A data backflow optimization method based on big data

Abstract

The application relates to the technical field of data backflow optimization, and specifically discloses a data backflow optimization method based on big data, which comprises the following steps: S1: obtaining shopping frequency and total transaction amount and calculating a detection period; S2: obtaining the total number of times of browsing a commodity interface and the single browsing duration, calculating the average of the single browsing duration, and correcting the detection period; S3: calculating the discount strength of a user's collected commodities, obtaining the average of the number of times of purchasing the collected commodities and the purchase price, calculating a correction value, and correcting the first period; S4: calculating a recommendation accuracy rate, screening unqualified periods, and resetting the detection period if the proportion of the unqualified periods is too high. The application uses a data analysis algorithm for personalized recommendation, realizes a backflow optimization method based on big data to reduce the influence caused by data backflow delay, and thereby speeds up the data backflow speed of big data, so that higher processing efficiency is obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data backflow optimization, and particularly relates to a data backflow optimization method based on big data. BACKGROUND

[0002] Data backflow based on big data refers to the process of returning data from various sources or processing links to a specific system, platform or application according to certain rules and processes, so as to realize further processing, analysis, integration or optimization of data. In the context of big data, data backflow refers to the process of returning massive and diversified data to a specific data system or business process according to business needs or data analysis results after the data has undergone collection, storage, processing, analysis and other processes, so as to realize data reuse, reanalysis and value mining.

[0003] Through data backflow optimization, an e-commerce platform can deeply understand the shopping behavior, preferences and needs of users. For example, the platform can use data analysis algorithms to make personalized recommendations based on users' historical purchase records, browsing records, favorites and other information. By real-time backflow of user behavior data, the platform can timely adjust the recommendation strategy to provide users with more suitable product recommendations, thereby improving the user's purchase conversion rate and shopping experience.

[0004] In the prior art, the e-commerce platform cannot provide feedback in the first time after data update. The user's living conditions, consumption preferences and other information will change over time, but the data backflow may be delayed, resulting in the platform making recommendations and marketing based on outdated data, which reduces the marketing effect and user satisfaction. Therefore, a backflow optimization method based on big data is needed to reduce the impact of data backflow delay and improve the data backflow optimization level of big data to achieve higher processing efficiency. SUMMARY

[0005] The purpose of the present application is to provide a data backflow optimization method based on big data to solve the above technical problems.

[0006] The purpose of the present application can be achieved by the following technical solutions:

[0007] A data backflow optimization method based on big data, comprising the following steps:

[0008] S1: obtaining the number of shopping times N and the total transaction amount E of a user on an e-commerce platform, and calculating a detection period T , wherein T ave represents the mean value of the user's shopping time interval, λ1 is a preset first coefficient, λ2 is a preset second coefficient, and 0 < λ1 < λ2 < 1;

[0009] S2: Obtain the total number of times I and the duration of each visit IT of the user browsing the product interface within the detection period T, and calculate the average duration IT of each visit. ave The detection period T is corrected to obtain the first period T1 = T * (T / IT) ave );

[0010] S3: Obtain the original price S and current price S of the product the user has saved. now Calculate the discount level ΔS=SS now Get the number of times a user purchases or favorites items (IS) and the average purchase price (GS). ave Calculate the correction value Where δ is a preset correction coefficient, the first period is corrected to obtain the second period T2=T1*X;

[0011] S4: Record the products displayed on the e-commerce platform's homepage as recommended products, obtain the number of recommended product types TJ that appear on the e-commerce platform within the detection period T, and calculate the recommendation accuracy. , of which M j m represents the number of times the j-th recommended product appears. j This represents the number of times a user clicks on the j-th recommended product. If the recommendation accuracy Z ≤ 60%, the detection period T is recorded as an unqualified period. If the proportion of unqualified periods is greater than 20%, the correction time XT = T2 * Z is calculated, and the detection period is reset with the correction time XT.

[0012] As a further aspect of the present invention: in step S1, the number of times the user returned goods TH and the amount of the refund TE are obtained, and the number of times the user returned goods TH and the amount of the refund TE are removed from the number of shopping trips N and the total transaction amount E, and are not included in the calculation of the detection period T.

[0013] As a further aspect of the present invention: in step S2, the duration of a single browsing session IT is less than a preset minimum browsing duration IT. min Browsing behavior that does not meet the minimum requirement is recorded as invalid browsing, and invalid browsing is not included in the total number of times the product interface is viewed.

