Intelligent recommendation management system of e-commerce platform

Abstract

The invention discloses an intelligent recommendation management system of an e-commerce platform, which relates to the technical field of intelligent recommendation management of e-commerce platforms and comprises a recommendation data acquisition module, a server, a comparison module and a prompt module. And the recommendation data acquisition module is used for acquiring numeric data information and performance information when the intelligent recommendation management system of the e-commerce platform performs related content recommendation, processing the numeric data information and the performance information after acquisition, and transmitting the processed information to the server. According to the method, the intelligent recommendation management system of the e-commerce platform monitors the strategies during related content recommendation, when the system performs related content recommendation and the recommendation strategies may have hidden dangers, related personnel of the e-commerce platform are timely notified to know, and the strategies for performing related content recommendation by the system are timely adjusted, so that the system is more convenient to use. And the user is effectively prevented from losing interest in the platform, so that the loss of the user and the stagnation of product and service sales are effectively prevented.

Description

Technical Field

[0001] This invention relates to the field of intelligent recommendation management technology for e-commerce platforms, and more specifically to an intelligent recommendation management system for e-commerce platforms. Background Technology

[0002] An intelligent recommendation management system for e-commerce platforms utilizes artificial intelligence technology to automatically recommend products and services that best match a user's interests and needs based on their personal preferences, behavioral data, and other relevant information. This system aims to improve user experience, increase sales, and foster user loyalty to the platform. Intelligent recommendation management systems for e-commerce platforms can significantly enhance user experience, helping users find products that match their interests more quickly, thereby increasing platform activity and sales.

[0003] The existing technology has the following shortcomings:

[0004] When an e-commerce platform's intelligent recommendation management system recommends relevant content based on users' personal preferences, behavioral data, and other relevant information, if the system's recommendation strategy malfunctions (recommended content does not meet expectations, duplicate recommendations, etc.) and the relevant personnel of the e-commerce platform fail to detect the problem, users may eventually lose interest in the platform. This will not only lead to user churn but also cause stagnation in the sales of products and services.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide an intelligent recommendation management system for e-commerce platforms. This invention monitors the strategies used by the intelligent recommendation management system of e-commerce platforms when recommending relevant content. When there are potential problems with the recommendation strategies, the invention promptly notifies the relevant personnel of the e-commerce platform and adjusts the strategies accordingly. This effectively prevents users from losing interest in the platform, thereby preventing user churn and stagnation of product and service sales, thus solving the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: an intelligent recommendation management system for an e-commerce platform, comprising a recommendation data collection module, a server, a comparison module, and a prompting module;

[0008] The recommendation data acquisition module collects numerical data and performance information from the intelligent recommendation management system of the e-commerce platform when recommending relevant content. After collection, the numerical data and performance information are processed and transmitted to the server.

[0009] The server comprehensively analyzes the numerical data and performance information processed by the system when recommending relevant content, generates a risk index, and transmits the risk index to the comparison module.

[0010] The comparison module compares and analyzes the hazard index generated by the system when recommending relevant content with the pre-set hazard index reference threshold, generates a hazard signal, and transmits the signal to the alert module, which then issues a hazard warning.

[0011] Preferably, the numerical data information used by the intelligent recommendation management system of the e-commerce platform when recommending relevant content includes a low conversion rate index and an average purchase frequency index. After collection, the recommendation data collection module calibrates the low conversion rate index and the average purchase frequency index as follows: The performance information of the intelligent recommendation management system of the e-commerce platform when recommending relevant content includes the time delay abnormal change index. After collection, the recommendation data collection module calibrates the time delay abnormal change index as μμ.

[0012] The preferred logic for obtaining the low conversion rate index is as follows:

[0013] A101. Obtain the historical best conversion rate range when the system recommends relevant content, and define the historical best conversion rate range as...

[0014] A102. Obtain the actual conversion rate of the system during different time periods at time T when recommending relevant content, and label the actual conversion rate as... y represents the number of the actual conversion rate at different time periods of time T when the system recommends relevant content, y = 1, 2, 3, 4, ..., n, where n is a positive integer;

[0015] A103, will be less than the historical best conversion rate range Minimum value The actual conversion rate is calibrated as x indicates a conversion rate less than the historical best range. Minimum value The actual conversion rate is assigned a number, x = 1, 2, 3, 4, ..., N, where N is a positive integer;

[0016] A104, Through and actual conversion rate The conversion rate slump index is calculated using the following expression: In the formula, n represents the total number of actual conversion rates obtained by the system at time T when recommending relevant content.

