A user portrait generation cloud platform for enterprise digital operations

Abstract

The present invention relates to the technical field of enterprise digital operation, and discloses a user portrait generation cloud platform in enterprise digital operation, including a data collection unit, a data processing unit, a feature analysis unit and a portrait integration unit. The present invention can deeply mine data value through a sophisticated data processing flow, such as calculating various frequency features, interval duration features, stay duration ratio features and consumption amount features, so as to make the obtained user portrait more accurate. Through multi-dimensional analysis of the time interval, frequency and stay duration of user browsing and purchasing behaviors, it can scientifically and objectively evaluate the characteristic indicators of users in terms of behavior, transaction and preference. By using these indicators, it can provide enterprises with clear and quantifiable user feature descriptions. The generated user portraits including activity levels and consumption levels can help enterprises accurately locate user groups and formulate targeted marketing strategies.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise digital operation, and in particular to a user portrait generation cloud platform in enterprise digital operation. Background Art

[0002] In today's era of rapid digital development, digital operations have become a key link in enterprise development. With the intensification of market competition and the increasing diversification of consumer demand, enterprises need to understand users more deeply and accurately in order to provide them with personalized products and services, improve user satisfaction and loyalty, and enhance market competitiveness.

[0003] However, traditional user analysis methods have many limitations. On the one hand, they can only obtain and analyze part of the user's data, such as simple purchase records, while ignoring the details of the user's behavior when browsing products, such as the number of views, length of stay, etc. On the other hand, traditional methods lack systematicity and comprehensiveness when processing user data, making it difficult to comprehensively evaluate the user's behavioral characteristics, transaction characteristics, and preference characteristics from multiple dimensions.

[0004] In addition, the existing user portrait generation methods are not precise and detailed enough to accurately reflect the real needs and behavior patterns of users, resulting in a lack of effective basis for enterprises to formulate marketing strategies, optimize product design, etc. Therefore, in order to meet the needs of enterprises for comprehensive, in-depth and accurate analysis of users in digital operations, a cloud platform that can integrate multi-dimensional user data and generate accurate and comprehensive user portraits through scientific and rigorous calculation and analysis methods has become an urgent need. Summary of the invention

[0005] The purpose of the present invention is to provide a user portrait generation cloud platform in enterprise digital operations, which solves the technical problems raised in the background technology.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A user portrait generation cloud platform in enterprise digital operations, comprising:

[0008] A data collection unit is used to collect relevant data of the target user, including the number of views and purchases of products of different categories, the timestamp of each product view and the timestamp of each product purchase, the corresponding stay time of each product view, and the purchase amount of each product purchase;

[0009] A data processing unit, used to perform feature processing on the relevant data of the target user, and obtain multiple data features according to the feature processing results;

[0010] The multiple data features are browsing frequency features, purchase frequency features, browsing interval duration features, purchase interval duration features, stay time ratio features, and consumption amount features corresponding to all categories of products, as well as browsing frequency features, purchase frequency features, and stay time ratio features corresponding to different categories of products;

[0011] A feature analysis unit, used to perform feature index analysis based on the multiple data features obtained by the data processing unit, and obtain multiple feature indexes corresponding to the user portrait;

[0012] The multiple characteristic indicators corresponding to the user portrait are behavioral characteristic indicators, transaction characteristic indicators, and preference characteristic indicators;

[0013] The portrait integration unit is used to integrate the behavior feature indicators, transaction feature indicators and preference feature indicators obtained by the feature analysis unit to obtain a user portrait.

[0014] As a further solution of the present invention: the feature processing method is as follows:

[0015] StepA1, obtain the number of views and purchases of all categories of products by users in a specified period, and mark them as n and m respectively;

[0016] Step A2, by using PL = n / t and PG = m / t, calculate the browsing frequency characteristics PL and the purchasing frequency characteristics PG of all categories of products of users in the specified period respectively;

[0017] Among them, t represents the duration corresponding to the specified period;

[0018] StepA3: Get the timestamps of each browsing and purchasing of a product, and based on them, get the intervals between the corresponding adjacent timestamps of each browsing and the intervals between the corresponding adjacent timestamps of each purchasing, and mark them as LS i and GS j , i=1, 2,...n-1, j=1, 2,...m-1;

