Commodity sales platform recommendation method and system based on data analysis

Abstract

The invention relates to the technical field of commodity recommendation management, in particular to a commodity sales platform recommendation method and system based on data analysis. According to the method, the user demand is determined according to the historical data of the user so as to realize personalized commodity recommendation on the commodity sales platform, the commodity categories of the recommended personalized commodities are counted, and the interaction rate, the instant interaction rate and the negative feedback rate of each commodity category in the preset time are periodically detected; and judging whether the recommendation of the commodity sales platform meets the standard or not based on the instant interaction rate, judging whether the recommendation of the commodity sales platform meets the standard or not based on the negative feedback rate, performing corresponding adjustment according to the similarity diffusion coefficient of the clicked commodities, the category exploration flow proportion and the negative feedback penalty weight coefficient, and giving out corresponding alarms under different conditions. According to the invention, real-time monitoring and intelligent adjustment of commodity recommendation are effectively realized, and the recommendation efficiency of the commodity sales platform is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of product recommendation management technology, and in particular to a product sales platform recommendation method and system based on data analysis. Background Technology

[0002] A product sales platform is a digital marketplace or store that connects sellers and buyers, providing a range of functions such as product display, transactions, payment, and logistics tracking to facilitate online sales. A product recommendation system is a key software subsystem within this platform. Its core task is to automatically filter and present products that each user might be interested in from a vast array of goods, thereby improving user discovery efficiency, shopping experience, and platform transaction conversion rates. Data-driven product sales platform recommendations refer to collecting, processing, and analyzing large amounts of user and product data, using algorithms to predict user preferences, and achieving personalized recommendations. Traditional product recommendation systems suffer from a lack of personalization, reliance on static rules, coarse recommendation results, and difficulty in real-time updates. With the development of the times, research on data-driven product sales platform recommendations has significant practical implications for improving user experience and industry development.

[0003] Chinese Patent Publication No. CN113283960A discloses a method and cloud service platform for intelligent product recommendation on a vertical e-commerce platform based on big data analysis and cloud computing. This method includes: obtaining basic information about the user; counting the number of times the user has purchased alcoholic beverages in the past; obtaining alcoholic beverage information corresponding to each of the user's historical purchases on the e-commerce platform; obtaining basic information and sales information of each alcoholic beverage seller on the e-commerce platform; matching and filtering the user's preferred alcoholic beverage information with the basic information of each alcoholic beverage seller; comparing the alcoholic beverage sellers to be recommended and making recommendations.

[0004] Therefore, the above solution effectively addresses the problem of existing product recommendation methods not analyzing users' historical shopping records, achieving intelligent recommendations for beverages based on the user's preferences. However, this solution cannot perform real-time monitoring and intelligent adjustment of product recommendations, thus failing to guarantee the recommendation efficiency of the product sales platform. Summary of the Invention

[0005] To address this issue, the present invention provides a product sales platform recommendation method and system based on data analysis, which overcomes the problem of low recommendation efficiency in existing technologies due to the inability to monitor and intelligently adjust product recommendations in real time.

[0006] On one hand, the present invention provides a recommendation method for a product sales platform based on data analysis, comprising: Determine user needs based on historical user data to enable personalized product recommendations on the product sales platform; The system statistically analyzes the product categories of the personalized recommended products and periodically monitors the real-time interaction rate and negative feedback rate of each product category within a preset time period. The recommendation of the product sales platform is determined based on the real-time interaction rate to determine whether it meets the standards. Based on the real-time interaction rate determination result, the platform determines whether the product sales platform's recommendations meet the standards based on the negative feedback rate, or generates corresponding processing instructions. Based on the negative feedback rate determination result, periodic determinations are made on the commodity sales platform, or corresponding processing instructions are generated; Adjust the similarity diffusion coefficient of clicked products based on the real-time interaction rate difference ratio, adjust the category exploration traffic ratio based on the interaction rate difference, adjust the negative feedback penalty weight coefficient based on the negative feedback rate difference ratio, issue a potential product discovery stagnation alarm, issue a category attractiveness lack alarm, and issue a product quality risk alarm. The similarity diffusion coefficient of the clicked product, the category exploration traffic ratio, and the negative feedback penalty weight coefficient are stored.

[0007] Furthermore, the process of determining whether the product sales platform's recommendations meet the standards based on the real-time interaction rate includes: The number of clicks on recommended products of the same category within the preset time period is counted, and the obtained number is recorded as the number of clicks on recommended products of that category; Determine the total number of times recommended products of the same category are displayed within the detection period, and record the total number as the total number of times recommended products of that category are displayed; Calculate the ratio of the number of clicks on the recommended product to the total number of times the recommended product in the corresponding category is displayed, and record the obtained ratio as the real-time interaction rate; When the instant interaction rate is greater than or equal to the preset instant interaction rate, the recommendation of the product sales platform is determined to meet the standard based on the negative feedback rate. When the instant interaction rate is less than the preset instant interaction rate, it is determined that the product sales platform's recommendation does not meet the standard, and the reason why the product sales platform's recommendation does not meet the standard is determined based on the click behavior.