[0017] The preferred logic for obtaining the average purchase frequency index is as follows:

[0018] B101. Obtain the number of user purchases and the total number of times the system recommends content to users within time T when the system makes relevant content recommendations;

[0019] B102. Let P and Q be the number of times each user makes a purchase and the total number of times the system recommends content to the user within time T, respectively. Calculate the purchase frequency for each user. The expression for the calculation is: δ = P / Q.

[0020] B103. Obtain the purchase frequency of each user within time T when the system recommends relevant content, and recalibrate the purchase frequency as δ. k k is the user ID used by the system to recommend relevant content, k = 1, 2, 3, 4, ..., m, where m is a positive integer;

[0021] B104. When the system recommends relevant content, the purchase frequency δ of each user within time T. k The average purchase frequency index is calculated using the following expression: In the formula, This represents the average purchase frequency of all users when the system recommends relevant content.

[0022] Preferably, the logic for obtaining the time delay anomaly index is as follows:

[0023] C101. Obtain several actual time delays generated within time T when the system recommends relevant content, and calibrate the actual time delays as μ. f f represents the number of the actual time delay generated within time T when the system makes relevant content recommendations, f = 1, 2, 3, 4, ..., p, where p is a positive integer;

[0024] C102. Calculate the standard deviation of the actual time delay generated within time T when the system recommends relevant content, and denote the standard deviation of the actual time delay as L. Then:

[0025]

[0026] ;in, The average actual time delay generated within time T when the system recommends relevant content is calculated using the following formula:

[0027] C103. Calculate the time delay anomaly index using the actual time delay standard deviation L. The expression for the calculation is: μ = L * ln(L)2 +1).

[0028] Preferably, the server obtains a low conversion rate index. After establishing the average purchase frequency index δδ and the time delay anomaly variation index μμ, a data analysis model is built to generate the hidden danger index θ. YH The formula used is:

[0029]

[0030] In the formula, r1, r2, and r3 are the indices of low conversion rate. The preset proportional coefficients of the average purchase frequency index δδ and the time delay abnormal change index μμ, and r1, r2 and r3 are all greater than 0.

[0031] Preferably, the comparison module compares and analyzes the hazard index generated when the system recommends relevant content with a pre-set hazard index reference threshold. If the hazard index is greater than or equal to the hazard index reference threshold, a high hazard signal is generated through the comparison module and transmitted to the prompting module, which then issues a hazard warning. If the hazard index is less than the hazard index reference threshold, a low hazard signal is generated through the comparison module and transmitted to the prompting module, but no hazard warning is issued through the prompting module.

[0032] Preferably, it also includes a recommendation strategy feedback module;

[0033] The recommendation strategy adjustment feedback module comprehensively analyzes several potential risk indices output by the server when the system recommends relevant content, and judges the overall situation after the system adjusts the recommendation strategy.

[0034] When the recommendation strategy adjustment feedback module recommends relevant content to the system, it establishes an analysis set based on several hidden danger indices output by the server, and labels the analysis set as W. Then, W = {θ} YH y}, where y represents the index number of the hidden danger index within the analysis set, y = 1, 2, 3, 4, ..., q, and q is a positive integer;

[0035] Calculate the mean and standard deviation of the hazard index based on the hazard index within the analysis set, and denote the mean and standard deviation of the hazard index as G. θ1 and G θ2 ,but: but:

[0036] The average value of the hazard index G θ1 and the standard deviation of the hazard index G θ2Each is compared with the pre-set hazard index reference threshold F θ1 and standard deviation reference threshold F θ2 The comparative analysis yielded the following results:

[0037] If G θ1 ≥F θ1 If the recommendation strategy adjustment fails, a signal indicating that the recommendation strategy adjustment has failed will be generated through the recommendation strategy adjustment feedback module and transmitted to the mobile terminal. The mobile terminal will then notify the relevant personnel of the e-commerce platform that the recommendation strategy adjustment has failed and that further adjustments to the recommendation strategy are needed.

