E-commerce user screening method based on big data

Abstract

The invention discloses an e-commerce user screening method based on big data, and belongs to the technical field of e-commerce, and the method comprises the steps: 1, building an e-commerce user information base which is used for storing e-commerce user data, and the e-commerce user data comprises e-commerce user portraits and e-commerce user information; 2, setting a classification label graph, and marking corresponding classification labels for the e-commerce users according to the classification label graph; 3, establishing a demand analysis model which is used for analyzing the user screening data of the e-commerce enterprise and determining a user screening target of the e-commerce enterprise; the user screening target is composed of corresponding target categories; 4, obtaining user screening data of the e-commerce enterprise, and analyzing the user screening data through the demand analysis model to obtain a user screening target; and 5, identifying a target category corresponding to the user screening target, forming a screening label combination according to the target category, and screening the e-commerce users according to the screening label combination.

Description

Technical Field

[0001] The present invention belongs to the technical field of e-commerce, and specifically relates to an e-commerce user screening method based on big data. Background Art

[0002] As an important product of the information age, e-commerce has profoundly changed people's shopping methods and consumption habits. With the booming development of e-commerce platforms, the accumulation and mining of e-commerce user data have become the key for e-commerce enterprises to enhance their competitiveness. However, there are many challenges and deficiencies in existing e-commerce user screening methods.

[0003] Traditional e-commerce user screening mainly relies on simple e-commerce user attributes (such as age, gender, region, etc.) and purchase history. This method often ignores the complexity and diversity of e-commerce user behavior. For example, e-commerce users' browsing behavior, search behavior, social media interactions, etc. can all reflect their potential needs and preferences, but this information has not been fully utilized in traditional screening methods. Moreover, with the change of demand, there will be different screening requirements, making it difficult for traditional screening methods to meet the dynamically changing screening needs.

[0004] Based on this, the present invention provides an e-commerce user screening method based on big data. Summary of the Invention

[0005] In order to solve the problems existing in the above solution, the present invention provides an e-commerce user screening method based on big data.

[0006] The object of the present invention can be achieved by the following technical solutions: An e-commerce user screening method based on big data, the method comprising: Step 1: Establish an e-commerce user information database, which is used to store e-commerce user data, and the e-commerce user data includes e-commerce user portraits and e-commerce user information; Further, the method for optimizing and adjusting the e-commerce user portraits in the e-commerce user information database includes: Set e-commerce user portrait standards, and check the e-commerce user portraits stored in the e-commerce user information database according to the e-commerce user portrait standards to obtain portraits to be optimized; Adjust the portraits to be optimized according to the e-commerce user portrait standards.

[0007] Further, checking the e-commerce user portraits stored in the e-commerce user information database according to the e-commerce user portrait standards includes: Establish a portrait checking model, and the expression of the portrait checking model is: ; Where: (hi, BZ) is the input data, hi is the corresponding e-commerce user portrait in the e-commerce user information database, i = 1, 2, ……, n, and n is the number of un-verified e-commerce user portraits in the e-commerce user information database; BZ is the e-commerce user portrait standard; hi→BZ indicates that the corresponding e-commerce user portrait meets the e-commerce user portrait standard; the output data is the portrait verification value HK(hi, BZ), and the portrait verification value is 1 or 0; Analyze the e-commerce user portraits stored in the e-commerce user information database and the e-commerce user portrait standard through the portrait verification model to obtain the portrait verification value of the e-commerce user portrait; Mark the e-commerce user portraits with a portrait verification value of 0 as portraits to be optimized.

[0008] Step two: Set up a classification label map, and assign corresponding classification labels to e-commerce users according to the classification label map; Furthermore, the method for setting up the classification label map includes: Obtain the classification types in real time and identify the classification indicators corresponding to the classification types; Collect the historical classification data corresponding to the classification types, and identify the classification standard combinations possessed by the classification types according to the historical classification data; Screen the classification standard combinations according to the historical classification data, determine the target standard combination of the classification types, and determine the classification standards corresponding to the classification types according to the target standard combination; Generate a classification label map according to the classification types, classification indicators and classification standards.