[0019] StepA4, through and Calculate the average values ​​LS0 and GS0 of the intervals between adjacent timestamps corresponding to each browsing of the product and the intervals between adjacent timestamps corresponding to each purchasing of the product, and then record them as the browsing interval duration feature and the purchase interval duration feature respectively;

[0020] Step A5, obtain the corresponding dwell time of each product browsing, and mark it as Tk, k = 1, 2, ... n;

[0021] StepA6. Pass Calculate the residence time ratio feature T0;

[0022] Step A7, obtain the purchase amount of each product purchase and mark it as Ur, r = 1, 2, ... m;

[0023] StepA8, through Calculate the consumption amount feature U0;

[0024] StepA9: Get the number of times users browse and purchase products of each category within a specified period, and mark them as n1 respectively. g and m1 g ; g = 1, 2, ... z, z represents the number of categories of all products;

[0025] StepA10, through PL1 g =n1 g / t and PG1 g =m1 g / t, calculate the browsing frequency characteristics PL1 of each category of products in the specified period g and purchase frequency feature PG1 g ;

[0026] Among them, t represents the duration corresponding to the specified period;

[0027] StepA11. Get the corresponding dwell time of each category product for each browsing product, and mark it as T k1，g , k1=1、2、……n1 g ;

[0028] StepA12, pass Calculate the dwell time ratio feature T1 corresponding to each category of products g .

[0029] As a further solution of the present invention: the characteristic index analysis is as follows:

[0030] Step B1. Calculation of user behavior characteristic indicators

[0031] Substitute browsing frequency characteristics, purchase frequency characteristics, browsing interval duration characteristics, purchase interval duration characteristics, and stay duration ratio characteristics into a preset behavior characteristic calculation formula, and calculate the user's behavior characteristic index;

[0032] Step B2: Calculation of user transaction characteristic indicators

[0033] Substitute the purchase frequency feature, purchase interval duration feature, and consumption amount feature into the preset transaction feature calculation formula, and calculate the user's transaction feature index;

[0034] Step B3: Calculation of user preference characteristic index

[0035] Substitute the browsing frequency characteristics, purchase frequency characteristics and dwell time ratio characteristics of each category of products into the preset preference processing formula, and calculate the user's preference characteristic value for each type of product;

[0036] Then, from the preference characteristic values ​​for each type of product, select a type of product with the largest preference characteristic value;

[0037] Then, the preference characteristic values ​​of each type of product and the preference characteristic value with the largest value are substituted into the preset preference characteristic calculation formula to calculate the user's preference characteristic index.

[0038] As a further solution of the present invention: the behavior characteristic calculation formula is as follows:

[0039]

[0040] In the formula, TX is the user's behavioral characteristic index, γ1, γ2, γ3, and γ4 are all corresponding preset weight values; β is a fixed value, and β is 1;

[0041] The transaction characteristics calculation formula is as follows:

[0042]

[0043] In the formula, TJ is the user's transaction characteristic index, λ1 and λ2 are the corresponding preset weight values, β is a fixed value, and β is 1;

[0044] The preference processing formula is as follows:

[0045] TP g =PL1 g ×T1 g ×μ1+PG1 g ×μ2

[0046] In the formula, TPg is the user's preference characteristic value for each type of product, μ1 and μ2 are the corresponding preset weight values;

[0047] The formula for calculating preference characteristics is as follows:

[0048]

[0049] In the formula, TP is the user preference characteristic index, TP max is the preference characteristic value with the largest value, β is a fixed value, and β is 1.

[0050] As a further solution of the present invention: the integrated processing method is as follows:

[0051] Then, according to the preset classification threshold set [y1, y2], the indicator threshold sets [TX×y1, TX×y2] and [TJ×y1, TJ×y2] corresponding to the behavior characteristic index and transaction characteristic index, as well as the indicator threshold corresponding to the preference characteristic index are calculated respectively.

[0052] Among them, y1>y2;

[0053] Then, the behavior characteristic index, transaction characteristic index and preference characteristic index are compared with the corresponding indicator threshold set respectively, and based on the comparison results, the user's activity level, consumption level and product type corresponding to the preference mark are determined:

[0054] Then, the user's activity level, consumption level and product type corresponding to the preference mark are used as the user portrait.