[0038] like The recommendation strategy adjustment feedback module generates a signal indicating that the recommendation strategy adjustment is unstable and transmits the signal to the mobile terminal. The mobile terminal then notifies the relevant personnel of the e-commerce platform that the recommendation strategy adjustment is unstable and requires further adjustment.

[0039] like The recommendation strategy adjustment feedback module then generates a signal indicating that the recommendation strategy adjustment has been successful and transmits the signal to the mobile terminal, which then notifies the relevant personnel on the e-commerce platform that the recommendation strategy adjustment has been successful.

[0040] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0041] This invention monitors the content recommendation strategies of an e-commerce platform's intelligent recommendation management system. When the system's recommendation strategy may have potential problems, it promptly notifies the relevant personnel of the e-commerce platform to adjust the system's content recommendation strategy in a timely manner. This effectively prevents users from losing interest in the platform, thereby effectively preventing user churn and stagnation of product and service sales.

[0042] This invention comprehensively analyzes the output results of the system when recommending relevant content after adjusting the strategy, and judges the situation of the system when recommending relevant content after adjusting the strategy. In this way, problems such as failure or instability of the recommendation strategy adjustment can be detected in time. By continuously adjusting the recommendation strategy, the system can be kept in a stable and low-risk state when recommending relevant content, which can further effectively prevent users from losing interest in the platform, thereby further effectively preventing user churn and stagnation of product and service sales. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0044] Figure 1 This is a schematic diagram of the modules of an intelligent recommendation management system for an e-commerce platform according to the present invention. Detailed Implementation

[0045] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0046] This invention provides, for example Figure 1 The intelligent recommendation management system for an e-commerce platform shown includes a recommendation data collection module, a server, a comparison module, and a prompt module.

[0047] The recommendation data acquisition module collects numerical data and performance information from the intelligent recommendation management system of the e-commerce platform when recommending relevant content. After collection, the numerical data and performance information are processed and transmitted to the server.

[0048] The intelligent recommendation management system of e-commerce platforms uses numerical data information, including a low conversion rate index and an average purchase frequency index, when recommending relevant content. After collection, the recommendation data collection module calibrates the low conversion rate index and the average purchase frequency index as follows: and δδ;

[0049] When the conversion rate of recommendations made by the intelligent recommendation management system of an e-commerce platform is low, it may mean that there are some abnormal issues with the content recommended by the system. Here are some possible abnormal issues:

[0050] Recommended content does not match user interests: Recommended content may not match the user's interests, needs, or preferences, causing the user to lose interest and thus reducing the conversion rate;

[0051] Insufficient relevance of recommended content: Recommended content may have a weak correlation with users' historical behavior and preferences, resulting in low user interest in the recommended content and thus affecting conversion rates;

[0052] Recommended product or service quality issues: The recommended product or service may have quality issues, causing users to have doubts about purchasing, thereby reducing the conversion rate;

[0053] Poor performance of recommendation algorithms: Recommendation algorithms may fail to accurately predict user behavior, resulting in low-quality recommended content and thus affecting conversion rates;

[0054] Inappropriate timing of recommendations: The timing of the recommended content may be inappropriate, such as recommending it at a time when users are unlikely to convert, thus affecting the conversion rate;

[0055] The recommended content is not attractive enough: The recommended content may fail to attract the user's attention and lacks sufficient appeal, making the user unwilling to convert.

[0056] Cumbersome purchase process: If the purchase process is complicated and lengthy, users may give up halfway, which will affect the conversion rate.

[0057] Pricing issues: If the recommended products or services are priced unreasonably, it may deter users and affect conversion rates;

[0058] Therefore, monitoring the conversion rate of content recommendations made by the intelligent recommendation management system of e-commerce platforms can promptly detect abnormal issues such as low conversion rates of the recommended content.

[0059] The logic behind obtaining the low conversion rate index is as follows:

[0060] A101. Obtain the historical best conversion rate range when the system recommends relevant content, and define the historical best conversion rate range as...

