E-commerce big data service management system based on user behavior data analysis

Abstract

The present invention relates to the field of e-commerce management technology, and specifically, to an e-commerce big data service management system based on user behavior data analysis. It includes a commodity pre-sale status prediction module, a user correlation analysis module, a user behavior information analysis module, and an in-sale order sequence management module, wherein: the commodity pre-sale status prediction module is used to publish advertisements corresponding to the commodities to be sold, and obtain the richness of the advertisement content. In the e-commerce big data service management system based on user behavior data analysis, by using advertisements to collect interested users before sales, and analyzing the distribution status and density of interested users to provide a basis for identifying hired users, at the same time, all users are centrally managed in combination with waiting channels, so that all users have the opportunity to place orders. Therefore, in the process of identifying hired users, the impact on normal users is reduced, and the shopping experience of normal users is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of e-commerce management, and in particular to an e-commerce big data service management system based on user behavior data analysis. Background Art

[0002] E-commerce big data refers to the various traces and data left by users after they make purchases on e-commerce platforms, including product information, prices, express delivery costs, etc. These data are stored in the big databases of various platforms, and data sharing between platforms is achieved through API interfaces. E-commerce big data covers various behavioral data of users in e-commerce activities, such as purchase time, purchased goods, payment amount, etc. These data are of great significance to the operation and decision-making of e-commerce platforms.

[0003] As e-commerce platforms become more aware of prevention and optimize the shopping experience of normal users, they are stepping up their crackdown on abnormal traders (those who illegally resell or hoard certain goods to obtain illicit profits). E-commerce platforms use big data such as users' consumption habits, order addresses, and historical purchase information to identify abnormal traders, and then terminate the transaction of the order, making it impossible for the abnormal traders' personal accounts to trade goods normally.

[0004] However, for some abnormal traders who are willing to take risks, they can still hire normal users (referring to users hired by abnormal traders for the purpose of illegally reselling or hoarding certain goods) to buy up goods. After a purchase, the abnormal traders will replace a new group of hired personnel, which leads to uncertainty in the hired users, making it difficult for traditional identification methods to obtain sufficient data, making it difficult to respond and prone to misjudgment, resulting in some users finding it difficult to purchase the goods they need, affecting the purchasing experience of normal users. Summary of the invention

[0005] The purpose of the present invention is to provide an e-commerce big data service management system based on user behavior data analysis to solve the problems raised in the above background technology.

[0006] To achieve the above purpose, an e-commerce big data service management system based on user behavior data analysis is provided, including a product pre-sale status prediction module, a user correlation analysis module, a user behavior information analysis module, and an in-sale order sequence management module, wherein:

[0007] The commodity pre-sale status prediction module is used to publish advertisements corresponding to the commodities to be sold, obtain the richness of the advertisement content, and establish an interested user information collection platform to obtain interested users through advertisement information;

[0008] The user association analysis module obtains the location distribution of interested users and analyzes the predicted sales information of the goods to be sold in the area based on the distribution status;

[0009] The user behavior information analysis module establishes an order verification platform to collect information about users who submit orders, and identifies hired users in combination with the predicted sales information;

[0010] The on-sale order sequence management module adjusts the queue for submitting orders so that the priority of normal users' orders is higher than that of hired users' orders.

[0011] As a further improvement of the technical solution, the advertisements published by the commodity pre-sale status prediction module include text advertisements, picture advertisements and video advertisements.

[0012] As a further improvement of the technical solution, the method steps for obtaining the richness of the advertisement content in the commodity pre-sale status prediction module are as follows:

[0013] S1. Collect the parameter information of the product and obtain the parameter quantity P corresponding to the parameter information q ;

[0014] S2. Obtain the parameter quantity P in each type of advertisement q The sum of P s ;

[0015] S3. Calculate the correlation between each type of advertisement and product B e , where the correlation B e = Total P s ÷Total number of parameters P z ;

[0016] S4. Set correlation threshold B t , when the correlation B e ≥Correlation threshold B t When the relevance B e <Correlation threshold B t , mark the ad as not of interest.