[0055] As a further solution of the present invention: wherein, the activity level includes highly active type, moderately active type, and inactive type; the consumption type level includes high consumption type, moderate consumption type, and low consumption type.

[0056] As a further solution of the present invention: the comparison method in the integrated processing is as follows:

[0057] If TX≥TX×y1, the user is judged to be highly active;

[0058] If TX×y1>TX≥TX×y2, ​​the user is judged to be moderately active;

[0059] If TX<TX×y2, ​​the user is judged to be inactive;

[0060] If TJ ≥ TJ × y1, the user is judged to be a high-consumption type;

[0061] If TJ×y1>TJ≥TJ×y2, the user is judged to be a medium-consumption type;

[0062] If TJ<TJ×y2, the user is judged to be a low-consumption type;

[0063] like It is determined that the user has a product type that he / she particularly likes among all types of products, and the product of this type is marked as a favorite;

[0064] like It is determined that the user does not have a product type that he / she particularly likes among all product types, and no preference mark is added to the product of this type.

[0065] Beneficial effects of the present invention:

[0066] The present invention can widely collect a variety of detailed data related to target users and different categories of products, including number of views, number of purchases, timestamps, length of stay and purchase amount, etc., laying a solid foundation for generating accurate and comprehensive user portraits.

[0067] The present invention has a sophisticated data processing flow, such as calculating various frequency characteristics, interval duration characteristics, stay duration ratio characteristics and consumption amount characteristics, which can deeply mine the data value and make the obtained user portrait more accurate.

[0068] The present invention can deeply understand the user's behavior patterns and habits through multi-dimensional analysis of the time interval, frequency, and length of stay of the user's browsing and purchasing behaviors.

[0069] The present invention clearly distinguishes the feature calculation of all categories of products and each category of products, which helps to find the difference in user preferences in overall and specific product categories.

[0070] The present invention presets the calculation formulas of the behavior characteristics, transaction characteristics and preference characteristics and the corresponding weights, which can scientifically and objectively evaluate the characteristic indicators of the user in terms of behavior, transaction and preference.

[0071] The present invention, using these indicators, can clearly outline the user's activity, consumption ability and product preference, and provide enterprises with a clear and quantifiable description of user characteristics.

[0072] The user portraits generated by the present invention, which include activity levels and consumption levels, can help enterprises accurately locate user groups and formulate targeted marketing strategies.

[0073] The present invention can accurately identify and mark the product types preferred by users, which helps enterprises optimize product recommendations and improve sales conversion rates and user satisfaction.

[0074] The present invention uses a cloud platform to perform data processing and portrait generation, which can efficiently process large-scale data, quickly respond to market changes, and improve enterprise operational efficiency.

[0075] The present invention, based on accurate user portraits, enables enterprises to allocate resources more effectively, optimize product design and services, and enhance market competitiveness.

[0076] With the present invention, enterprises can provide personalized services and product recommendations to users based on user portraits, meet users' personalized needs, and improve user experience.

[0077] The present invention can better meet user needs, help improve user loyalty, and promote long-term user participation and consumption.

[0078] The present invention has a flexible and scalable design for the entire system, and can continuously optimize and adjust parameters such as thresholds and weights as the business of the enterprise develops and data accumulates, so as to adapt to the ever-changing market environment and user behavior.

[0079] In summary, the user portrait generation cloud platform in the enterprise digital operation of the present invention can provide enterprises with comprehensive, accurate and in-depth user insights, strongly support the enterprise's decision-making, operation optimization and user service improvement, so as to gain advantages in the fiercely competitive market. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] The present invention will be further described below in conjunction with the accompanying drawings.

[0081] Figure 1 It is a system block diagram of a user portrait generation cloud platform in enterprise digital operation according to the present invention.

[0082] Figure 2 It is a system block diagram of a feature analysis unit in a user portrait generation cloud platform in enterprise digital operations of the present invention. DETAILED DESCRIPTION

[0083] The following will be combined with the drawings in the embodiments of 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.