[0061] It should be noted that A / B testing is a commonly used method that can apply different recommendation strategies to different user groups and then compare the conversion rates and other key indicators between different groups. By comparing the control group and the experimental group, it can be determined which recommendation strategy works better in a specific situation, thereby determining the historical best conversion rate range when the system recommends relevant content. The historical best conversion rate range when the system recommends relevant content is not specifically limited here and can be adjusted according to actual needs.

[0062] A102. Obtain the actual conversion rate of the system when recommending relevant content at different time periods (the time periods can be the same, different, or a combination of both; no specific limitation is made here) at time T, and label the actual conversion rate as... y represents the number of the actual conversion rate at different time periods of time T when the system recommends relevant content, y = 1, 2, 3, 4, ..., n, where n is a positive integer;

[0063] When an e-commerce platform's intelligent recommendation management system recommends content, the conversion rate measures the percentage of users who ultimately complete a purchase or other key action after clicking on a piece of content through the recommendation system. The conversion rate is usually expressed as a percentage and is calculated as follows:

[0064]

[0065] Among them, the number of users who actually completed a purchase or action refers to the number of users who actually took key actions such as purchasing, subscribing, or registering after receiving the recommended content, and the number of users who clicked on the content refers to the number of users who clicked on the recommended content.

[0066] For example: Suppose a recommendation system recommends a product to 100 users, 20 of whom click on the product, and ultimately 4 users purchase the product through those clicks. The conversion rate of this system would be calculated as follows:

[0067]

[0068] It should be noted that data analytics tools, such as Google Analytics and Adobe Analytics, can be used to monitor user behavior on the platform, including clicks and purchases. These tools typically provide reports on key metrics such as conversion rates.

[0069] A103, will be less than the historical best conversion rate range Minimum value The actual conversion rate is calibrated as x indicates a conversion rate less than the historical best range. Minimum value The actual conversion rate is assigned a number, x = 1, 2, 3, 4, ..., N, where N is a positive integer;

[0070] A104, Through and actual conversion rate The conversion rate slump index is calculated using the following expression: In the formula, n represents the total number of actual conversion rates obtained by the system at time T when recommending relevant content;

[0071] As can be seen from the calculation formula of the conversion rate depression index, the larger the conversion rate depression index generated by the intelligent recommendation management system of the e-commerce platform in time T when recommending relevant content, the greater the potential for anomalies in the system when recommending relevant content; conversely, the smaller the conversion rate depression index, the smaller the potential for anomalies in the system when recommending relevant content.

[0072] A low average purchase frequency recommended by the intelligent recommendation management system of an e-commerce platform may indicate some anomalies in the system. The following are some anomalies that could lead to a low purchase frequency:

[0073] Recommended content mismatch: The recommended content may not match the user's interests, needs, or purchase intentions, causing the user to lose interest in the recommended content and thus reduce the purchase frequency;

[0074] Recommendation algorithm failure: The recommendation algorithm may fail to accurately analyze user behavior and interests, resulting in low-quality recommended content and low user purchase interest;

[0075] Poor timing of recommendations: The timing of the recommended content may be inappropriate, such as recommending it at a time when users are unlikely to make a purchase, thereby affecting the purchase frequency;

[0076] Unappealing Recommended Content: The recommended content may lack sufficient appeal, thus discouraging users from purchasing it.

[0077] Cumbersome purchase process: If the purchase process is too complicated and lengthy, users may abandon the purchase halfway, thus affecting the frequency of purchase;

[0078] Difficulty in making a purchase decision: The recommended content may not provide enough information or comparisons, making it difficult for users to make a purchase decision;

[0079] Therefore, by monitoring the average purchase frequency of the intelligent recommendation management system of e-commerce platforms when recommending relevant content, abnormal issues such as low average purchase frequency of the recommended content can be detected in a timely manner.

[0080] The logic for obtaining the average purchase frequency index is as follows:

[0081] B101. Obtain the number of user purchases and the total number of times the system recommends content to users within time T when the system makes relevant content recommendations;

[0082] It should be noted that data analytics tools such as Google Analytics and Adobe Analytics can be used to monitor user behavior, including purchasing behavior and click behavior on recommended content, thereby calculating the number of purchases and the total number of recommendations;

[0083] B102. Let P and Q be the number of times each user makes a purchase and the total number of times the system recommends content to the user within time T, respectively. Calculate the purchase frequency for each user. The expression for the calculation is: δ = P / Q.