[0009] Furthermore, the method for screening the classification standard combinations according to the historical classification data includes: Statistically calculate the market share corresponding to the classification standard combination according to the historical classification data; identify the e-commerce background data corresponding to the classification standard combination; Obtain the target background, and calculate the similarity between the target background and the e-commerce background data corresponding to the classification standard combination; Calculate the screening value of the classification standard combination according to the screening formula, and the screening formula is: ; Where: SA is the screening value; δ is the similarity; HA is the market share; Select the classification standard combination with the largest screening value as the target standard combination.

[0010] Furthermore, the method for assigning corresponding classification labels to e-commerce users according to the classification label map includes: Identify the corresponding classification types and the lower-level categories corresponding to the classification types in real time according to the classification label map; identify the classification standards corresponding to the lower-level categories; Extract features from the e-commerce user data according to the classification metrics to obtain the classification metric data of the e-commerce users; Analyze the classification metric data and classification criteria through a preset classification recognition model to determine the target category; generate a classification label according to the target category, and mark the classification label on the e-commerce user.

[0011] Further, the method for analyzing the classification metric data and classification criteria through a classification recognition model includes: The expression of the classification recognition model is: ; In the formula: (YA, KZj) is the input data, YA is the classification metric data of the e-commerce user for the corresponding classification type; KZj is the classification criteria corresponding to the corresponding lower-level category within the classification type, j represents the corresponding lower-level category, j = 1, 2,..., m, and m is the number of lower-level categories corresponding to the classification type; YA→KZj means that the classification metric data meets the classification criteria of the corresponding lower-level category; the output data is the classification recognition value FL(YA, KZj), and the classification recognition value is 1 or 0; Analyze the classification metric data and classification criteria through a classification recognition model to obtain the classification recognition value of the e-commerce user for the corresponding lower-level classification; Mark the lower-level category with a classification recognition value of 1 as the target category.

[0012] Step three: Establish a demand analysis model, which is used to analyze the user screening data of the e-commerce enterprise to determine the user screening target of the e-commerce enterprise; the user screening target consists of the corresponding target categories; Step four: Obtain the user screening data of the e-commerce enterprise, and analyze the user screening data through the demand analysis model to obtain the user screening target; Step five: Identify the target category corresponding to the user screening target, form a screening label combination according to the target category, and screen the e-commerce users according to the screening label combination.

[0013] Compared with the prior art, the beneficial effects of the present invention are: The screening method of the present invention realizes automation and intelligence, can quickly process a large amount of user data, and output the screening result in real time. This greatly shortens the screening cycle, reduces the labor cost, improves the screening efficiency. At the same time, based on user portraits and precise screening, the present invention can provide personalized recommendations and services for different user groups. Moreover, the screening method of the present invention has good flexibility and scalability, and can be customized and optimized according to different business requirements, which helps e-commerce enterprises to respond to market changes, quickly adjust marketing strategies, and maintain competitive advantages. Brief Description of the Drawings

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or in the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0015] Figure 1 This is the flowchart of the method of the present invention. Specific embodiments

[0016] The following will clearly and completely describe the technical solutions of the present invention in conjunction with the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0017] As Figure 1 shown, a method for screening e-commerce users based on big data, the method includes: Step 1: Establish an e-commerce user information database, which is used to store e-commerce user data of e-commerce enterprises, platforms, etc. The e-commerce user data includes e-commerce user portraits and e-commerce user information; The e-commerce user portrait can be generated from the original e-commerce user information source channels, or can be generated based on the existing e-commerce user portrait generation technology according to the e-commerce user information. When obtaining some e-commerce user information, the corresponding e-commerce user portrait can be collected, because many e-commerce user analysis systems and modules will be configured with the function of generating e-commerce user portraits.

[0018] The e-commerce user information includes personal related information, shopping information, etc.