[0084] Embodiment 1

[0085] See also Figure 1 and Figure 2 As shown, the present invention is a user portrait generation cloud platform in enterprise digital operation, including:

[0086] A data collection unit is used to collect relevant data of target users from various systems within the enterprise and external data sources;

[0087] The relevant data includes the number of views and purchases of products in different categories, as well as the timestamps of each product view and purchase;

[0088] It also includes the corresponding length of stay when browsing products and the purchase amount when purchasing products;

[0089] A data processing unit, used for performing feature processing on relevant data of target users;

[0090] The specific method is as follows:

[0091] First, obtain the number of views and purchases of all categories of products by users in a specified period, and mark them as n and m respectively;

[0092] Then, through PL=n / t and PG=m / t, the browsing frequency characteristics PL and the purchasing frequency characteristics PG of all categories of products of the user in the specified period are calculated respectively;

[0093] Among them, t represents the duration corresponding to the specified period;

[0094] At the same time, obtain the timestamps of each product browsing and the timestamps of each product purchase, and based on them, obtain the interval length between the corresponding adjacent timestamps when each product is browsed and the interval length between the corresponding adjacent timestamps when each product is purchased, and mark them as LS i and GS j , i=1, 2,...n-1, j=1, 2,...m-1;

[0095] Then through and Calculate the average values ​​LS0 and GS0 of the intervals between adjacent timestamps corresponding to each browsing of the product and the intervals between adjacent timestamps corresponding to each purchasing of the product, and then record them as the browsing interval duration feature and the purchase interval duration feature respectively;

[0096] At the same time, the corresponding stay time of each product browsing is obtained and marked as T k , k = 1, 2, ... n;

[0097] Then through Calculate the residence time ratio feature T0;

[0098] A feature analysis unit, used to analyze feature indicators based on the results obtained by the data processing unit and obtain behavior feature indicators;

[0099] The indicator analysis method of behavioral characteristic indicators is as follows:

[0100] Substitute browsing frequency characteristics, purchase frequency characteristics, browsing interval duration characteristics, purchase interval duration characteristics, and stay duration ratio characteristics into a preset behavior characteristic calculation formula, and calculate the user's behavior characteristic index;

[0101] The behavior characteristic calculation formula is as follows:

[0102]

[0103] Wherein, TX is the user's behavior characteristic index, γ1, γ2, γ3, and γ4 are corresponding preset weight values, and β is a fixed value. In this embodiment, β is 1;

[0104] A portrait integration unit is used to integrate and process the behavior feature indicators obtained by the feature analysis unit to obtain a user portrait;

[0105] The integration process is as follows:

[0106] Then, according to the preset classification threshold set [y1, y2], the indicator threshold set [TX×y1, TX×y2] corresponding to the behavior feature indicator is calculated, where y1>y2;

[0107] Then, the behavior characteristic index is compared with the corresponding index threshold set, and the user's activity level is determined based on the comparison result;

[0108] Among them, the activity levels include highly active, moderately active, and inactive;

[0109] If TX≥TX×y1, the user is judged to be highly active;

[0110] If TX×y1>TX≥TX×y2, ​​the user is judged to be moderately active;

[0111] If TX<TX×y2, ​​the user is judged to be inactive;

[0112] The user's activity level is then used as the user profile.

[0113] This embodiment can collect relevant data of target users from multiple sources, provide a comprehensive basis for subsequent analysis, and deeply understand the user's behavior pattern through calculation and analysis of browsing and purchasing frequency, interval duration, and residence time ratio. According to the preset behavior characteristic calculation formula and classification threshold, the user's activity level can be accurately judged, which is helpful for enterprises to formulate targeted interaction strategies. For highly active users, more exclusive services can be provided, for moderately active users, guidance can be strengthened to improve their activity, and for inactive users, incentives can be taken to attract their attention again.