[0084] B103. Obtain the purchase frequency of each user within time T when the system recommends relevant content, and recalibrate the purchase frequency as δ. k k is the user ID used by the system to recommend relevant content, k = 1, 2, 3, 4, ..., m, where m is a positive integer;

[0085] B104. When the system recommends relevant content, the purchase frequency δ of each user within time T. k The average purchase frequency index is calculated using the following expression: In the formula, This represents the average purchase frequency of all users when the system recommends relevant content.

[0086] As can be seen from the calculation formula of the average purchase frequency index, the higher the average purchase frequency index generated by the intelligent recommendation management system of the e-commerce platform in time T when recommending relevant content, the smaller the potential for anomalies in the system when recommending relevant content; conversely, the lower the average purchase frequency index, the greater the potential for anomalies in the system when recommending relevant content.

[0087] The performance information of the intelligent recommendation management system of the e-commerce platform when recommending relevant content includes the time delay abnormal change index. After collection, the recommendation data collection module calibrates the time delay abnormal change index as μμ.

[0088] Time delay refers to the time interval between a user performing an action (such as clicking on recommended content or browsing products) and the generation of relevant recommended content by the recommendation system. In the intelligent recommendation management system of an e-commerce platform, time delay refers to how long it takes for the recommendation system to generate and display recommended content related to the user's behavior after the user interacts with the platform. When the time delay stability of the intelligent recommendation management system of an e-commerce platform is poor, the following abnormal problems may occur:

[0089] Inaccurate recommendations: If the recommendation system generates recommended content in a short period of time, there may not be enough time to analyze user behavior and interests. As a result, the recommended content may not be accurate and may not meet the user's expectations.

[0090] Outdated recommendations: If the system's time delay fluctuates significantly, it can sometimes cause recommended content to lag behind, which may make the recommended content outdated and no longer match the user's current interests.

[0091] Missed business opportunities: The unstable time delay of recommended content may cause users to miss some potential purchase opportunities. If users are waiting for the recommendation delay, they may miss products or promotions that they are interested in.

[0092] Inconsistent user experience: Users may experience different recommended content at different times, which can cause confusion when using the platform and reduce the user experience.

[0093] Decreased recommendation performance: Fluctuations in time delay may affect the overall performance of the recommendation system, leading to a decrease in the quality and relevance of the recommended content;

[0094] Purchase decisions are affected: Unstable time delays may make it difficult for users to make purchase decisions because they cannot be sure when they will receive the appropriate recommended content;

[0095] Decreased user satisfaction: Users may be dissatisfied with the quality and timeliness of the recommended content, thus reducing their overall satisfaction with the platform;

[0096] The logic for obtaining the time delay anomaly index is as follows:

[0097] C101. Obtain several actual time delays generated within time T when the system recommends relevant content, and calibrate the actual time delays as μ. f f represents the number of the actual time delay generated within time T when the system makes relevant content recommendations, f = 1, 2, 3, 4, ..., p, where p is a positive integer;

[0098] It should be noted that the system usually records user behavior, such as browsing, searching, purchasing, and system-recommended content. By analyzing these logs, we can understand the timestamp of each recommendation request and the timestamp of the actual recommended content, thereby calculating the actual time delay of the recommended content.

[0099] C102. Calculate the standard deviation of the actual time delay generated within time T when the system recommends relevant content, and denote the standard deviation of the actual time delay as L. Then:

[0100]

[0101] ;in, The average actual time delay generated within time T when the system recommends relevant content is calculated using the following formula:

[0102] It should be noted that the larger the standard deviation L of the actual time delay generated by the system within time T when recommending relevant content, the greater the fluctuation and the worse the stability of the several actual time delays generated by the system within time T when recommending relevant content; conversely, the smaller the fluctuation and the better the stability of the several actual time delays generated by the system within time T when recommending relevant content.