[0014] As a further aspect of the present invention: in step S2, the maximum duration of browsing the same product interface is taken as the value of the single browsing duration IT.

[0015] As a further aspect of the present invention: In step S1, if the number of times a user makes purchases N on the e-commerce platform is less than a preset minimum number of purchases N... min If so, then stop the subsequent operations.

[0016] As a further scheme of the present application: in the step S3, if the discount strength AS = 0 and the number of times IS = 0 that the user purchases the collection goods, the subsequent steps are stopped, and the staff is prompted to adjust the recommended content.

[0017] As a further scheme of the present application: in the step S4, if the number of times m j = 0 that the user clicks the jth recommended goods, the calculation of the recommendation accuracy Z is stopped, and the detection period T is directly recorded as an unqualified period.

[0018] As a further scheme of the present application: in the step S4, if the recommendation accuracy Z > 60%, the detection period is recorded as a qualified period, and the adjustment of the detection period T for the qualified period is suspended until the recommendation accuracy Z ≤ 60%.

[0019] The present application has the following advantages: In the present application, it is of great significance for e-commerce platforms to deeply understand the shopping behavior patterns and consumption habits of users. First of all, two key data indicators, the number of shopping times and the total transaction amount of users on the e-commerce platform, need to be obtained.

[0020] The number of shopping times is an important measure dimension, which directly reflects the frequency of users using the e-commerce platform to shop. A user who frequently places orders to purchase goods on the e-commerce platform will naturally have a relatively high number of shopping times. The total transaction amount reflects the consumption scale and economic strength of the user on the platform from another angle. A higher total transaction amount means that the user is willing to invest more funds on the platform to purchase goods, which not only indicates that the user has a high degree of recognition for the goods and services provided by the platform, but also reflects the user's activity level in e-commerce shopping.

[0021] By comprehensively analyzing these two data points, the degree of dependence of the user on the e-commerce shopping platform can be roughly reflected. Generally speaking, the higher the number of shopping times and the total transaction amount, the stronger the user's dependence on the e-commerce platform. This is because when a user frequently shops on a certain e-commerce platform and the cumulative transaction amount reaches a certain scale, it means that the platform has met the user's diverse needs, whether in terms of product variety, price advantage, or service quality, and has gained the trust and favor of the user.

[0022] In order to more accurately grasp the degree of dependence of the user on the platform and provide more personalized and high-quality services to the user accordingly, a specific calculation formula can be introduced to determine the detection period. This detection period is closely related to the number of shopping times and the total transaction amount. Specifically, the higher the number of shopping times and the total transaction amount, the smaller the detection period calculated by the formula.

[0023] The reason for this design is that the more dependent users are on the shopping platform, the more representative and forward-looking their behavior and needs are. These users are not only the core customer group of the platform, but also the important driving force for the continuous development and optimization of the platform's services. Therefore, for this part of the users, the detection frequency needs to be accelerated. By collecting their shopping data and feedback information more frequently, the platform can timely understand the problems and expectations they encounter in the shopping process, so as to adjust the product recommendation strategy, optimize the page layout, and improve the after-sales service quality. In this way, users can have a better experience in using the platform, further enhancing their loyalty and stickiness to the platform, forming a virtuous cycle, and promoting the long-term stable development of the e-commerce platform.

[0024] In addition to the above two key indicators of the number of user shopping times and total transaction amount, attention should also be paid to the relevant data of users browsing the product interface, mainly including the total number of times users browse the product interface and the duration of single browsing. These two data dimensions can reveal the user's shopping tendency and interest point from a more subtle perspective.

[0025] The total number of times users browse the product interface directly reflects the user's attention and exploration desire for various products on the e-commerce platform. When a user frequently browses the product interface, it may mean that the user has strong consumption motivation and shopping willingness, and is actively looking for goods that meet their needs. The duration of single browsing further reflects the user's interest in a particular product. If a user spends a long time on a product page, it indicates that the product has attracted enough attention from the user, and the user may be carefully checking the product's detailed information, pictures, user reviews, etc., trying to fully understand the characteristics and advantages of the product in order to make a purchase decision.

[0026] According to the general objective law, the user's purchase desire is positively correlated with the browsing behavior. That is, the more interested a user is in a product, the more time and effort they spend browsing it, and the greater the likelihood of purchase. For example, when a user repeatedly browses a certain product within a short period of time, and the duration of each browsing reaches the specified standard, it strongly suggests that the user has a strong interest in the product and a purchase intention. At this time, it can be considered that the user has a high probability of purchasing the product.