[0008] Furthermore, the process of determining whether the product sales platform's recommendations meet the standards based on the negative feedback rate includes: The number of negative feedback behaviors for the same product category within the specified testing period is counted. Calculate the ratio of the number of negative feedback behaviors to the total number of times the recommended products are displayed in the corresponding category, and record the ratio as the negative feedback rate of the category; When the negative feedback rate is less than or equal to the preset negative feedback rate, it is determined that the recommendations of the product sales platform meet the standard, thus completing the periodic assessment of the product sales platform, and determining whether the product sales platform meets the standard in the next period; When the negative feedback rate is greater than the preset negative feedback rate, it is determined that the recommendation of the product sales platform does not meet the standard, and the negative feedback penalty weight coefficient is adjusted based on the difference in negative feedback rates.

[0009] Furthermore, the process of determining why the product sales platform's recommendations do not meet the standards based on the click behavior includes: The click behavior of recommended products in categories where the real-time interaction rate is less than the preset real-time interaction rate is determined; When the click condition exists, the similarity diffusion coefficient of the clicked product is adjusted based on the instant interaction rate difference ratio; When the click condition is not present, the category exploration traffic ratio is adjusted based on the interaction rate difference.

[0010] Furthermore, the process of increasing the similarity diffusion coefficient of the clicked product based on the instant interaction rate difference ratio includes: Calculate the difference between the preset real-time interaction rate and the real-time interaction rate, and record the obtained difference as the interaction rate difference; Calculate the ratio of the interaction rate difference to the preset real-time interaction rate and record the obtained ratio as the real-time interaction rate difference ratio; The similarity diffusion coefficient of the clicked product is increased based on the real-time interaction rate difference ratio, and the increase in the similarity diffusion coefficient of the clicked product is proportional to the real-time interaction rate difference ratio.

[0011] Furthermore, the process of increasing the proportion of category exploration traffic based on the interaction rate difference includes: Determine the minimum value within the interval formed by the interaction rate difference and the upper limit of the flow adjustment, and record the minimum value as the increase value; Calculate the sum of the category exploration traffic ratio and the increase value, and record the sum as the secondary category exploration traffic ratio; Increase the proportion of traffic for the category exploration to the proportion of traffic for the secondary category exploration.

[0012] Furthermore, after the increase in the clicked product similarity diffusion coefficient and the category exploration traffic ratio, the process of determining whether the product sales platform's recommendation meets the standard based on the real-time interaction rate includes: When the instant interaction rate is greater than or equal to the preset instant interaction rate, the recommendation of the product sales platform is determined to meet the standard based on the negative feedback rate. When the instant interaction rate is less than the preset instant interaction rate after the similarity diffusion coefficient of the clicked product has increased, a potential product discovery stall alarm is issued. When the instant interaction rate is less than the preset instant interaction rate after the proportion of category exploration traffic has increased, an alarm for lack of category attractiveness is issued.

[0013] Furthermore, the process of increasing the negative feedback penalty weight coefficient based on the negative feedback rate difference ratio includes: Calculate the difference between the negative feedback rate and the preset negative feedback rate, and record the obtained difference as the negative feedback rate difference; Calculate the ratio of the negative feedback rate difference to the preset negative feedback rate, and record the obtained ratio as the negative feedback rate difference ratio; The negative feedback penalty weight coefficient is increased based on the negative feedback rate difference ratio, and the increase in the negative feedback penalty weight coefficient is proportional to the negative feedback rate difference ratio.

[0014] Furthermore, the process of determining whether the product sales platform's recommendations meet the standards based on the negative feedback rate after the negative feedback penalty weight coefficient has been increased includes: When the negative feedback rate is less than or equal to the preset negative feedback rate, it is determined that the recommendations of the product sales platform meet the standard, thus completing the periodic assessment of the product sales platform, and determining whether the product sales platform meets the standard in the next period; When the negative feedback rate is greater than the preset negative feedback rate, it is determined that the product sales platform's recommendation does not meet the standard, and a product quality risk alarm is issued.