[0114] Embodiment 2

[0115] As the second embodiment of the present invention, when the present application is implemented, compared with the first embodiment, the technical solution of the present embodiment is different from that of the first embodiment only in that:

[0116] In this embodiment, the relevant data also includes the corresponding stay time when browsing products each time and the purchase amount when purchasing products each time;

[0117] The feature processing method is also as follows:

[0118] First, obtain the purchase amount of each product purchase and mark it as Ur, r = 1, 2, ... m;

[0119] Then through Calculate the consumption amount feature U0;

[0120] A feature analysis unit, used to analyze feature indicators based on the results obtained by the data processing unit and obtain transaction feature indicators;

[0121] The indicator analysis method of the transaction characteristic indicator is as follows:

[0122] Substitute the purchase frequency feature, purchase interval duration feature, and consumption amount feature into the preset transaction feature calculation formula, and calculate the user's transaction feature index;

[0123] The transaction characteristics calculation formula is as follows:

[0124]

[0125] Wherein, TJ is the transaction characteristic index of the user, λ1 and λ2 are corresponding preset weight values, β is a fixed value, and in this embodiment, β is 1;

[0126] The integration process is as follows:

[0127] Then, according to the preset classification threshold set [y1, y2], the indicator threshold set [TJ×y1, TJ×y2] corresponding to the transaction characteristic indicator is calculated, where y1>y2;

[0128] Then the transaction characteristic index is compared with the corresponding index threshold set, and the user's consumption level is determined based on the comparison result:

[0129] Among them, the consumption level includes high consumption, medium consumption, and low consumption;

[0130] If TJ ≥ TJ × y1, the user is judged to be a high-consumption type;

[0131] If TJ×y1>TJ≥TJ×y2, the user is judged to be a medium-consumption type;

[0132] If TJ<TJ×y2, the user is judged to be a low-consumption type;

[0133] Then the user's consumption level is used as the user profile.

[0134] Based on the first embodiment, this embodiment adds the important data of purchase amount, so that the feature processing is more comprehensive. Through the analysis of transaction feature indicators, the user's consumption level can be accurately judged. Enterprises can provide differentiated product recommendations and promotion strategies for users with different consumption types based on this, such as providing high-end customized products for high-spending users, providing cost-effective product combinations for medium-spending users, and providing preferential activities for low-spending users to stimulate consumption.

[0135] Embodiment 3

[0136] As the third embodiment of the present invention, when the present application is implemented, compared with the first and second embodiments, the technical solution of this embodiment is different from the first and second embodiments only in that:

[0137] In this embodiment, the feature processing method is also as follows:

[0138] First, obtain the number of views and purchases of each category of products by users in the specified period, and mark them as n1 respectively g and m1 g ; g = 1, 2, ... z, z represents the number of categories of all products;

[0139] Then through PL1 g =n1 g / t and PG1 g =m1 g / t, calculate the browsing frequency characteristics PL1 of each category of products in the specified period g and purchase frequency feature PG1 g ;

[0140] Among them, t represents the duration corresponding to the specified period;

[0141] At the same time, the corresponding stay time of each category product corresponding to each browsing product is obtained and marked as T k1，g , k1=1、2、……n1 g ;

[0142] Then through Calculate the dwell time ratio feature T1 corresponding to each category of products g ;

[0143] A feature analysis unit, used to analyze feature indicators according to the results obtained by the data processing unit, and obtain a preferred feature indicator;

[0144] The indicator analysis method of the preference characteristic indicator is as follows:

[0145] Substitute the browsing frequency characteristics, purchase frequency characteristics and dwell time ratio characteristics of each category of products into the preset preference processing formula, and calculate the user's preference characteristic value for each type of product;

[0146] The preference processing formula is as follows:

[0147] TP g =PL1 g ×T1 g ×μ1+PG1 g ×μ2

[0148] In the formula, TPg is the user's preference characteristic value for each type of product, μ1 and μ2 are the corresponding preset weight values;

[0149] Then, from the preference characteristic values ​​of each type of product, select a type of product with the largest preference characteristic value, and record the preference characteristic value as TP max ;

[0150] Then, the preference characteristic calculation formula is used Calculate the user's preference characteristic index TP, where β is a fixed value. In this embodiment, β is 1;

[0151] The integration process is as follows:

[0152] Then, according to the preset classification threshold set [y1, y2], the index threshold corresponding to the preference feature index is calculated Among them, y1>y2;

[0153] Then, the preference feature index is compared with the corresponding index threshold, and based on the comparison result, the product type corresponding to the user preference mark is determined:

[0154] in,

[0155] like It is determined that the user has a product type that he / she particularly likes among all types of products, and the product of this type is marked as a favorite;

[0156] like It is determined that the user does not have a preferred product type among all product types, and no preference mark is added to the product type;

[0157] In this embodiment, the preference mark is a clear mark indicating that the user is interested or not interested in a certain product, content, service, etc.;

[0158] Then, the product type corresponding to the user's preference tag is used as the user portrait.