[0027] By recording and analyzing the user's browsing behavior in detail, the abstract concept of purchase desire can be concretized. This concretization not only makes our judgment of user purchase behavior more objective and accurate, but also makes the subsequent results calculated from these data more convincing. Based on these data, we can optimize the display and marketing strategies of products, improve the user's purchase conversion rate, and achieve sales growth.

[0028] The price is one of the key factors affecting the user's purchase decision, and its importance cannot be ignored. In the context of e-commerce shopping, when users face a large number of goods, in addition to considering the quality, function and user evaluation of the goods, the price is often an important consideration point for them to decide whether to place an order. Especially for users who are not buying goods for the first time, they usually have some understanding and experience of the goods they want to buy, so in this case, users will pay more attention to whether the price of the goods is preferential.

[0029] When the goods show a large discount, this advantage in price will form a great attraction to the user. Because for users who shop frequently, they know how to save expenses by capturing preferential activities while ensuring the quality of the goods. At this time, the huge discount of the goods can stimulate their stronger desire to buy, and they are more inclined to choose these high-value-for-money goods.

[0030] Therefore, for e-commerce platforms or merchants, when they identify that users show strong interest in such deeply discounted goods, they should pay more attention to the behavior dynamics of this part of users. One of the best strategies to strengthen attention is to achieve it by adjusting the detection frequency. Specifically, according to the pre-set formula, when the discount of the goods is greater, the corresponding detection period should be set smaller. This design fully conforms to the objective fact, because it means that the platform or the merchant needs to collect the shopping data and feedback information of this part of users more frequently, in order to understand their demand changes and shopping preferences in time, so as to respond to market changes quickly, provide more accurate personalized recommendations and services for users, and further improve the shopping experience and satisfaction of users.

[0031] Finally, according to the recommendation accuracy of the e-commerce platform, it is determined whether the detection period needs to be reset, if so, the specific correction duration is calculated by the formula, and the detection period is reset with the correction duration.

[0032] In summary, the present application uses data analysis algorithms for personalized recommendation according to the user's historical purchase records, browsing records, favorites information, etc., and realizes a backflow optimization method based on big data to reduce the impact of data backflow delay, thereby optimizing the data backflow speed of big data to obtain higher processing efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0033] The present application will be further described below in conjunction with the drawings.

[0034] Figure 1 It is a flowchart of the data backflow optimization method based on big data of the present application. DETAILED DESCRIPTION

[0035] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0036] Please refer to Figure 1 The present application is a data backflow optimization method based on big data, which comprises the following steps:

[0037] S1: Obtain the shopping frequency N and total transaction amount E of a user on an e-commerce platform, and calculate a detection period T according to the following formula: T=λ1*IT+λ2*IT ave , wherein T represents the average of the shopping time interval of the user, λ1 is a preset first coefficient, λ2 is a preset second coefficient, and 0<λ1<λ2<1;

[0038] S2: Obtain the total number I of times that the user browses the product interface within the detection period T and the single browsing time IT, and calculate the average single browsing time IT ave . Modify the detection period T to obtain a first period T1=T*(T / IT ave );

[0039] S3: Obtain the original price S and the current price S now of the product collected by the user, calculate the discount strength ΔS=S-S now , obtain the number IS of times that the user purchases the collected product and the average purchase price GS ave , and calculate the correction value , wherein δ is a preset correction coefficient, and the first period is modified to obtain a second period T2=T1*X;

[0040] S4: Record the product displayed on the homepage of the e-commerce platform as a recommended product, obtain the number TJ of types of recommended products appearing on the e-commerce platform within the detection period T, and calculate the recommendation accuracy , wherein M j represents the number of times that the jth recommended product appears, m j represents the number of times that the user clicks the jth recommended product, if the recommendation accuracy Z≤60%, the detection period T is recorded as an unqualified period, and if the proportion of the unqualified period is greater than 20%, the correction time XT=T2*Z is calculated to reset the detection period with the correction time XT.

[0041] It should be noted that for e-commerce platforms, it is of vital importance to understand the shopping behavior patterns and consumption habits of users. First of all, it is necessary to obtain two key data indicators: the number of shopping times and the total transaction amount of users on the e-commerce platform.