[0103] C103. Calculate the time delay anomaly index using the actual time delay standard deviation L. The expression for the calculation is: μ = L * ln(L) 2 +1);

[0104] As can be seen from the calculation formula of the time delay anomaly variation index, the larger the time delay anomaly variation index generated within time T when the intelligent recommendation management system of the e-commerce platform recommends relevant content, the greater the potential for anomalies when the system recommends relevant content; conversely, the smaller the time delay anomaly variation index, the smaller the potential for anomalies when the system recommends relevant content.

[0105] The server comprehensively analyzes the numerical data and performance information processed by the system when recommending relevant content, generates a risk index, and transmits the risk index to the comparison module.

[0106] The server obtained a low conversion rate index. After establishing the average purchase frequency index δδ and the time delay anomaly variation index μμ, a data analysis model is built to generate the hidden danger index θ. YH The formula used is:

[0107]

[0108] In the formula, r1, r2, and r3 are the indices of low conversion rate. The preset proportional coefficients of the average purchase frequency index δδ and the time delay abnormal change index μμ, and r1, r2 and r3 are all greater than 0;

[0109] As can be seen from the calculation formula, the larger the conversion rate depression index, the smaller the average purchase frequency index, and the larger the time delay abnormal fluctuation index generated by the intelligent recommendation management system of the e-commerce platform within time T when recommending relevant content, the greater the potential risk index θ generated by the system within time T. YH The higher the performance value, the greater the potential for anomalies when the system recommends relevant content. For e-commerce platforms, a smaller conversion rate depression index, a larger average purchase frequency index, and a smaller time delay anomaly fluctuation index generated within time T when the intelligent recommendation management system recommends relevant content indicates a greater potential risk index θ. YH The smaller the performance value, the lower the potential for anomalies when the system recommends relevant content;

[0110] The comparison module compares and analyzes the hazard index generated by the system when recommending relevant content with the pre-set hazard index reference threshold, generates a hazard signal, and transmits the signal to the prompting module, which then issues a hazard warning.

[0111] The comparison module compares the potential risk index generated when the system recommends relevant content with a pre-set potential risk index reference threshold. If the potential risk index is greater than or equal to the potential risk index reference threshold, a high potential risk signal is generated through the comparison module and transmitted to the alert module. The alert module then issues a potential risk alert, prompting relevant personnel on the e-commerce platform to promptly identify potential risks in the system's recommendation strategy and adjust the strategy accordingly. If the potential risk index is less than the potential risk index reference threshold, a low potential risk signal is generated through the comparison module and transmitted to the alert module. No potential risk alert is issued through the alert module.

[0112] It should be noted that the above-mentioned time T is a relatively short time period. The time within this period is not specifically limited and can be set according to the actual situation. The purpose is to monitor the situation of the system when making relevant content recommendations within time T, so as to monitor the situation of the system when making relevant content recommendations in different time periods (within time T).

[0113] It also includes a recommendation strategy feedback module;

[0114] The recommendation strategy adjustment feedback module comprehensively analyzes several potential risk indices output by the server when the system recommends relevant content, and judges the overall situation after the system adjusts the recommendation strategy.

[0115] When the recommendation strategy adjustment feedback module recommends relevant content to the system, it establishes an analysis set based on several hidden danger indices output by the server, and labels the analysis set as W. Then, W = {θ} YH y}, where y represents the index number of the hidden danger index within the analysis set, y = 1, 2, 3, 4, ..., q, and q is a positive integer;

[0116] Calculate the mean and standard deviation of the hazard index based on the hazard index within the analysis set, and denote the mean and standard deviation of the hazard index as G. θ1 and G θ2 ,but: but:

[0117] The average value of the hazard index G θ1 and the standard deviation of the hazard index G θ2 Each is compared with the pre-set hazard index reference threshold F θ1 and standard deviation reference threshold F θ2 The comparative analysis yielded the following results:

[0118] If G θ1 ≥Fθ1 If the recommendation strategy adjustment fails, a signal indicating that the recommendation strategy adjustment has failed will be generated through the recommendation strategy adjustment feedback module and transmitted to the mobile terminal. The mobile terminal will then notify the relevant personnel of the e-commerce platform that the recommendation strategy adjustment has failed and that further adjustments to the recommendation strategy are needed.

