewal page, it is a key conversion node page with specific business attributes and user intentions. When the user reaches this page, it indicates that he has completed the selection and comparison of goods and other pre-decision processes, has a clear purchase or renewal intention, and is in the last stage before the transaction is completed. At this time, the user's attention is highly concentrated, and the psychological acceptance of related recommendations is higher. The subscription page is to let the user confirm and complete the subscription or renewal operation, usually containing order summary, payment method selection, final confirmation button and other elements. The first subscription page is the confirmation page when the user purchases a certain benefit product (such as a membership card or a package service) for the first time. The renewal page is the page that guides the user to perform the renewal operation when the user's existing benefit service is about to expire or has expired. Triggering the virtual counter on this page can greatly improve the conversion rate and success rate of marketing.

[0049] Further, the virtual counter interaction interface is a composite and intelligent interaction interface, including virtual clerk interaction and dynamic product display counter. For virtual clerk interaction, it is an AI digital human interaction interface with multi-round dialogue, intent recognition and semantic understanding capabilities. The interaction content can include mining user needs, such as initiating greetings ("Hello, I see you are subscribing to X service, do you need to know about the recent popular Y service?") or answering user questions; understanding user potential needs based on dialogue content, and making precise and personalized product recommendations; answering user questions about platform rules, product details, etc. Virtual clerk interaction builds trust through anthropomorphic interaction, deeply mines user needs, and guides users to focus on recommended products.

[0050] Specifically, the virtual teller multi-round interaction, the virtual teller interacts with the user through natural language processing (NLP) technology in multiple rounds of dialogue, actively excavates user demand in the dialogue, identifies user intent and recommends goods, and finally guides to accurate recommendation and transaction of benefit goods. First, intent recognition, the initial question of the user input is quickly classified, whether the current dialogue of the user is in the core business scene of "goods recommendation" or other (such as consultation, complaint, etc.); receive user text or voice (converted to text by ASR) input, use NLP technology to extract core words from the question; for example, the user asks: "What is the recent good movie discount?" The system will extract the keywords "movie" and "discount"; match the extracted keywords with the pre-set "intent library"; the intent library is usually pre-configured by the operation personnel, and contains a large number of business-related scenes; determine whether the current dialogue intent belongs to the recommendation scene or the non-recommendation scene. Then semantic understanding, if the current dialogue intent is not a recommendation scene, call the existing knowledge base (FAQ) or business process engine to directly answer the user's specific question or guide the user to complete the process; if it is a recommendation scene, start multi-round question and answer interaction.

[0051] Specifically, starting multi-round question and answer interaction, further extracting more specific "entity" information from the user's initial question; since the user's initial demand is often vague, further questioning is initiated to confirm and supplement key information to build a complete user portrait; the entity information confirmed in each round of dialogue will be transmitted as real-time input parameters to the recommendation algorithm in the back end; the recommendation algorithm dynamically filters and sorts the most matched goods list from the goods library according to the increasingly clear user portrait; this dynamically generated recommendation list will refresh the "dynamic goods display counter" in real time; the user does not need to jump out of the current dialogue interface to directly add to the shopping cart or place an order in bulk for the recommended goods displayed in the counter.

[0052] For the dynamic goods display counter, it is a visual and operable goods recommendation area that can display multiple (for example, 8-15) recommended goods in grid, list or 3D shelf form side by side, each goods module contains picture, name, key benefit, price and other core information. Among them, the displayed goods are not fixed, but are dynamically updated in real time or quasi-real time according to algorithm calculation (attractiveness score a, counter display effect F), to ensure that the displayed goods are the most attractive ones; the user can directly click to view details or one-click bulk add to shopping cart / order on the counter interface.

[0053] S3: Collecting goods data and calculating the attractiveness score of the goods, forming a list of goods to be replaced based on the attractiveness score.

[0054] In this embodiment, the core user behavior data of each commodity in a specific time period is collected in real time or at a fixed time, including the number of displays, clicks and purchases of the commodity; the number of displays represents the number of times the commodity is displayed to the user in the counter, the number of clicks represents the number of times the user clicks to view the details of the commodity, and the number of purchases represents the number of times the user finally purchases the commodity; the data collection is realized through the embedding technology, and the front end (virtual counter interface) sends a log record to the server when the user performs the corresponding behavior; the back-end service aggregates these logs to form the historical and real-time data stream of each commodity.