[0042] The number of shopping times is an important measure dimension, which intuitively reflects the frequency of users using the e-commerce platform to shop. A user who frequently places orders to purchase goods on the e-commerce platform will naturally have a relatively high number of shopping times. The total transaction amount, on the other hand, reflects the user's consumption scale and economic strength on the platform. A higher total transaction amount means that the user is willing to invest more money in purchasing goods on the platform, which not only indicates that the user has a high degree of recognition for the goods and services provided by the platform, but also reflects the user's activity level in e-commerce shopping.

[0043] By comprehensively analyzing these two data points, we can roughly reflect the user's dependence on the e-commerce shopping platform. Generally speaking, the higher the number of shopping times and the total transaction amount, the stronger the user's dependence on the e-commerce platform. This is because when a user frequently shops on a certain e-commerce platform and the cumulative transaction amount reaches a certain scale, it means that the platform has met the user's diverse needs, whether in terms of product variety, price advantage, or service quality, and has gained the user's trust and favor.

[0044] In order to more accurately grasp the user's dependence on the platform and provide more personalized and high-quality services to users accordingly, a specific calculation formula can be introduced to determine the detection period. This detection period is closely related to the number of shopping times and the total transaction amount. Specifically, the higher the number of shopping times and the total transaction amount, the smaller the detection period calculated by the formula.

[0045] The reason for this design is that users who are more dependent on the shopping platform tend to have more representative and forward-looking behavior and needs. These users are not only the core customer group of the platform, but also an important driving force for the platform's continuous development and service optimization. Therefore, for this part of users, the detection frequency needs to be accelerated. By collecting their shopping data and feedback information more frequently, the platform can timely understand the problems and expectations they encounter in the shopping process, and accordingly adjust the product recommendation strategy, optimize the page layout, and improve the quality of after-sales service. In this way, users can obtain a better experience in using the platform, further enhancing their loyalty and stickiness to the platform, forming a virtuous cycle, and promoting the long-term stable development of the e-commerce platform.

[0046] In addition to the above-mentioned acquisition of user shopping frequency and total transaction amount, two key indicators, attention should also be paid to the relevant data of user browsing the product interface, which mainly includes the total number of times the user browses the product interface and the duration of single browsing. These two data dimensions can reveal the user's shopping tendency and interest point from a more subtle perspective.

[0047] The total number of times the user browses the product interface directly reflects the user's attention and desire to explore various types of goods on the e-commerce platform. When a user frequently browses the product interface, it may mean that the user has a strong consumption motivation and shopping willingness, and is actively looking for goods that meet their needs. The duration of single browsing further reflects the user's interest in a particular product. If the user stays on a product page for a long time, it indicates that the product has attracted enough attention from the user, and the user may be carefully checking the product's detailed information, pictures, user reviews, etc., trying to fully understand the characteristics and advantages of the product in order to make a purchase decision.

[0048] According to the general objective law, the user's purchase desire is positively correlated with the browsing behavior. That is, the more interested the user is in a product, the more time and effort the user spends browsing the product, and the greater the likelihood of purchase. For example, when a user browses a product multiple times in a short period of time, and each browsing duration meets the specified standard, it strongly suggests that the user has a strong interest in the product and a purchase intention. At this time, it can be considered that the user has a greater likelihood of purchasing the product.

[0049] By recording and analyzing the user's browsing behavior in detail, the abstract concept of purchase desire can be concretized. This concretization not only makes our judgment of user purchase behavior more objective and accurate, but also makes the subsequent results calculated based on these data more convincing. Based on these data, we can optimize the display and marketing strategies of goods, improve the user's purchase conversion rate, and achieve sales performance growth.

[0050] Price, as one of the key factors affecting user purchase decision, is of great importance and cannot be ignored. In the context of e-commerce shopping, when users face numerous goods, in addition to considering the quality, function and user reviews of the goods, price is often an important consideration point for them to decide whether to place an order. Especially for users who are not buying the goods for the first time, they usually have some understanding and experience of the goods they want to buy, so in this case, users will pay more attention to whether the price of the goods is favorable.

[0051] When the goods present a large discount strength, this price advantage will form a great attraction to the user. Because for the user who often shops, they know how to save expenses by capturing the preferential activities on the premise of ensuring the quality of goods. At this time, the huge discount of goods can stimulate their stronger purchase desire, so they tend to choose these high cost-effective goods.