[0119] like The recommendation strategy adjustment feedback module generates a signal indicating that the recommendation strategy adjustment is unstable, and transmits the signal to the mobile terminal. The mobile terminal then notifies the relevant personnel of the e-commerce platform that the recommendation strategy adjustment is unstable. The occurrence of unstable recommendation strategy adjustment indicates that after the strategy adjustment, the system cannot maintain a stable and low-risk state when recommending relevant content, and further adjustment of the recommendation strategy is required.

[0120] like The recommendation strategy adjustment feedback module generates a signal indicating that the recommendation strategy adjustment has been successful and transmits the signal to the mobile terminal. The mobile terminal then notifies the relevant personnel of the e-commerce platform that the recommendation strategy adjustment has been successful. By maintaining this strategy, the system can keep the relevant content recommendations in a stable and low-risk state.

[0121] This invention monitors the content recommendation strategies of an e-commerce platform's intelligent recommendation management system. When the system's recommendation strategy may have potential problems, it promptly notifies the relevant personnel of the e-commerce platform to adjust the system's content recommendation strategy in a timely manner. This effectively prevents users from losing interest in the platform, thereby effectively preventing user churn and stagnation of product and service sales.

[0122] This invention comprehensively analyzes the output results of the system when recommending relevant content after adjusting the strategy, and judges the situation of the system when recommending relevant content after adjusting the strategy. In this way, problems such as failure or instability of the recommendation strategy adjustment can be detected in time. By continuously adjusting the recommendation strategy, the system can be kept in a stable and low-risk state when recommending relevant content, which can further effectively prevent users from losing interest in the platform, thereby further effectively preventing user churn and stagnation of product and service sales.

[0123] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0124] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0125] It should be understood that, in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0126] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0127] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0128] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0129] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0130] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. An intelligent recommendation management system for an e-commerce platform, characterized in that, It includes a recommendation data collection module, a server, a comparison module, and a prompting module; The recommendation data acquisition module collects numerical data and performance information from the intelligent recommendation management system of the e-commerce platform when recommending relevant content. After collection, the numerical data and performance information are processed and transmitted to the server. The server comprehensively analyzes the numerical data and performance information processed by the system when recommending relevant content, generates a risk index, and transmits the risk index to the comparison module. The comparison module compares and analyzes the hazard index generated when the system recommends relevant content with the pre-set hazard index reference threshold, generates a hazard signal, and transmits the signal to the alert module, which then issues a hazard warning.

2. The intelligent recommendation management system for an e-commerce platform according to claim 1, characterized in that, The intelligent recommendation management system of e-commerce platforms uses numerical data, including a low conversion rate index and an average purchase frequency index, when recommending relevant content. After collection, the recommendation data collection module calibrates the low conversion rate index and the average purchase frequency index as follows: The performance information of the intelligent recommendation management system of the e-commerce platform when recommending relevant content includes the time delay abnormal change index. After collection, the recommendation data collection module calibrates the time delay abnormal change index as μμ.

3. The intelligent recommendation management system for an e-commerce platform according to claim 2, characterized in that, The logic behind obtaining the low conversion rate index is as follows: A101. Obtain the historical best conversion rate range when the system recommends relevant content, and define the historical best conversion rate range as... A102. Obtain the actual conversion rate of the system during different time periods at time T when recommending relevant content, and label the actual conversion rate as... y represents the number of the actual conversion rate at different time periods of time T when the system recommends relevant content, y = 1, 2, 3, 4, ..., n, where n is a positive integer; A103, will be less than the historical best conversion rate range Minimum value The actual conversion rate is calibrated as x indicates a conversion rate less than the historical best range. Minimum value The actual conversion rate is assigned a number, x = 1, 2, 3, 4, ..., N, where N is a positive integer; A104, Through and actual conversion rate The conversion rate slump index is calculated using the following expression: In the formula, n represents the total number of actual conversion rates obtained by the system at time T when recommending relevant content.