[0017] As a further improvement of the technical solution, the method steps for the commodity pre-sale status prediction module to establish an interested user information collection platform are as follows:

[0018] S1. Record the user's behavior information and obtain the number of ads that the user is interested in watching. PCS ;

[0019] S2. Obtain the number of interesting advertisements delivered to users per unit time I PCS The sum of I s, and calculate the proportion of users who watch interesting ads A ccount ; Among them, the proportion of users watching interesting ads is A ccount = Quantity I PCS ÷Total I s ;

[0020] S3. Set the ratio threshold P k , the proportion of users watching interesting ads A ccount ≥ Ratio threshold P k When the user views the advertisement of interest, the user is marked as an interested user; when the user views the advertisement of interest, the proportion A ccount <Ratio threshold P k , mark the user as a non-interesting user.

[0021] As a further improvement of the technical solution, the method steps for predicting sales information by the user association analysis module are as follows:

[0022] S1. Collect the number of interested users G PCS and location information to obtain the dense and non-dense areas of the users of interest in the same city;

[0023] S2. Obtain shopping data of interested users in non-dense areas, and mark the shopping data as comparison data.

[0024] As a further improvement of the technical solution, the method steps for establishing an order verification platform in the user behavior information analysis module are as follows:

[0025] S1. Establish an order waiting channel so that users who submit orders can queue up in the waiting channel according to the order sequence;

[0026] S2. Obtain the user's information through the order and retrieve the user's historical behavior;

[0027] S3. Based on historical behaviors, interested users are marked as normal users and uninterested users are marked as sensitive users.

[0028] As a further improvement of the technical solution, the method steps for identifying employed users in the user behavior information analysis module are as follows:

[0029] S1. Get the number of normal users U PCS ;

[0030] S2. Calculate the number of normal users U PCS The number of interested users G PCS The proportion of G ccount ; Among them, the ratio G ccount = quantity U PCS ÷Quantity G PCS ;

[0031] S3. Set safety threshold S t , when the ratio G ccount ≥Safety threshold S t When the ratio G ccount <Safety threshold S t When receiving a sensitive user's delivery address,

[0032] S4. Determine the area where the delivery address is located. When the delivery address is in a dense area, keep the mark of the sensitive user. When the address is in a non-dense area, compare the historical shopping data of the sensitive user with the comparison data.

[0033] S5. Based on the relevance comparison results, irrelevant sensitive users are marked as employed users.

[0034] As a further improvement of the technical solution, the method steps for the order sequence management module to adjust the queue are as follows:

[0035] S1. Get all orders in the waiting channel;

[0036] S2. Assign priorities according to the status of users. The order after assigning priorities is: normal users > sensitive users > hired users;

[0037] S3. Adjust the order of all orders according to priority;

[0038] S4. Send the adjusted order status to the user.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] In this e-commerce big data service management system based on user behavior data analysis, advertisements are used to collect interested users before sales, and the distribution and density of interested users are analyzed to provide a basis for identifying hired users. At the same time, all users are centrally managed in combination with waiting channels, so that all users have the opportunity to place orders. After identification, the order is adjusted in a sorting manner, thereby reducing the impact on normal users in the process of identifying hired users and improving the shopping experience of normal users. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a schematic diagram of the flow of the overall module of the present invention;

[0042] The meaning of each number in the figure is:

[0043] 100. Prediction module of commodity status before sale; 200. User correlation analysis module; 300. User behavior information analysis module; 400. Order sequence management module during sale. DETAILED DESCRIPTION

[0044] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] It should be noted that the hired users mentioned in the present invention refer to users hired by abnormal traders for the purpose of illegally reselling or hoarding certain commodities.

[0046] See also Figure 1 As shown, an e-commerce big data service management system based on user behavior data analysis is provided, including a commodity pre-sale status prediction module 100, a user relevance analysis module 200, a user behavior information analysis module 300, and an in-sale order sequence management module 400, wherein:

[0047] The commodity pre-sale status prediction module 100 is used to publish advertisements corresponding to the commodities to be sold, obtain the richness of the advertisement content, and establish an interested user information collection platform to obtain interested users through advertisement information;

[0048] The user association analysis module 200 obtains the location distribution of interested users and analyzes the predicted sales information of the goods to be sold in the area based on the distribution status;

[0049] The user behavior information analysis module 300 establishes an order verification platform, collects information about users who submit orders, and identifies hired users in combination with predicted sales information;

[0050] The order sequence management module 400 adjusts the queue for submitting orders so that the priority of the orders of normal users is higher than that of the orders of hired users.