[0052] In view of this, for the e-commerce platform or the merchant, when it is identified that the user shows strong interest in such deeply discounted goods, the behavior dynamics of this part of the user should be more closely observed. One of the best strategies to strengthen attention is to achieve it by adjusting the detection frequency. Specifically, according to the pre-set formula, when the discount strength of the goods is greater, the corresponding detection period should be set smaller. This design fully conforms to the objective fact, because it means that the platform or the merchant needs to collect the shopping data and feedback information of this part of the user more frequently, so as to understand their demand changes and shopping preferences in time, so as to quickly respond to market changes, provide more accurate personalized recommendations and services for users, and further improve the shopping experience and satisfaction of users.

[0053] Finally, according to the recommendation accuracy of the e-commerce platform, it is determined whether the detection period value needs to be reset. If it is needed, the specific correction duration is calculated by the formula, and the detection period is reset with the correction duration.

[0054] In another preferred embodiment of the present application, the number of times of returning goods TH and the amount of returned goods TE of the user are obtained, and the number of times of returning goods TH and the amount of returned goods TE are excluded from the shopping times N and the total transaction amount E, and do not participate in the calculation of the detection period T.

[0055] And it is worth noting that in the actual process of goods transaction, the situation of returning goods often occurs. This returning behavior may be caused by many reasons, such as quality problems of goods, specifications, subjective dissatisfaction of users, etc. However, no matter what the reason is, the return of goods is an important link in the transaction process, which not only affects the result of single transaction, but also may have a profound impact on the overall sales data and analysis.

[0056] In order to ensure the rigor and accuracy of experimental data, we must exclude the returned goods from the data record. Because the returned goods have not actually completed the complete transaction process, their existence may distort the authenticity and integrity of the sales data. If the data of returned and exchanged goods is not timely excluded, they may be incorrectly included in key indicators such as total sales and total transaction times, thereby causing serious interference and deviation to the final result.

[0057] In another preferred embodiment of the present application, the single browsing duration IT is less than the preset minimum browsing duration ITmin The browsing behavior of the user is recorded as invalid browsing, and the invalid browsing is not counted in the total number of times I of browsing the commodity interface.

[0058] It can be understood that, in order to further improve the accuracy and reliability of the experimental data, we decide to eliminate the records of too short browsing duration. In the daily operation of the e-commerce platform, users may generate a large number of short-term commodity browsing records due to misoperation, curiosity clicks or quick browsing. These too short browsing durations often cannot truly reflect the user's interest or purchase intention for the commodity, but may introduce noise and interfere with our accurate judgment of user behavior patterns.

[0059] By setting a reasonable browsing duration threshold and eliminating all records below the threshold from the data set, we can effectively reduce the data bias caused by user errors or unintentional browsing. This step helps to ensure the rigor of the experiment, so that we can focus more on the user behavior data of those who are really interested in the commodity and have potential purchase intention.

[0060] After eliminating the records of too short browsing duration, we can more accurately identify the user's shopping preferences and interest points, providing more accurate data support for subsequent work such as commodity recommendation and marketing strategy formulation. At the same time, this also helps to improve user experience and avoid user confusion or dissatisfaction caused by irrelevant commodity recommendations, thereby further consolidating user trust and loyalty to the e-commerce platform.

[0061] In another preferred embodiment of the present application, the maximum duration of browsing the same commodity interface is taken as the value of the single browsing duration IT.

[0062] It should be noted that by tracking each browsing behavior of the user on the same commodity interface and recording the maximum duration therein, the most in-depth and continuous attention period of the user to the commodity can be captured. This maximum duration not only reflects the user's initial interest in the commodity, but also may include their careful study of the details of the commodity, their thinking process of comparing different specifications or color options, and even important reference for their final decision whether to purchase.

[0063] In another preferred embodiment of the present application, if the number of shopping times N of the user in the e-commerce platform is less than the preset minimum number of shopping times N min , the subsequent operation is stopped.

[0064] It should be noted that too little experimental data and large sample errors cannot accurately determine, in order to ensure accuracy, stop operating.

[0065] In another preferred embodiment of the present application, if the discount strength AS = 0 and the number of times IS = 0 that the user purchases the collectible item, the subsequent steps are stopped, and the staff is prompted to adjust the recommended content.

[0066] It can be understood that in order to more comprehensively improve the rigor and anti-interference ability of the experimental scheme, not only the data accuracy in the normal situation needs to be concerned, but also various special situations need to be fully considered, and corresponding rejection solutions are formulated. Special situations may include user abnormal behavior, system failure, data entry error, external factor influence, etc., which may cause potential interference and deviation to the experimental results.