4. The intelligent recommendation management system for an e-commerce platform according to claim 3, characterized in that, The logic for obtaining the average purchase frequency index is as follows: B101. Obtain the number of user purchases and the total number of times the system recommends content to users within time T when the system makes relevant content recommendations; B102. Let P and Q be the number of times each user makes a purchase and the total number of times the system recommends content to the user within time T, respectively. Calculate the purchase frequency for each user. The expression for the calculation is: δ = P / Q. B103, the purchase frequency of each user in T time when the system makes relevant content recommendation, and recalibrate the purchase frequency to δ k , k the user number when the system makes relevant content recommendation, k = 1, 2, 3, 4, …, m, m is a positive integer; B104. When the system recommends relevant content, the purchase frequency δ of each user within time T. k The average purchase frequency index is calculated using the following expression: In the formula, This represents the average purchase frequency of all users when the system recommends relevant content.

5. The intelligent recommendation management system for an e-commerce platform according to claim 4, characterized in that, The logic for obtaining the time delay anomaly index is as follows: C101, a plurality of actual time delays generated within T time when the system makes relevant content recommendation, and the actual time delay is marked as μ f f represents the number of actual time delays generated within T time when the system makes relevant content recommendation, f = 1, 2, 3, 4, …, p, p is a positive integer; C102. Calculate the standard deviation of the actual time delay generated within time T when the system recommends relevant content, and denote the standard deviation of the actual time delay as L. Then: ； in, The average actual time delay generated within time T when the system recommends relevant content is calculated using the following formula: C103. Calculate the time delay anomaly index using the actual time delay standard deviation L. The expression for the calculation is: μ = L * ln(L) 2 +1).

6. The intelligent recommendation management system for an e-commerce platform according to claim 5, characterized in that, The server obtained a low conversion rate index. After establishing the average purchase frequency index δδ and the time delay anomaly variation index μμ, a data analysis model is built to generate the hidden danger index θ. YH The formula used is: In the formula, r1, r2, and r3 are the indices of low conversion rate. The preset proportional coefficients of the average purchase frequency index δδ and the time delay abnormal change index μμ, and r1, r2 and r3 are all greater than 0.

7. The intelligent recommendation management system for an e-commerce platform according to claim 6, characterized in that, The comparison module compares the hazard index generated when the system recommends relevant content with a pre-set hazard index reference threshold. If the hazard index is greater than or equal to the hazard index reference threshold, a high hazard signal is generated through the comparison module and transmitted to the prompting module, which then issues a hazard warning. If the hazard index is less than the hazard index reference threshold, a low hazard signal is generated through the comparison module and transmitted to the prompting module, but no hazard warning is issued.

8. The intelligent recommendation management system for an e-commerce platform according to claim 7, characterized in that, It also includes a recommendation strategy feedback module; The recommendation strategy adjustment feedback module comprehensively analyzes several potential risk indices output by the server when the system recommends relevant content, and judges the overall situation after the system adjusts the recommendation strategy. When the recommendation strategy adjustment feedback module recommends relevant content to the system, it establishes an analysis set based on several hidden danger indices output by the server, and labels the analysis set as W. Then, W = {θ} YH y }, where y represents the index number of the hidden danger index within the analysis set, y = 1, 2, 3, 4, ..., q, and q is a positive integer; Calculate the mean and standard deviation of the hazard index based on the hazard index within the analysis set, and denote the mean and standard deviation of the hazard index as G. θ1 and G θ2 ,but: but: The average value of the hazard index G θ1 and the standard deviation of the hazard index G θ2 Each is compared with the pre-set hazard index reference threshold F θ1 and standard deviation reference threshold F θ2 The comparative analysis yielded the following results: If G θ1 ≥F θ1 If the recommendation strategy adjustment fails, a signal indicating that the recommendation strategy adjustment has failed will be generated through the recommendation strategy adjustment feedback module and transmitted to the mobile terminal. The mobile terminal will then notify the relevant personnel of the e-commerce platform that the recommendation strategy adjustment has failed and that further adjustments to the recommendation strategy are needed. like The recommendation strategy adjustment feedback module generates a signal indicating that the recommendation strategy adjustment is unstable and transmits the signal to the mobile terminal. The mobile terminal then notifies the relevant personnel of the e-commerce platform that the recommendation strategy adjustment is unstable and requires further adjustment. like The recommendation strategy adjustment feedback module then generates a signal indicating that the recommendation strategy adjustment has been successful and transmits the signal to the mobile terminal, which then notifies the relevant personnel on the e-commerce platform that the recommendation strategy adjustment has been successful.

Images