[0019] In one embodiment, due to the differences in different e-commerce user portrait generation methods, requirements, etc., some e-commerce user portraits will contain incomplete information. And to facilitate the subsequent update of the e-commerce user portrait according to the e-commerce user information, in this embodiment, the stored e-commerce user portraits are uniformly adjusted and supplemented. The method includes: Obtain the e-commerce user portrait standard, such as the data to be included, multi-dimensional information such as age, gender, region, occupation, income level, consumption habits, hobbies, etc.; check the e-commerce user portraits stored in the e-commerce user information database according to the e-commerce user portrait standard, determine the e-commerce user portraits that need to be adjusted, and mark them as portraits to be optimized; Adjust the optimized portrait according to the e-commerce user portrait standard, such as extracting corresponding e-commerce user feature information, and generating a new compliant e-commerce user portrait according to the preset e-commerce user portrait standard.

[0020] In one embodiment, check the e-commerce user portraits stored in the e-commerce user information database according to the e-commerce user portrait standard, including: Establish a portrait checking model, and the expression of the portrait checking model is: ; In the formula: (hi, BZ) is the input data, hi is the corresponding e-commerce user portrait in the e-commerce user information database, i = 1, 2,..., n, n is the number of un-checked e-commerce user portraits in the e-commerce user information database; BZ is the e-commerce user portrait standard; hi → BZ means that the corresponding e-commerce user portrait meets the e-commerce user portrait standard; the output data is the portrait checking value HK(hi, BZ), and the portrait checking value is 1 or 0; Analyze the e-commerce user portrait and the e-commerce user portrait standard through the portrait checking model to obtain the portrait checking value of the corresponding e-commerce user portrait; Mark the e-commerce user portraits with a portrait checking value of 0 as portraits to be optimized.

[0021] Step two: Set up a classification label map, and assign corresponding classification labels to e-commerce users according to the classification label map.

[0022] The classification label map is used to count various classifications, classification indicators, and classification standards, such as classification based on purchase behavior: according to indicators such as the purchase frequency and purchase amount of e-commerce users, e-commerce users are divided into high-value e-commerce users, medium-value e-commerce users, and low-value e-commerce users, etc. Classification based on preferences: According to the browsing and search behaviors of e-commerce users, identify the preferences of e-commerce users for specific categories or brands, and divide e-commerce users into different interest groups; Classification based on promotion sensitivity: By analyzing the proportion and amount of promotional offers in the historical orders of e-commerce users, measure the sensitivity of e-commerce users to promotions. E-commerce users are divided into categories such as extremely sensitive, relatively sensitive, generally sensitive, relatively insensitive, and extremely insensitive. This step helps e-commerce enterprises formulate more targeted promotional strategies.

[0023] Set the corresponding classifications and classification indicators according to the above description, and then determine the classification standards corresponding to different classifications within the classification according to the classification indicators. For example, for e-commerce users in classifications such as generally sensitive and medium-value, the classification standards can be set according to the classification standards of the same industry.

[0024] In one embodiment, the classification label map is established by existing methods.

[0025] In one embodiment, the method for setting up a classification label map includes: Obtain in real time the classification types of e-commerce users in the current e-commerce field based on big data or other means, and identify the classification metrics corresponding to the classification types; Collect the historical classification data corresponding to the classification types, and identify the classification standard combinations of the classification types according to the historical classification data. That is, for the same classification type, due to differences in platforms and other factors, its classification standards may be different, so different classification standard combinations will be formed; screen the classification standard combinations according to the historical classification data, determine the target standard combination of the classification type, and determine the classification standard corresponding to the classification type according to the target standard combination; Generate a classification label map according to the classification types, classification metrics, and classification standards. That is, when the classification types, classification metrics, and classification standards are clear, generate a classification label map based on a preset classification label map template.

[0026] In one embodiment, when screening the classification standard combinations, various metrics such as classification effect, share ratio, and scenario similarity can be used for screening, and a representative classification standard combination is selected as the target standard combination.

[0027] Exemplarily, count the market share corresponding to the corresponding classification standard combination according to the historical classification data, that is, the quantity ratio of the historical classification data corresponding to the classification standard combination; identify the e-commerce background data corresponding to the classification standard combination. The e-commerce background is the background of what kind of products an e-commerce enterprise sells, such as clothing background, snack background, etc.; the e-commerce background data is the combined data of various e-commerce backgrounds of the classification standard combination; Identify the target background, which is the e-commerce background of the e-commerce enterprise; Calculate the similarity between the target background and the e-commerce background data corresponding to each classification standard combination; Calculate the screening value of the corresponding classification standard combination according to the screening formula. The screening formula is: ; In the formula: SA is the screening value; δ is the similarity; HA is the market share; Select the classification standard combination with the largest screening value as the target standard combination.