[0159] This embodiment further considers the relevant data of each category of products to make the analysis more refined, and can accurately judge the user's preferences for different types of products, thereby providing users with more accurate personalized product recommendations, improving user satisfaction and purchase conversion rate, and determining the product type corresponding to the user's preference mark based on the preference feature index, which helps enterprises optimize product layout and supply chain management.

[0160] Embodiment 4

[0161] As the fourth embodiment of the present invention, when the present application is specifically implemented, compared with the first, second and third embodiments, the technical solution of this embodiment is to combine and implement the solutions of the above-mentioned first, second and third embodiments.

[0162] This embodiment combines all the advantages of embodiments one, two, and three, realizes an all-round, multi-level in-depth analysis of user behaviors, transactions, and preferences, and can generate a comprehensive and accurate user portrait, including activity level, consumption level, and product types corresponding to preference tags, providing more powerful support for the digital operation of enterprises. Enterprises can formulate comprehensive and personalized marketing strategies, product development plans, and user service strategies based on this comprehensive portrait, thereby maximizing user satisfaction, loyalty, and the market competitiveness of enterprises.

[0163] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0164] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A user portrait generation cloud platform in enterprise digital operation, characterized in that: include: A data collection unit, used to collect relevant data of target users; The relevant data includes the number of views and purchases of products of different categories, the timestamps of each product view and purchase, the duration of each product view, and the purchase amount of each product purchase; A data processing unit, used to perform feature processing on the relevant data of the target user, and obtain multiple data features according to the feature processing results; The multiple data features are browsing frequency features, purchase frequency features, browsing interval duration features, purchase interval duration features, stay time ratio features, and consumption amount features corresponding to all categories of products, as well as browsing frequency features, purchase frequency features, and stay time ratio features corresponding to different categories of products; A feature analysis unit is used to perform feature index analysis based on the multiple data features obtained by the data processing unit, and obtain behavior feature indexes, transaction feature indexes, and preference feature indexes corresponding to the user portrait; The analysis method of behavioral characteristic indicators is as follows: Substitute browsing frequency characteristics, purchase frequency characteristics, browsing interval duration characteristics, purchase interval duration characteristics, and stay duration ratio characteristics into a preset behavior characteristic calculation formula, and calculate the user's behavior characteristic index; The behavior characteristic calculation formula is as follows: ; In the formula, TX is the user's behavior characteristic index, γ1, γ2, γ3, and γ4 are corresponding preset weight values, PL is the browsing frequency characteristic, PG is the purchase frequency characteristic, LS0 is the browsing interval duration characteristic, GS0 is the purchase interval duration characteristic, β is a fixed value, and β is 1; The analysis of the trading characteristic indicator is as follows: Substitute the purchase frequency feature, purchase interval duration feature, and consumption amount feature into the preset transaction feature calculation formula, and calculate the user's transaction feature index; The transaction characteristics calculation formula is as follows: ; In the formula, TJ is the user's transaction characteristic index, λ1 and λ2 are the corresponding preset weight values, PG is the purchase frequency characteristic, U0 is the consumption amount characteristic, and GS0 is the purchase interval duration characteristic; The analysis method of preference characteristic indicators is as follows: Substitute the browsing frequency characteristics, purchase frequency characteristics and dwell time ratio characteristics of each category of products into the preset preference processing formula, and calculate the user's preference characteristic value for each type of product; The preference processing formula is as follows: ; In the formula, TPg is the user's preference characteristic value for each type of product, μ1 and μ2 are the corresponding preset weight values, PL1 g PG1 is the browsing frequency characteristics of each category of products by users in a specified period. g T1 is the purchase frequency characteristics of each category of products by users in a specified period, g is the residence time ratio characteristic corresponding to each category of products, where g=1, 2, ... z, and z represents the number of categories of all products; Then, from the preference characteristic values ​​for each type of product, select a type of product with the largest preference characteristic value; Then, the preference characteristic values ​​of each type of product and the maximum preference characteristic value are substituted into the preset preference characteristic calculation formula to calculate the user's preference characteristic index; The formula for calculating preference characteristics is as follows: ; In the formula, TP is the user preference characteristic index, TP max is the preference characteristic value with the largest value, and TPg is the user's preference characteristic value for each type of product; The portrait integration unit is used to integrate and process the multiple feature indicators obtained by the feature analysis unit. It calculates the indicator threshold set of the behavioral feature indicator, the transaction feature indicator and the threshold of the preference feature indicator based on the preset classification threshold set, and then compares each indicator with the corresponding threshold set to determine the user's activity level, consumption level and preferred product type, and use this as the user portrait.