[0015] On the other hand, the present invention also provides a data analysis-based product sales platform recommendation system using the above-described method, comprising: The display module is used to determine user needs based on historical user data in order to achieve personalized product recommendations on the product sales platform; An analysis module, connected to the display module, is used to statistically analyze the product categories of the recommended personalized products and periodically detect the instant interaction rate and negative feedback rate of each product category within the preset time period. The analysis module is also used to determine whether the recommendations of the product sales platform meet the standards based on the real-time interaction rate; The analysis module is also used to determine whether the recommendations of the product sales platform meet the standards based on the negative feedback rate according to the real-time interaction rate determination result, or to generate corresponding processing instructions. The analysis module is also used to make periodic judgments on the commodity sales platform based on the negative feedback rate judgment results, or to generate corresponding processing instructions. The adjustment modules are respectively connected to the display module and the analysis module, and are used to adjust the similarity diffusion coefficient of the clicked product based on the real-time interaction rate difference ratio, adjust the category exploration traffic ratio based on the interaction rate difference, adjust the negative feedback penalty weight coefficient based on the negative feedback rate difference ratio, issue a potential product mining stagnation alarm, issue a category attractiveness lack alarm, and issue a product quality risk alarm. The strategy configuration center, which is connected to the display module and the adjustment module respectively, is used to store the similarity diffusion coefficient of the clicked product, the category exploration traffic ratio, and the negative feedback penalty weight coefficient.

[0016] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention determines whether the recommendations of a product sales platform meet the standards based on the real-time interaction rate and the negative feedback rate. It can timely and accurately determine whether the recommendations of a product sales platform meet the standards, effectively realizes rapid and accurate analysis of whether the recommended products of the product sales platform meet user needs, effectively realizes real-time monitoring of product recommendations, and adjusts the similarity diffusion coefficient of clicked products, the proportion of category exploration traffic, and the negative feedback penalty weight coefficient, issuing corresponding alarms under different circumstances. While effectively realizing intelligent adjustment of product recommendations, it effectively improves the recommendation efficiency of the product sales platform.

[0017] Furthermore, this invention determines whether the recommendations of a product sales platform meet the standards based on the real-time interaction rate, accurately identifying the reasons why the platform's recommendations do not meet the standards, whether the negative feedback rate is needed to determine this, or whether the click behavior is needed to determine why the platform's recommendations do not meet the standards. This avoids misjudgments and further enables real-time monitoring of product recommendations, thereby improving the recommendation efficiency of the product sales platform.

[0018] Furthermore, this invention determines whether the recommendations of a product sales platform meet the standards based on the negative feedback rate, accurately identifies whether the product sales platform's recommendations for product categories meet the standards, and promptly determines whether the negative feedback penalty weight coefficient needs to be adjusted based on the difference in negative feedback rates, ensuring the system's recommendation accuracy. It further enables real-time monitoring of product recommendations, thereby further improving the recommendation efficiency of the product sales platform.

[0019] Furthermore, this invention determines the reasons why product recommendations on a sales platform do not meet standards based on click behavior, accurately identifies whether the similarity diffusion coefficient of clicked products needs to be adjusted based on the real-time interaction rate difference ratio or whether the category exploration traffic ratio needs to be adjusted based on the interaction rate difference, avoids misjudgment, and further realizes real-time monitoring of product recommendations, thereby further improving the recommendation efficiency of the sales platform.

[0020] Furthermore, this invention increases the similarity diffusion coefficient of clicked products based on the real-time interaction rate difference ratio, effectively avoiding situations where the product sales platform's recommendations do not meet the standards due to the similarity diffusion coefficient of clicked products not meeting the standards. This effectively ensures that the product sales platform's recommendations meet user needs, further realizing intelligent adjustment of product recommendations while further improving the recommendation efficiency of the product sales platform.

[0021] Furthermore, this invention increases the proportion of category exploration traffic based on the interaction rate difference, effectively avoiding situations where the product sales platform's recommendations do not meet the standards due to the proportion of category exploration traffic not meeting the standards. This effectively prevents category recommendations from failing to meet the standards, and while further realizing intelligent adjustment of product recommendations, it further improves the recommendation efficiency of the product sales platform.

[0022] Furthermore, after the increase in the similarity diffusion coefficient of clicked products and the proportion of category exploration traffic, this invention determines whether the recommendations of the product sales platform meet the standards based on the real-time interaction rate. It can judge the adjustment effect and promptly determine whether to issue an alarm for potential product discovery stagnation or an alarm for lack of category attractiveness, avoiding misjudgment. This further enables real-time monitoring of product recommendations and improves the recommendation efficiency of the product sales platform.

[0023] Furthermore, this invention increases the negative feedback penalty weight coefficient based on the negative feedback rate difference ratio, effectively avoiding situations where the product sales platform's recommendations do not meet the standards due to the negative feedback penalty weight coefficient not meeting the standards. This effectively ensures that the product sales platform's recommendations meet user needs, further realizing intelligent adjustment of product recommendations while improving the recommendation efficiency of the product sales platform.

[0024] Furthermore, after the negative feedback penalty weight coefficient is increased, the present invention determines whether the recommendations of the product sales platform meet the standards based on the negative feedback rate. It can judge the effect of increasing the negative feedback penalty weight coefficient and determine in a timely manner whether a product quality risk alarm should be issued. This further realizes real-time monitoring of product recommendations and further improves the recommendation efficiency of the product sales platform.