[0051] In the above, the advertisements released by the commodity pre-sale status prediction module 100 include text advertisements, picture advertisements and video advertisements.

[0052] Specifically, the steps of the method for obtaining the richness of the advertisement content in the commodity pre-sale status prediction module 100 are as follows:

[0053] S1. Collect the parameter information of the product and obtain the parameter quantity P corresponding to the parameter information q ;

[0054] S2. Obtain the parameter quantity P in each type of advertisement q The sum of P s ;

[0055] S3. Calculate the correlation between each type of advertisement and product B e , where the correlation B e = Total P s ÷Total number of parameters P z ;

[0056] S4. Set correlation threshold B t , when the correlation B e ≥Correlation threshold B t When the relevance B e <Correlation threshold B t , mark the ad as not of interest.

[0057] For example, when the product is a mobile phone, first obtain the parameter quantity P of the mobile phone q For example, the four parameters of "phone size", "phone color", "screen quality" and "camera function" are P q ; Then, obtain the parameter quantity P in the text advertisement, picture advertisement and video advertisement respectively q The sum of P s For example, text ads only have one parameter, “screen quality”. q , picture ads have two parameters: "screen quality" and "photo function". q Video ads have four parameters: "phone size", "phone color", "screen quality" and "camera function". q ; Then, the parameter quantity P of text ads q The sum of P s =1, the parameter quantity P of picture ads q The sum of P s =2, the parameter quantity P of video ads q The sum of P s is 4; this means that the relevance between text ads and products is B e 25%, the relevance of image ads to products B e 50%, the relevance of video ads to products B e is 100%. Assume that the correlation threshold B t is 75%, at this time, the correlation B e More than 75% of the ads, i.e. video ads, are interest-based ads.

[0058] Among them, the correlation B e Also represents richness, correlation B eThe higher the value, the richer the content, which means the product parameters introduced in the advertisement are more comprehensive.

[0059] Furthermore, the steps of the method for the commodity pre-sale status prediction module 100 to establish an interested user information collection platform are as follows:

[0060] S1. Record the user's behavior information and obtain the number of ads that the user is interested in watching. PCS ;

[0061] S2. Obtain the number of interesting advertisements delivered to users per unit time I PCS The sum of I s , and calculate the proportion of users who watch interesting ads A ccount ; Among them, the proportion of users watching interesting ads is A ccount = Quantity I PCS ÷Total I s ;

[0062] S3. Set the ratio threshold P k , the proportion of users watching interesting ads A ccount ≥ Ratio threshold P k When the user views the advertisement of interest, the user is marked as an interested user; when the user views the advertisement of interest, the proportion A ccount <Ratio threshold P k , mark the user as a non-interesting user.

[0063] Specifically, since the interested ads introduce more parameters of the product, if the user watches all the interested ads, it means that the user may be interested in the product, but it cannot be ruled out that the user hired by the abnormal trader has also watched the ad. For this reason, it is also necessary to analyze the number of users' views. For example, within a week, 10 interested ads are displayed to users. If user A watches 7 of the 10 interested ads and user B watches 2 of the 10 interested ads, then the proportion of user A watching interested ads is A. ccount The proportion of ads that user B is interested in is 70%. ccount Assuming the ratio threshold P is 20%, k is 70%, indicating that user a is an interested user and user b is a non-interested user.

[0064] It should be noted that the contents of the 10 advertisements of interest to users are different. For example, the products are introduced by different people. This can avoid content duplication and prevent users from experiencing viewing fatigue.

[0065] Specifically, the method steps for the user association analysis module 200 to predict sales information are as follows:

[0066] S1. Collect the number of interested users G PCS and location information to obtain the dense and non-dense areas of the interested users in the same city;

[0067] S2. Obtain shopping data of interested users in non-dense areas, and mark the shopping data as comparison data.

[0068] That is to say, the higher the density of interested users in the area, the better the advertising effect in the area, the more people buy the product, and some users who have not seen the advertisement will learn about the product through friends' dissemination and make a purchase. The lower the density, the worse the advertising effect in the area, and it is difficult for users who have not seen the advertisement to learn about the product through other means.