[0055] Further, the commodity attraction score is calculated based on the collected commodity data to objectively measure the comprehensive attraction of a commodity to users in the near future. The calculation formula is:

[0056] ;

[0057] α is the attraction score; w y is the weight factor of time period y, which is a parameter pre-configured by the operator according to the marketing calendar. During important promotional festivals (such as Double 11), w y is set to 1.6, and during ordinary promotional days, w y is set to 1.2). The click and purchase behaviors of all commodities are given higher weights, so that the best-selling products during festivals can continue to be retained in the counter for a period of time after the festival, extending their popularity; is the basic conversion rate, is the comprehensive (click and purchase) of user positive behavior, C y represents the number of commodity clicks in time period y; B y represents the number of commodity purchases in time period y; S y represents the number of commodity displays in time period y; is the time decay factor, which amplifies the influence of recent data and weakens the influence of long-term data. t is the current time, and σ is the decay coefficient, which is configured by the operator. The larger the value, the faster the forgetting speed. The value of σ ranges from 0.1 to 0.5, and is configured to use a larger decay coefficient value for high-frequency users and a smaller decay coefficient value for low-frequency users.

[0058] Further, according to the commodity attraction score, it is determined whether to replace the corresponding commodity, and the attraction score α of all commodities is compared with the single product replacement threshold pre-set by the operator in the console. The attraction scores α of all commodities are sorted from low to high; replace the low-attraction commodity (α is lower than the threshold), and automatically mark it as "to be replaced" to generate a list of commodities to be replaced; further, when batch replacement of commodities needs to be performed subsequently, the N commodities with the lowest attraction score α ranking are replaced preferentially.

[0059] S4: Based on the attractiveness score and the real-time display status of the products, evaluate the counter display effect to determine the batch replacement of products.

[0060] Counter display effectiveness evaluation refers to assessing the overall performance of the current product mix at the counter; the real-time display status of the products. , indicating whether product i is displayed in the counter at a specific time t, is represented by a binary function (0 or 1). Based on the attractiveness score calculated in step S3 and the real-time display status of the product, the counter display effect score F is calculated using the following formula:

[0061] ;

[0062] in, This means that at any time t, all currently displayed ( The attractiveness score of the product. The sum of these values ​​represents the instantaneous total attractiveness of all items on the counter at that moment. The logarithmic scores of the instantaneous attractiveness calculated at each discrete time point over a period of time (from t0 to t1) are summed together to assess the sustained performance over the entire period.

[0063] Furthermore, the calculated counter display effect score F is compared with the attention threshold set in the control panel beforehand; if F is greater than or equal to the threshold, it means that the overall performance of the current counter's product mix is ​​good and no batch update is required; if F is less than the threshold, batch replacement of products is triggered.

[0064] S5: Update the products displayed in the counter based on the determination result of the batch replacement of the products and the list of products to be replaced.

[0065] like Figure 2 As shown, the triggering conditions for batch product replacement include: the counter display effect score F is less than a threshold, and a forced refresh is performed according to a pre-configured display cycle (e.g., every 10 minutes). When batch product replacement is triggered, all products in the current counter are sorted from lowest to highest based on their attractiveness score, generating a list of products to be replaced. Based on the operationally configured number of products to be replaced per batch (N), the N products with the lowest attractiveness scores in this list are selected for batch replacement. New products are selected from the activity product library, prioritizing new products in the same category as the replaced products. Within this category, they are sorted by the highest pageviews or other operational metrics, and the top N new products are selected. To avoid interfering with the user's current browsing and operation, and to prevent full-page refreshes, front-end partial refresh technologies (such as Ajax or WebSocket) can be used to update the page. For entirely new products, since there is no historical data, their initial attractiveness score is 0 or very low. They need to be manually added to the counter by operations personnel or replaced through a fixed display cycle refresh mechanism to gain display opportunities and accumulate data.

[0066] Further, a new round of monitoring and decision-making cycle is started, and the commodity data is continuously collected and the commodity attraction calculation, display effect evaluation and commodity update are repeatedly performed, so as to realize continuous self-adaptive optimization of the commodity display strategy.

[0067] It should be noted that the method described in the application is not only applicable to the benefit mall, but also can be applied to any online business platform that needs to carry out commodity recommendation and marketing, such as a digital content subscription platform, an online member service system, a software as a service (SaaS) product renewal scene, etc. Embodiment 2

[0068] This embodiment exemplarily presents an AI virtual counter marketing system based on dynamic evaluation, which comprises a configuration management module, a virtual counter interaction module, an attraction calculation module, a display effect evaluation module and a commodity update module.

[0069] The configuration management module is configured for an operator to configure operation parameters; the operation parameters include but are not limited to user configuration, counter display commodity quantity, display period, attention threshold, single product replacement threshold and single replacement commodity quantity; the user configuration includes providing a user filtering interface, supporting combined query and filtering according to one or more conditions of user basic attributes, ordered commodity conditions and recent login time.

[0070] The virtual counter interaction module is configured to dynamically display commodities to target users, interact with users in multiple rounds of dialogue, recognize user intentions and recommend commodities.