[0028] In one embodiment, the method for tagging e-commerce users with corresponding classification labels according to the classification label map includes: Identify in real time the classification types in the classification label map and the subordinate categories corresponding to the classification types; for example, classification based on purchase behavior: according to indicators such as the purchase frequency and purchase amount of e-commerce users, divide e-commerce users into subordinate categories such as high-value e-commerce users, medium-value e-commerce users, and low-value e-commerce users; identify the classification standards corresponding to the corresponding subordinate categories; Extract the feature of e-commerce user data according to the classification index to obtain the classification index data of e-commerce users; Analyze the classification index data and classification criteria through a preset classification recognition model to determine the corresponding lower-level category of the e-commerce user in this classification type, and mark it as the target category; Generate corresponding classification labels according to the target category; mark the classification labels on the corresponding e-commerce users.

[0029] In one embodiment, the classification recognition model can be established based on existing intelligent technologies for realizing classification recognition judgment, such as being established based on neural networks such as CNN network or DNN network.

[0030] In one embodiment, the method for analyzing the classification index data and classification criteria through the classification recognition model includes: The expression of the classification recognition model is: ; In the formula: (YA, KZj) is the input data, YA is the classification index data of the e-commerce user for the corresponding classification type; KZj is the classification criteria corresponding to the corresponding lower-level category within the classification type, j represents the corresponding lower-level category, j = 1, 2,..., m, and m is the number of lower-level categories corresponding to the classification type; YA→KZj means that the classification index data conforms to the classification criteria of the corresponding lower-level category; the output data is the classification recognition value FL(YA, KZj), and the classification recognition value is 1 or 0; Analyze the classification index data and classification criteria through the classification recognition model to obtain the classification recognition value of the e-commerce user for the corresponding lower-level classification; Mark the lower-level category with a classification recognition value of 1 as the target category.

[0031] Step 3: Establish a demand analysis model, which is used to analyze the user screening data of e-commerce enterprises to determine the user screening target of e-commerce enterprises; the user screening data is the purpose data, requirement data, etc. of the e-commerce enterprise for screening e-commerce users, which are relevant description data for screening e-commerce users; the user screening target is each required user category, that is, the user screening requirements of the enterprise are transformed into each user category it screens, and the user screening target is composed of the corresponding target categories.

[0032] In one embodiment, the demand analysis model is established based on existing methods, such as being established based on neural networks such as CNN network or DNN network, and a corresponding training set is established and trained manually. The training set includes input data and output data. The input data is the user screening data, and the output data is the user screening target; analyze through the demand analysis model after successful training.

[0033] Step 4: Obtain the user screening data of the e-commerce enterprise, and analyze the user screening data through the demand analysis model to obtain the user screening target; Step 5: Identify the target category corresponding to the user screening target, form a screening label combination according to the target category, that is, composed of the target category labels of the corresponding target category; screen the e-commerce users according to the screening label combination.

[0034] The above formulas are all calculated by removing the dimension and taking their numerical values. The formula is a formula obtained by collecting a large amount of data for software simulation to be closest to the actual situation. The preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained by simulating a large amount of data.

[0035] The above embodiments are only used to illustrate the technical method of the present invention and not to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An e-commerce user screening method based on big data, characterized in that, The method includes: Step 1: Establish an e-commerce user information database for storing e-commerce user data, where the e-commerce user data includes e-commerce user portraits and e-commerce user information; Step 2: Set up a classification label map, and assign corresponding classification labels to e-commerce users according to the classification label map; Step 3: Establish a demand analysis model for analyzing the user screening data of an e-commerce enterprise to determine the user screening target of the e-commerce enterprise; the user screening target consists of corresponding target categories; Step 4: Obtain the user screening data of an e-commerce enterprise, and analyze the user screening data through the demand analysis model to obtain the user screening target; Step 5: Identify the target category corresponding to the user screening target, form a screening label combination according to the target category, and screen the e-commerce users according to the screening label combination.