2. The user portrait generation cloud platform in enterprise digital operation according to claim 1, characterized in that: The integration process is as follows: Then, according to the preset classification threshold set [y1, y2], the indicator threshold sets [TX×y1, TX×y2] and [TJ×y1, TJ×y2] corresponding to the behavior characteristic index and transaction characteristic index, as well as the indicator threshold corresponding to the preference characteristic index are calculated respectively. ; Among them, y1>y2; Then, the behavior characteristic index, transaction characteristic index and preference characteristic index are compared with the corresponding indicator threshold set respectively, and based on the comparison results, the user's activity level, consumption level and product type corresponding to the preference mark are determined: Then, the user's activity level, consumption level and product type corresponding to the preference mark are used as the user portrait.

3. The user portrait generation cloud platform in enterprise digital operation according to claim 2 is characterized in that: in, The features are processed as follows: StepA1. Obtain the number of views and purchases of all categories of products by users within a specified period; StepA2: Based on the result of StepA1, the browsing frequency characteristics of all categories of products in the specified period are calculated by dividing the number of browsing times by the duration of the specified period. By dividing the number of purchases by the duration of the specified period, the user's purchase frequency characteristics for all categories of products in the specified period are calculated; Step A3, obtain the timestamps of each product browsing and the timestamps of each product purchase, and obtain the intervals between the corresponding adjacent timestamps of each product browsing and the intervals between the corresponding adjacent timestamps of each product purchase based on the timestamps; StepA4, based on the result of StepA3, calculate the average values ​​of the intervals between the corresponding adjacent timestamps when browsing the product and the intervals between the corresponding adjacent timestamps when purchasing the product, and then record them as the browsing interval duration feature and the purchase interval duration feature respectively; StepA5, obtain the corresponding dwell time of each product browsing; StepA6: Based on the result of StepA5, first calculate the average of the corresponding dwell time of each product browsing, then divide the average by the duration corresponding to the specified period, and record the result value as the dwell time ratio feature; StepA7. Get the purchase amount of each product purchased; StepA8: Based on the result of StepA7, calculate the average purchase amount of each product purchase and record it as the consumption amount feature; StepA9, obtain the number of views and purchases of each category of products by users within a specified period; StepA10, based on the results of StepA9 and combined with the method of StepA2, calculate the browsing frequency characteristics and purchase frequency characteristics of users for each category of products in the specified period; StepA11. Obtain the corresponding dwell time of each category product for each browsing product; StepA12: Based on the results of StepA11 and combined with the method of StepA6, calculate the residence time ratio characteristics corresponding to each category of products.

4. The user portrait generation cloud platform in enterprise digital operation according to claim 3 is characterized in that: in, The activity levels include highly active, moderately active, and inactive; the consumption levels include high consumption, medium consumption, and low consumption.

5. The user portrait generation cloud platform in enterprise digital operation according to claim 4, characterized in that: The comparison method in the integration process is as follows: If TX≥TX×y1, the user is judged to be highly active; If TX×y1>TX≥TX×y2, ​​the user is judged to be moderately active; If TX<TX×y2, ​​the user is judged to be inactive; If TJ ≥ TJ × y1, the user is judged to be a high-consumption type; If TJ×y1>TJ≥TJ×y2, the user is judged to be a medium-consumption type; If TJ<TJ×y2, the user is judged to be a low-consumption type; If TP≥ , it is determined that the user has a favorite product type among all product types, and the product of this type is marked as a favorite; If TP< , it is determined that the user does not have a prominent favorite product type among all types of products, and no favorite mark is added to the product of this type.

Images