[0025] Furthermore, by setting up a display module, an analysis module, an adjustment module, and a strategy configuration center, and connecting these modules into a system, the present invention effectively ensures the efficiency of the data analysis-based product sales platform recommendation method. While further realizing real-time monitoring and intelligent adjustment of product recommendations, it further improves the recommendation efficiency of the product sales platform. Attached Figure Description

[0026] Figure 1This is a structural block diagram of a product sales platform recommendation system based on data analysis, according to an embodiment of the present invention. Figure 2 This is a flowchart of a product sales platform recommendation method based on data analysis, as described in an embodiment of the present invention. Figure 3 This is a flowchart illustrating how an embodiment of the present invention determines whether a product sales platform's recommendations meet the standards and the reasons for determining whether they do not meet the standards; Figure 4 This is a flowchart illustrating how an embodiment of the present invention determines why a product sales platform's recommendations do not meet the standards. Detailed Implementation

[0027] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0028] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0029] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0030] Please see Figure 1 The diagram shown is a structural block diagram of a product sales platform recommendation system based on data analysis, according to an embodiment of the present invention. The structure of this embodiment includes a display module, an analysis module, an adjustment module, and a strategy configuration center; wherein, The display module is used to determine user needs based on user historical data in order to realize personalized product recommendations on the product sales platform; The analysis module is connected to the display module and is used to statistically analyze the product categories of the recommended personalized products and periodically detect the interaction rate, real-time interaction rate and negative feedback rate of each product category within a preset time. The analysis module is also used to determine whether the recommendations of the product sales platform meet the standards based on the real-time interaction rate; The analysis module is also used to determine whether the recommendations of the product sales platform meet the standards based on the negative feedback rate according to the real-time interaction rate determination result, or to generate corresponding processing instructions. The analysis module is also used to make periodic judgments on the commodity sales platform based on the negative feedback rate judgment results, or to generate corresponding processing instructions. The adjustment module is connected to the display module and the analysis module respectively, and is used to adjust the similarity diffusion coefficient of clicked products based on the real-time interaction rate difference ratio, adjust the category exploration traffic ratio based on the interaction rate difference, adjust the negative feedback penalty weight coefficient based on the negative feedback rate difference ratio, issue a potential product mining stagnation alarm, issue a category attractiveness lack alarm, and issue a product quality risk alarm. The strategy configuration center is connected to the display module and the adjustment module respectively, and is used to store the similarity diffusion coefficient of the clicked product, the category exploration traffic ratio, and the negative feedback penalty weight coefficient.

[0031] Please see Figure 2 The diagram shown is a flowchart of a product sales platform recommendation method based on data analysis, according to an embodiment of the present invention. The method described in this embodiment includes: Determine user needs based on historical user data to enable personalized product recommendations on the product sales platform; The system statistically analyzes the product categories of the recommended personalized products and periodically detects the instant interaction rate and negative feedback rate of each product category within the preset time period. The recommendation of the product sales platform is determined based on the real-time interaction rate to determine whether it meets the standards. Based on the real-time interaction rate determination result, the platform determines whether the product sales platform's recommendations meet the standards based on the negative feedback rate, or generates corresponding processing instructions. Based on the negative feedback rate determination result, periodic determinations are made on the commodity sales platform, or corresponding processing instructions are generated; Adjust the similarity diffusion coefficient of the clicked product based on the real-time interaction rate difference ratio, adjust the category exploration traffic ratio based on the interaction rate difference, adjust the negative feedback penalty weight coefficient based on the negative feedback rate difference ratio, issue a potential product discovery stagnation alarm, issue a category attractiveness lack alarm, and issue a product quality risk alarm. The similarity diffusion coefficient of the clicked product, the category exploration traffic ratio, and the negative feedback penalty weight coefficient are stored.