[0069] In addition, the method steps for establishing the order verification platform in the user behavior information analysis module 300 are as follows:

[0070] S1. Establish an order waiting channel so that users who submit orders can queue up in the waiting channel according to the order sequence;

[0071] S2. Obtain the user's information through the order and retrieve the user's historical behavior;

[0072] S3. Based on historical behaviors, interested users are marked as normal users and uninterested users are marked as sensitive users.

[0073] For example, the order verification platform receives four orders, and then obtains user information through the orders, namely user a, user b, user c, and user d. At this time, the historical behavior information of user a, user b, user c, and user d is obtained respectively to check the proportion of interested ads viewed by these four users before. ccount Is it greater than the threshold P? k , assuming that user a and user b watch the advertisements of interest in a ratio of A ccount Greater than or equal to the threshold P k , indicating that user a and user b are interested users; user c and user d watch the ads of interest in proportion A ccount is less than the threshold P k , indicating that user c and user d are non-interested users; at this time, it is necessary to mark user a and user b as normal users, and user c and user d as sensitive users for further judgment.

[0074] There is no limit on the number of users in the waiting channel. For example, if a product is limited to 10,000 units and 20,000 people rush to buy it, these 20,000 people will enter the waiting channel to queue up, and then re-queue and submit orders according to the ranking. In other words, among these 20,000 people, only the top 10,000 can successfully complete the order.

[0075] Furthermore, the method steps for identifying the employed users in the user behavior information analysis module 300 are as follows:

[0076] S1. Get the number of normal users U PCS ;

[0077] S2. Calculate the number of normal users U PCS The number of interested users G PCS The proportion of G ccount ; Among them, the ratio G ccount = quantity U PCS ÷Quantity G PCS ;

[0078] S3. Set safety threshold S t , when the ratio G ccount ≥Safety threshold S t When the ratio G ccount <Safety threshold S t When receiving a sensitive user's delivery address,

[0079] S4. Determine the area where the delivery address is located. When the delivery address is in a dense area, keep the mark of the sensitive user. When the address is in a non-dense area, compare the historical shopping data of the sensitive user with the comparison data.

[0080] S5. Based on the relevance comparison results, irrelevant sensitive users are marked as employed users.

[0081] After creating an order, in order to avoid misjudgment that affects the user's shopping experience, after obtaining the order of the user who is not interested, first count the number of normal users U PCS Count the number of interested users who have watched the ad G PCS For comparison, when the number of normal users U PCS The number of interested users G PCS The proportion of G ccount Greater than or equal to the set safety threshold S t , it indicates that the interested user has already purchased the product, and it will not affect the shopping experience of normal users. At this time, there is no need to judge sensitive users. You only need to keep the mark of sensitive users to facilitate subsequent queuing;

[0082] When the number of normal users U PCS The number of interested users G PCS The proportion of G ccount Less than the set safety threshold S t , indicating that some interested users did not purchase the product. At this time, it is necessary to analyze sensitive users. First, the delivery address is used to determine the area where sensitive users are located. When the delivery address is in a dense area, since the advertising dissemination effect in dense areas is good, such sensitive users can learn about it through friends and make purchases. Therefore, such sensitive users are less suspicious and are not marked in the next step.

[0083] When the delivery address is in a non-dense area, due to the poor advertising effect in non-dense areas, direct purchase by sensitive users is suspicious. At this time, by comparing the historical shopping information of sensitive users with users in non-dense areas, when the shopping information is similar, it indicates that the sensitive user has a demand for the product. When the shopping information is not similar, it indicates that the user has no demand for the product. At this time, the sensitive user is marked as an employed user.

[0084] Furthermore, the method steps for adjusting the queue by the order sequence management module 400 are as follows:

[0085] S1. Get all orders in the waiting channel;

[0086] S2. Assign priorities according to the status of users. The order after assigning priorities is: normal users > sensitive users > hired users;

[0087] S3. Adjust the order of all orders according to priority;

[0088] S4. Send the adjusted order status to the user.

[0089] That is to say, after obtaining normal users, sensitive users and hired users, normal users have the highest priority because they have watched the advertisements continuously. At the same time, sensitive users are behind normal users, and hired users are at the end because they are purchased on behalf of abnormal traders. In this way, after sorting, different users can check whether the order is completed according to the order. For example, if the product is limited to 10,000, there are 8,000 normal users, 1,900 suspicious users, and 300 hired users, then both normal users and suspicious users can complete the order, and the hired users in the first 100 can complete the order, while the hired users in the last 200 cannot complete the order.