[0067] By enumerating and rejecting these special situations and their solutions, not only the adaptability of the experimental scheme to complex environment is enhanced, but also the stability and reliability of the experimental results are improved. In the face of various uncertainties and challenges, higher anti-interference ability can be maintained to ensure that the final conclusion is accurate and reliable, providing more accurate market insight and decision support for businesses.

[0068] In another preferred embodiment of the present application, if the number of times m j = 0 that the user clicks the jth recommended item, the calculation of the recommendation accuracy Z is stopped, and the detection period T is directly recorded as an unqualified period.

[0069] Notably, omitting the calculation process and simplifying the operation process are beneficial to reduce the amount of calculation and thus achieve the purpose of saving time.

[0070] In another preferred embodiment of the present application, if the recommendation accuracy Z > 60%, the detection period is recorded as a qualified period, and the adjustment of the detection period T for the qualified period is suspended until the recommendation accuracy Z ≤ 60%.

[0071] Notably, when the recommendation accuracy Z > 60%, it means that the accuracy of the recommendation is high at this time, therefore, for the purpose of saving computing power, the adjustment of the detection period T is stopped until the recommendation accuracy Z ≤ 60%.

[0072] The above has described in detail one embodiment of the present application, but the content described is only the preferred embodiment of the present application, and cannot be considered as limiting the scope of the implementation of the present application. Any equivalent changes and improvements made within the scope of the present application shall still belong to the patent coverage range of the present application.

Claims

1. A data backflow optimization method based on big data, characterized in that, Includes the following steps: S1: Obtain the number of purchases N and total transaction amount E of the user on the e-commerce platform, and calculate the detection period. , among which, T ave λ1 represents the average time interval between user shopping sessions, λ2 is the preset first coefficient, and 0 < λ1 < λ2 < 1. S2: Obtain the total number of times I and the duration of each visit IT of the user browsing the product interface within the detection period T, and calculate the average duration IT of each visit. ave The detection period T is corrected to obtain the first period T1 = T * (T / IT) ave ), where the maximum time spent browsing the same product page is used as the value of the single browsing time IT; S3: Obtain the original price S and current price S of the product the user has saved. now Calculate the discount level ΔS=SS now Get the number of times a user purchases or favorites items (IS) and the average purchase price (GS). ave Calculate the correction value Where δ is a preset correction coefficient, the first period is corrected to obtain the second period T2=T1*X; S4: Record the products displayed on the e-commerce platform's homepage as recommended products, obtain the number of recommended product types TJ that appear on the e-commerce platform within the detection period T, and calculate the recommendation accuracy. , of which M j m represents the number of times the j-th recommended product appears. j This represents the number of times a user clicks on the j-th recommended product. If the recommendation accuracy Z ≤ 60%, the detection period T is recorded as an unqualified period. If the proportion of unqualified periods is greater than 20%, the correction time XT = T2 * Z is calculated, and the detection period is reset with the correction time XT.

2. The data backflow optimization method based on big data according to claim 1, characterized in that, In step S1, the number of times a user returns goods (TH) and the amount of the return (TE) are obtained. The number of times a user returns goods (TH) and the amount of the return (TE) are removed from the number of purchases (N) and the total transaction amount (E) and are not included in the calculation of the detection period (T).

3. The data backflow optimization method based on big data according to claim 1, characterized in that, In step S2, the duration of a single browsing session IT is set to be less than the preset minimum browsing duration IT. min Browsing behavior that does not meet the minimum requirement is recorded as invalid browsing, and invalid browsing is not included in the total number of times the product interface is viewed.

4. The data backflow optimization method based on big data according to claim 1, characterized in that, In step S1, if the number of times a user makes purchases N on the e-commerce platform is less than the preset minimum number of purchases N... min If so, then stop the subsequent operations.

5. The data backflow optimization method based on big data according to claim 1, characterized in that, In step S3, if the discount ΔS=0 and the number of times the user purchases the favorite item IS=0, then the subsequent steps are stopped and the staff is prompted to adjust the recommended content.

6. The data backflow optimization method based on big data according to claim 1, characterized in that, In step S4, if the user clicks on the j-th recommended product a number of times m... j =0, stop calculating the recommended accuracy rate Z, and directly record the detection cycle T as the unqualified cycle.

7. The data backflow optimization method based on big data according to claim 1, characterized in that, In step S4, if the recommended accuracy Z > 60%, the detection period is recorded as the qualified period, and the adjustment of the detection period T of the qualified period is paused until the recommended accuracy Z ≤ 60%.

Images