[0032] Please see Figure 3 The diagram shows a flowchart illustrating how an embodiment of the present invention determines whether a product sales platform's recommendations meet the standards and the reasons for any non-compliance. The process by which this embodiment determines whether a product sales platform's recommendations meet the standards based on the real-time interaction rate includes: The number of clicks on recommended products of the same category within the preset time T is counted, and the obtained number is recorded as the number of clicks on recommended products of that category; Determine the total number of times recommended products of the same category are displayed within the detection period, and record the total number as the total number of times recommended products of that category are displayed; Calculate the ratio of the number of clicks on the recommended product to the total number of times the recommended product in the corresponding category is displayed, and record the obtained ratio as the real-time interaction rate; When the instant interaction rate is greater than or equal to the preset instant interaction rate N, the recommendation of the product sales platform is determined to meet the standard based on the negative feedback rate. When the instant interaction rate is less than the preset instant interaction rate N, it is determined that the recommendation of the product sales platform does not meet the standard, and the reason why the recommendation of the product sales platform does not meet the standard is determined based on the click behavior. Specifically, in this embodiment, the preset time T=5s, where 5s is an industry experience threshold, and the detection cycle is 15min; The process of determining the preset real-time interaction rate N for the same product category includes: Statistically analyze several recommended products of the same category within a historical time period U, determine the instant interaction rate of each recommended product of the category in turn, sort the obtained instant interaction rates from low to high, determine the value at the 75th position and record this value as the instant interaction rate baseline, calculate the product of the instant interaction rate baseline and the instant interaction rate weight coefficient A, and record the obtained product as the preset instant interaction rate. The preset real-time interaction rate for each product category is determined sequentially by following the steps described above for determining the preset real-time interaction rate for the same product category. Wherein, the historical time U = 7 days, and the weighting coefficient of the instant interaction rate A = 0.8, where 0.8 is an empirical threshold; It should be noted that the weighting coefficient of the instant interaction rate is applicable to the determination of the preset instant interaction rate for each product category; This embodiment includes categories such as dresses, sweatshirts, coats, pants, snacks, and toiletries. The meaning of several items in the same category is that dresses include dresses of different styles, sweatshirts include sweatshirts of different styles, coats include coats of different styles, pants include pants of different styles, snacks include snacks of different styles, and toiletries include toiletries of different styles. In this embodiment, the preset real-time interaction rate N1 for dresses is 3.2%, the preset real-time interaction rate N2 for sweatshirts is 2.8%, the preset real-time interaction rate N3 for coats is 2.4%, the preset real-time interaction rate N4 for pants is 2.0%, the preset real-time interaction rate N5 for snacks is 4.8%, and the preset real-time interaction rate N6 for toiletries is 1.6%.

[0033] When comparing the instant interaction rate with the preset instant interaction rate, the category of the instant interaction rate is determined, and the instant interaction rate is compared with the preset instant interaction rate of the category. That is, the instant interaction rate is compared with the category corresponding to the preset instant interaction rate N of the category. The categories of the preset instant interaction rate N include the preset instant interaction rate N1, the preset instant interaction rate N2, the preset instant interaction rate N3, the preset instant interaction rate N4, the preset instant interaction rate N5, and the preset instant interaction rate N6.

[0034] Please continue reading. Figure 3 As shown, the process of determining whether the recommendations of a product sales platform meet the standards based on the negative feedback rate in this embodiment of the invention includes: The number of negative feedback behaviors for the same product category within the specified testing period is counted. Calculate the ratio of the number of negative feedback behaviors to the total number of times the recommended products are displayed in the corresponding category, and record the ratio as the negative feedback rate of the category; When the negative feedback rate is less than or equal to the preset negative feedback rate F, it is determined that the recommendations of the product sales platform meet the standard, thus completing the periodic assessment of the product sales platform, and determining whether the product sales platform meets the standard in the next period; When the negative feedback rate is greater than the preset negative feedback rate F, it is determined that the recommendation of the product sales platform does not meet the standard, and the negative feedback penalty weight coefficient is adjusted based on the difference in negative feedback rates.

[0035] Specifically, the number of negative feedback behaviors mentioned in this embodiment includes the sum of the number of times the user clicked "not interested," the number of times the user clicked "block the product," and the number of times the exposure time was less than the preset exposure time Y. The preset exposure duration Y = 500ms, where 500ms is an industry-experienced threshold. The process of determining the preset negative feedback rate F for the same product category includes: Statistically analyze several recommended products of the same category within the historical time U, determine the negative feedback rate of each recommended product of the category in turn, sort the obtained negative feedback rates from low to high, determine the value at the 75th position and record this value as the negative feedback rate baseline, calculate the product of the negative feedback rate baseline and the negative feedback rate weight coefficient B, and record the obtained product as the preset negative feedback rate. The predetermined negative feedback rate for each product category is determined sequentially by following the steps described above for determining the predetermined negative feedback rate for the same product category. Wherein, the negative feedback rate weighting coefficient B = 1.2, and 1.2 is an empirical threshold; It should be noted that the negative feedback rate weighting coefficient is applicable to the determination of the preset negative feedback rate for each product category; In this embodiment, the preset negative feedback rate F1 for dresses is 1.8%, F2 for sweatshirts is 1.44%, F3 for coats is 2.16%, F4 for pants is 2.4%, F5 for snacks is 0.96%, and F6 for toiletries is 1.2%. When comparing the negative feedback rate with the preset negative feedback rate, the category of the negative feedback rate is determined, and the negative feedback rate is compared with the preset negative feedback rate of the category. That is, the negative feedback rate is compared with the category corresponding to the preset negative feedback rate F of the category. The categories of the preset instant interaction rate F include the preset instant interaction rate F1, the preset instant interaction rate F2, the preset instant interaction rate F3, the preset instant interaction rate F4, the preset instant interaction rate F5, and the preset instant interaction rate F6.