[0090] In this way, by centrally managing all users during the rush to buy goods, and sorting the centrally managed users by recording interested users, most normal users are provided with the opportunity to place orders.

[0091] It can be seen that by using advertisements to collect interested users before sales, and analyzing the distribution status and density of interested users to provide a basis for identifying hired users, and at the same time, combining waiting channels to centrally manage all users, so that all users have the opportunity to place orders. After identification, the order is adjusted in a sorting manner, thereby reducing the impact on normal users in the process of identifying hired users and improving the shopping experience of normal users.

[0092] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. E-commerce big data service management system based on user behavior data analysis, characterized by: The system comprises a commodity pre-sale status prediction module (100), a user association analysis module (200), a user behavior information analysis module (300), and an in-sale order sequence management module (400), wherein: The commodity pre-sale status prediction module (100) is used to publish advertisements corresponding to the commodities to be sold, obtain the richness of the advertisement content, and establish an interested user information collection platform to obtain interested users through user behavior information; The user relevance analysis module (200) performs the following method steps: S1. Collect the number of interested users and location information to obtain the dense and non-dense areas of the users of interest in the same city; S2. Obtain shopping data of interested users in non-dense areas, and mark the shopping data as comparison data; The user behavior information analysis module (300) establishes an order verification platform, collects information of users who submit orders, and identifies hired users in combination with the comparison data; The method steps for the user behavior information analysis module (300) to establish an order verification platform are as follows: S1. Establish an order waiting channel so that users who submit orders can queue up in the waiting channel according to the order sequence; S2. Obtain user information through orders and retrieve user historical behavior information; S3. Based on historical behavior information, mark the interested users as normal users and mark the uninterested users as sensitive users; The user behavior information analysis module (300) identifies the employed user in the following steps: S1. Get the number of normal users ; S2. Calculate the number of normal users The number of interested users The proportion of Among them, the proportion =Quantity quantity ; S3. Set safety threshold , when the ratio Safety threshold When the ratio is Safety threshold When receiving a sensitive user's delivery address, S4. Determine the area where the delivery address is located. When the delivery address is in a dense area, keep the mark of the sensitive user. When the address is in a non-dense area, compare the historical shopping data of the sensitive user with the comparison data. S5. Based on the correlation comparison results, irrelevant sensitive users are marked as employed users; The on-sale order sequence management module (400) adjusts the queue for submitting orders so that the priority of orders from normal users is higher than that of orders from hired users.

2. The e-commerce big data service management system based on user behavior data analysis according to claim 1, characterized in that: The advertisements published by the commodity pre-sale status prediction module (100) include text advertisements, picture advertisements and video advertisements.

3. The e-commerce big data service management system based on user behavior data analysis according to claim 2 is characterized by: The steps of the method for the commodity pre-sale status prediction module (100) to obtain the richness of the advertisement content are as follows: S1. Collect the parameter information of the product and obtain the parameter quantity corresponding to the parameter information ; S2. Get the number of parameters in each type of advertisement The sum of ; S3. Calculate the relevance of each type of advertisement and product , where the correlation =Sum Total number of parameters ; S4. Set the correlation threshold , when the correlation Relevance Threshold When the ad is marked as an interesting ad; when the relevance Relevance Threshold , mark the ad as not of interest.

4. The e-commerce big data service management system based on user behavior data analysis according to claim 3 is characterized by: The method steps for the commodity pre-sale status prediction module (100) to establish an interested user information collection platform are as follows: S1. Record user behavior information and obtain the number of ads that users are interested in. ; S2. Get the number of interesting ads delivered to users per unit time The sum of , and calculate the proportion of ads that users are interested in watching ; Among them, the proportion of users watching interesting ads =Quantity ; S3. Set the ratio threshold , the proportion of ads that users are interested in watching Ratio Threshold When the user views the advertisement of interest Ratio Threshold , mark the user as a non-interesting user.

5. The e-commerce big data service management system based on user behavior data analysis according to claim 1 is characterized by: The method steps for the on-sale order sequence management module (400) to adjust the queue are as follows: S1. Get all orders in the waiting channel; S2. Assign priorities according to the status of users. The order after assigning priorities is: normal users > sensitive users > hired users; S3. Adjust the order of all orders according to priority; S4. Send the adjusted order status to the user.

Images