2. The e-commerce user screening method based on big data according to claim 1, wherein The method for optimizing and adjusting the e-commerce user portraits in the e-commerce user information database includes: Set the e-commerce user portrait standard, check the e-commerce user portraits stored in the e-commerce user information database according to the e-commerce user portrait standard to obtain the portraits to be optimized; Adjust the portraits to be optimized according to the e-commerce user portrait standard.

3. A method for screening e-commerce users based on big data according to claim 2, characterized in that Checking the e-commerce user portraits stored in the e-commerce user information database according to the e-commerce user portrait standard includes: Establish a portrait checking model, and the expression of the portrait checking model is: ； In the formula: (hi, BZ) is the input data, hi is the corresponding e-commerce user portrait in the e-commerce user information database, i = 1, 2,..., n, n is the number of un-checked e-commerce user portraits in the e-commerce user information database; BZ is the e-commerce user portrait standard; hi→BZ means that the corresponding e-commerce user portrait meets the e-commerce user portrait standard; the output data is the portrait checking value HK(hi, BZ), and the portrait checking value is 1 or 0; Analyze the e-commerce user portraits stored in the e-commerce user information database and the e-commerce user portrait standard through the portrait checking model to obtain the portrait checking value of the e-commerce user portraits; Mark the e-commerce user portraits with a portrait checking value of 0 as the portraits to be optimized.

4. A method for screening e-commerce users based on big data according to claim 1, characterized in that, The method for setting up the classification label map includes: Obtain the classification types in real time and identify the classification indicators corresponding to the classification types; Collect the historical classification data corresponding to the classification types, and identify the classification standard combinations of the classification types according to the historical classification data; Screen the classification standard combinations according to the historical classification data, determine the target standard combinations of the classification types, and determine the classification standards corresponding to the classification types according to the target standard combinations; Generate a classification label map according to the classification types, classification indicators and classification standards.

5. A method for screening e-commerce users based on big data according to claim 4, characterized in that, The method for screening the classification standard combinations according to the historical classification data includes: Statistically analyze the market share corresponding to the classification standard combinations according to the historical classification data; identify the e-commerce background data corresponding to the classification standard combinations; Obtain the target background, and calculate the similarity between the target background and the e-commerce background data corresponding to the classification standard combinations; Calculate the screening value of the classification standard combinations according to the screening formula, and the screening formula is: ； Where: SA is the screening value; δ is the similarity; HA is the market share; Select the classification standard combination with the largest screening value as the target standard combination.

6. The method for screening e-commerce users based on big data according to claim 4, characterized in that, The method of assigning corresponding classification labels to e-commerce users according to the classification label map includes: Real-time identify the corresponding classification categories and the lower-level categories corresponding to the classification categories according to the classification label map; identify the classification criteria corresponding to the lower-level categories; Extract the feature of the e-commerce user data according to the classification index to obtain the classification index data of the e-commerce user; Analyze the classification index data and the classification criteria through a preset classification recognition model to determine the target category; generate a classification label according to the target category and mark the classification label on the e-commerce user.

7. A method for screening e-commerce users based on big data according to claim 6, characterized in that, The method of analyzing the classification index data and the classification criteria through a classification recognition model includes: The expression of the classification recognition model is: ； Where: (YA, KZj) is the input data, YA is the classification index data of the e-commerce user for the corresponding classification category; KZj is the classification criteria corresponding to the corresponding lower-level category within the classification category, j represents the corresponding lower-level category, j = 1, 2,..., m, m is the number of lower-level categories corresponding to the classification category; YA→KZj means that the classification index data meets the classification criteria of the corresponding lower-level category; the output data is the classification recognition value FL(YA, KZj), and the classification recognition value is 1 or 0; Analyze the classification index data and the classification criteria through a classification recognition model to obtain the classification recognition value of the e-commerce user for the corresponding lower-level classification; Mark the lower-level category with a classification recognition value of 1 as the target category.

Images