[0036] Please continue reading. Figure 3 As shown, the process of determining why the product sales platform's recommendations do not meet the standards based on the click behavior in this embodiment of the invention includes: The click behavior of recommended products in categories where the real-time interaction rate is less than the preset real-time interaction rate is determined; When the click condition exists, the similarity diffusion coefficient of the clicked product is adjusted based on the instant interaction rate difference ratio; When the click condition is absent, adjust the category exploration traffic ratio based on the interaction rate difference; Specifically, the click behavior refers to whether a recommended product within the same product category has been clicked.

[0037] Please see Figure 4 The flowchart shown illustrates the reasons why product recommendations from a sales platform do not meet the standards, as described in this embodiment of the invention. The process of increasing the similarity diffusion coefficient of clicked products based on the real-time interaction rate difference ratio in this embodiment of the invention includes: Calculate the difference between the preset real-time interaction rate and the real-time interaction rate, and record the obtained difference as the interaction rate difference; Calculate the ratio of the interaction rate difference to the preset real-time interaction rate and record the obtained ratio as the real-time interaction rate difference ratio; When the interaction rate difference ratio is less than or equal to the first preset interaction rate difference ratio G1, it is determined that the similarity diffusion coefficient of the clicked product will be increased to 1.21 times the initial similarity diffusion coefficient of the clicked product, wherein, in this embodiment, the first preset interaction rate difference ratio G1 = 20%; When the interaction rate difference ratio is greater than the first preset interaction rate difference ratio G1 and less than or equal to the second preset interaction rate difference ratio G2, it is determined that the similarity diffusion coefficient of the clicked product will be increased to 1.53 times the initial similarity diffusion coefficient of the clicked product, wherein, in this embodiment, the second preset interaction rate difference ratio G2 = 50%; When the interaction rate difference ratio is greater than the second preset interaction rate difference ratio G2, it is determined that the similarity diffusion coefficient of the clicked product will be increased to 1.80 times the initial similarity diffusion coefficient of the clicked product; Specifically, the values ​​of the interaction rate difference ratio and the similarity diffusion coefficient of the clicked product are both derived from the actual debugging results; Increasing the similarity diffusion coefficient of the clicked product can increase the number of other products recommended that are similar to the clicked product in the same category, that is, increase the number of other products recommended that are similar to the clicked product.

[0038] Please continue reading. Figure 4 As shown, the process of increasing the category exploration traffic ratio based on the interaction rate difference in this embodiment of the invention includes: Determine the minimum value within the interval formed by the interaction rate difference and the upper limit of the flow adjustment, and record the minimum value as the increase value; Calculate the sum of the category exploration traffic ratio and the increase value, and record the sum as the secondary category exploration traffic ratio; Increase the proportion of traffic for the aforementioned category exploration to the proportion of traffic for the secondary category exploration. Specifically, the upper limit of the flow rate adjustment described in this embodiment is 30%, which is an empirical threshold. Increasing the proportion of category exploration traffic can increase the recommended traffic for products in that category, thereby increasing the probability that users will see recommended products in that category.

[0039] Please continue reading. Figure 4 As shown, the process of determining whether the product sales platform's recommendation meets the standard based on the real-time interaction rate after the increase of the clicked product similarity diffusion coefficient and the category exploration traffic ratio in this embodiment of the invention includes: When the instant interaction rate is greater than or equal to the preset instant interaction rate N, the recommendation of the product sales platform is determined to meet the standard based on the negative feedback rate. When the instant interaction rate is less than the preset instant interaction rate N after the similarity diffusion coefficient of the clicked product has increased, a potential product discovery stall alarm is issued. When the instant interaction rate is less than the preset instant interaction rate N after the category exploration traffic ratio has increased, an alarm for lack of category attractiveness is issued. Specifically, after the similarity diffusion coefficient of the clicked products and the category exploration traffic ratio are adjusted, the detection period is 2 hours.