[0071] The attraction calculation module is configured to calculate the attraction score of the commodity according to the commodity data, and generate a replacement commodity list; the calculation formula of the attraction score is as follows:

[0072] ;

[0073] α is the attraction score, w y represents a weight factor in the time period y, C y represents the number of commodity clicks in the time period y, B y represents the number of commodity purchases in the time period y, S y represents the number of commodity displays in the time period y, and σ is a decay coefficient and t is the current time; wherein w y and σ are preset parameters. The attraction score of the commodity is compared with the preset single product replacement threshold, if the attraction score is lower than the single product replacement threshold, the commodity is marked as a replacement commodity, and the attraction score is sorted to form a replacement commodity list.

[0074] The display effect evaluation module is configured to calculate a counter display effect score and make a judgment on whether to replace the goods in bulk. The counter display effect evaluation includes calculating the counter display effect score.

[0075] ;

[0076] F is the counter display effect score, n is the total number of goods in the counter, is the attraction score of the good i at time t, is a binary function indicating whether the good i is displayed at time t; the counter display effect score is compared with a preset attention threshold value, and if the counter display effect score is less than the preset attention threshold value, the replacement of the goods in bulk is triggered.

[0077] The goods updating module replaces and updates the goods according to the judgment result of the judgment on whether to replace the goods in bulk and a list of goods to be replaced, or according to a preset display cycle.

[0078] Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. It is intended that the specification and examples be considered as exemplary only, with the true scope and spirit of the application being indicated by the following claims. It is understood that the application is not limited to the precise structures and methods described herein and that various modifications can be made therein without departing from the scope of the application. The scope of the application is limited only by the claims that follow.

Claims

1. An AI-powered virtual counter marketing method based on dynamic evaluation, characterized in that, The method includes: Build marketing functions on the server side and connect to the online business system through standardized interfaces to receive the running parameters configured through the console; When the target user enters the order page, a virtual counter interaction interface is triggered and pops up; Collect product data and calculate product attractiveness scores, then generate a list of products to be replaced based on these attractiveness scores; the formula for calculating the attractiveness score is: ； α is the attraction score, w y C represents the weighting factor within the time period y. y B represents the number of product clicks within the time period y. y S represents the number of times a product is purchased within a time period y. y This represents the number of times a product is displayed within a time period y, where σ is the decay coefficient and t is the current time; where w y And σ are preset parameters; Based on the attractiveness score and the real-time display status of the products, a counter display effect evaluation is conducted to determine the batch replacement of products; the counter display effect evaluation includes calculating the counter display effect score: ； F represents the score for the display effect of the counter, and n represents the total number of products in the counter. Let i be the attractiveness score of product i at time t. This is a binary function representing whether product i is displayed at time t; Based on the determination result of the batch replacement of goods and the list of goods to be replaced, the goods displayed in the counter are updated.

2. The AI ​​virtual counter marketing method based on dynamic evaluation according to claim 1, characterized in that, The operating parameters include user configuration, number of products displayed on the counter, display period, attention threshold, single product replacement threshold, and number of products replaced in a single transaction.

3. The AI ​​virtual counter marketing method based on dynamic evaluation according to claim 2, characterized in that, The user configuration includes providing a user filtering interface that supports combined queries and filtering based on one or more of the following conditions: user basic attributes, ordered products, and most recent login time.

4. The AI ​​virtual counter marketing method based on dynamic evaluation according to claim 1, characterized in that, The attractiveness score of a product is compared with a preset single-item replacement threshold. If the attractiveness score is lower than the single-item replacement threshold, the product is marked as a product to be replaced, and the products are sorted according to the attractiveness score to form a list of products to be replaced.

5. The AI ​​virtual counter marketing method based on dynamic evaluation according to claim 1, characterized in that, The counter display effect score is compared with a preset attention threshold. If the counter display effect score is less than the preset attention threshold, a batch replacement of products is triggered.

6. The AI ​​virtual counter marketing method based on dynamic evaluation according to claim 1, characterized in that, The updating of the products displayed in the counter also includes periodically refreshing the products in the counter according to a preset display cycle.

7. The AI ​​virtual counter marketing method based on dynamic evaluation according to claim 1, characterized in that, The virtual counter interface includes a virtual teller and a dynamic product display counter; the virtual teller uses natural language processing technology to engage in multi-round dialogue with the user, identify the user's intent, and recommend products.

8. An AI virtual counter marketing system based on dynamic evaluation, used to implement the method as described in any one of claims 1-7, characterized in that, It includes a configuration management module, a virtual counter interaction module, an attractiveness calculation module, a display effect evaluation module, and a product update module; The configuration management module is used by operators to configure operating parameters; The virtual counter interaction module is used to dynamically display products, engage in multi-round dialogue with users, identify user intent, and recommend products. The attraction calculation module is used to calculate the attraction score of a product based on product data and generate a list of products to be replaced. The display effect evaluation module is used to calculate the counter display effect score and determine the batch replacement of goods; The product update module performs product replacement and update operations based on the determination result of the batch product replacement judgment and the list of products to be replaced, or according to a preset display cycle.

Images