[0040] Please continue reading. Figure 4 As shown, the process of increasing the negative feedback penalty weight coefficient based on the negative feedback rate difference ratio in this embodiment of the invention includes: Calculate the difference between the negative feedback rate and the preset negative feedback rate, and record the obtained difference as the negative feedback rate difference; Calculate the ratio of the negative feedback rate difference to the preset negative feedback rate, and record the obtained ratio as the negative feedback rate difference ratio; When the negative feedback rate difference ratio is less than or equal to the first preset negative feedback rate difference ratio R1, it is determined that the negative feedback penalty weight coefficient will be increased to 1.25 times the initial negative feedback penalty weight coefficient. In this embodiment, the first preset negative feedback rate difference ratio R1 = 22%. When the negative feedback rate difference ratio is greater than the first preset negative feedback rate difference ratio R1 and less than or equal to the second preset negative feedback rate difference ratio R2, it is determined that the negative feedback penalty weight coefficient will be increased to 1.56 times the initial negative feedback penalty weight coefficient. In this embodiment, the second preset negative feedback rate difference ratio R2 = 54%. When the negative feedback rate difference ratio is greater than the second preset negative feedback rate difference ratio R2, it is determined that the negative feedback penalty weight coefficient will be increased to 1.88 times the initial negative feedback penalty weight coefficient; specifically, the values ​​of the negative feedback rate difference ratio and the negative feedback penalty weight coefficient are both derived from the actual debugging results; Increasing the negative feedback penalty weighting coefficient can increase the de-weighting intensity of recommended products with a history of negative feedback.

[0041] Please continue reading. Figure 4 As shown, the process of determining whether the recommendations of the product sales platform meet the standards based on the negative feedback rate after the negative feedback penalty weight coefficient is increased in this embodiment of the invention includes: When the negative feedback rate is less than or equal to the preset negative feedback rate F, it is determined that the recommendations of the product sales platform meet the standard, thus completing the periodic assessment of the product sales platform, and determining whether the product sales platform meets the standard in the next period; When the negative feedback rate is greater than the preset negative feedback rate F, it is determined that the recommendation of the product sales platform does not meet the standard and a product quality risk alarm is issued. Specifically, after the negative feedback penalty weight coefficient is adjusted, the detection period is 2 hours.

[0042] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0043] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A recommendation method for a product sales platform based on data analysis, characterized in that, include: Determine user needs based on historical user data to enable personalized product recommendations on the product sales platform; The system statistically analyzes the product categories of the personalized recommended products and periodically monitors the real-time interaction rate and negative feedback rate of each product category within a preset time period. The recommendation of the product sales platform is determined based on the real-time interaction rate to determine whether it meets the standards. Based on the real-time interaction rate determination result, the platform determines whether the product sales platform's recommendations meet the standards based on the negative feedback rate, or generates corresponding processing instructions. Based on the negative feedback rate determination result, periodic determinations are made on the commodity sales platform, or corresponding processing instructions are generated; Adjust the similarity diffusion coefficient of clicked products based on the real-time interaction rate difference ratio, adjust the category exploration traffic ratio based on the interaction rate difference, adjust the negative feedback penalty weight coefficient based on the negative feedback rate difference ratio, issue a potential product discovery stagnation alarm, issue a category attractiveness lack alarm, and issue a product quality risk alarm. The similarity diffusion coefficient of the clicked product, the category exploration traffic ratio, and the negative feedback penalty weight coefficient are stored.

2. The product sales platform recommendation method based on data analysis according to claim 1, characterized in that, The process of determining whether a product sales platform's recommendations meet the standards based on the real-time interaction rate includes: The number of clicks on recommended products of the same category within the preset time period is counted, and the obtained number is recorded as the number of clicks on recommended products of that category; Determine the total number of times recommended products of the same category are displayed within the detection period, and record the total number as the total number of times recommended products of that category are displayed; Calculate the ratio of the number of clicks on the recommended product to the total number of times the recommended product in the corresponding category is displayed, and record the obtained ratio as the real-time interaction rate; When the instant interaction rate is greater than or equal to the preset instant interaction rate, the recommendation of the product sales platform is determined to meet the standard based on the negative feedback rate. When the instant interaction rate is less than the preset instant interaction rate, it is determined that the product sales platform's recommendation does not meet the standard, and the reason why the product sales platform's recommendation does not meet the standard is determined based on the click behavior.

3. The product sales platform recommendation method based on data analysis according to claim 2, characterized in that, The process of determining whether a product sales platform's recommendations meet the standards based on the negative feedback rate includes: The number of negative feedback behaviors for the same product category within the specified testing period is counted. Calculate the ratio of the number of negative feedback behaviors to the total number of times the recommended products are displayed in the corresponding category, and record the ratio as the negative feedback rate of the category; When the negative feedback rate is less than or equal to the preset negative feedback rate, it is determined that the recommendations of the product sales platform meet the standard, thus completing the periodic assessment of the product sales platform, and determining whether the product sales platform meets the standard in the next period; When the negative feedback rate is greater than the preset negative feedback rate, it is determined that the recommendation of the product sales platform does not meet the standard, and the negative feedback penalty weight coefficient is adjusted based on the difference in negative feedback rates.

4. The product sales platform recommendation method based on data analysis according to claim 2, characterized in that, The process of determining why the product sales platform's recommendations do not meet the standards based on the click behavior includes: The click behavior of recommended products in categories where the real-time interaction rate is less than the preset real-time interaction rate is determined; When the click condition exists, the similarity diffusion coefficient of the clicked product is adjusted based on the instant interaction rate difference ratio; When the click condition is not present, the category exploration traffic ratio is adjusted based on the interaction rate difference.

5. The product sales platform recommendation method based on data analysis according to claim 4, characterized in that, The process of increasing the similarity diffusion coefficient of the clicked product based on the real-time interaction rate difference ratio includes: Calculate the difference between the preset real-time interaction rate and the real-time interaction rate, and record the obtained difference as the interaction rate difference; Calculate the ratio of the interaction rate difference to the preset real-time interaction rate and record the obtained ratio as the real-time interaction rate difference ratio; The similarity diffusion coefficient of the clicked product is increased based on the real-time interaction rate difference ratio, and the increase in the similarity diffusion coefficient of the clicked product is proportional to the real-time interaction rate difference ratio.

6. The product sales platform recommendation method based on data analysis according to claim 5, characterized in that, The process of increasing the proportion of category exploration traffic based on the interaction rate difference includes: Determine the minimum value within the interval formed by the interaction rate difference and the upper limit of the flow adjustment, and record the minimum value as the increase value; Calculate the sum of the category exploration traffic ratio and the increase value, and record the sum as the secondary category exploration traffic ratio; Increase the proportion of traffic for the category exploration to the proportion of traffic for the secondary category exploration.

7. The product sales platform recommendation method based on data analysis according to claim 6, characterized in that, After the increase in the clicked product similarity diffusion coefficient and the category exploration traffic ratio is completed, the process of determining whether the product sales platform's recommendations meet the standards based on the real-time interaction rate includes: When the instant interaction rate is greater than or equal to the preset instant interaction rate, the recommendation of the product sales platform is determined to meet the standard based on the negative feedback rate. When the instant interaction rate is less than the preset instant interaction rate after the similarity diffusion coefficient of the clicked product has increased, a potential product discovery stall alarm is issued. When the instant interaction rate is less than the preset instant interaction rate after the category exploration traffic ratio has increased, an alarm for lack of category attractiveness is issued.

8. The product sales platform recommendation method based on data analysis according to claim 3, characterized in that, The process of increasing the negative feedback penalty weight coefficient based on the negative feedback rate difference ratio includes: Calculate the difference between the negative feedback rate and the preset negative feedback rate, and record the obtained difference as the negative feedback rate difference; Calculate the ratio of the negative feedback rate difference to the preset negative feedback rate, and record the obtained ratio as the negative feedback rate difference ratio; The negative feedback penalty weight coefficient is increased based on the negative feedback rate difference ratio, and the increase in the negative feedback penalty weight coefficient is proportional to the negative feedback rate difference ratio.

9. The product sales platform recommendation method based on data analysis according to claim 8, characterized in that, The process of determining whether the product sales platform's recommendations meet the standards based on the negative feedback rate after the negative feedback penalty weight coefficient has been increased includes: When the negative feedback rate is less than or equal to the preset negative feedback rate, it is determined that the recommendations of the product sales platform meet the standard, thus completing the periodic assessment of the product sales platform, and determining whether the product sales platform meets the standard in the next period; When the negative feedback rate is greater than the preset negative feedback rate, it is determined that the product sales platform's recommendation does not meet the standard, and a product quality risk alarm is issued.

10. A product sales platform recommendation system based on data analysis using the method of any one of claims 1-9, characterized in that, include: The display module is used to determine user needs based on historical user data in order to achieve personalized product recommendations on the product sales platform; An analysis module, connected to the display module, is used to statistically analyze the product categories of the recommended personalized products and periodically detect the instant interaction rate and negative feedback rate of each product category within the preset time period. The analysis module is also used to determine whether the recommendations of the product sales platform meet the standards based on the real-time interaction rate; The analysis module is also used to determine whether the recommendations of the product sales platform meet the standards based on the negative feedback rate according to the real-time interaction rate determination result, or to generate corresponding processing instructions. The analysis module is also used to make periodic judgments on the commodity sales platform based on the negative feedback rate judgment results, or to generate corresponding processing instructions. The adjustment modules are respectively connected to the display module and the analysis module, and are used to adjust the similarity diffusion coefficient of the clicked product based on the real-time interaction rate difference ratio, adjust the category exploration traffic ratio based on the interaction rate difference, adjust the negative feedback penalty weight coefficient based on the negative feedback rate difference ratio, issue a potential product mining stagnation alarm, issue a category attractiveness lack alarm, and issue a product quality risk alarm. The strategy configuration center, which is connected to the display module and the adjustment module respectively, is used to store the similarity diffusion coefficient of the clicked product, the category exploration traffic ratio, and the negative feedback penalty weight coefficient.

Images