A user portrait analysis platform and method based on big data

Abstract

The present invention discloses a user portrait analysis platform and method based on big data, and relates to the technical field of user portrait analysis. The platform discloses a portrait analysis module, a portrait update module, and an emergency analysis module. The portrait analysis module and the portrait update module are provided, and the characteristic data of the user portrait can be regularly deeply analyzed to ensure accurate and reasonable processing of the user portrait. The emergency analysis module is provided, and after determining the special portrait update user, it can be determined in time whether there are users with abnormal changes in potential user portraits, and the portrait analysis of the users with abnormal changes in potential user portraits can be performed in time to avoid the situation that the user portraits of some users are not updated in time after abnormal changes occur.

Description

Technical Field

[0001] The present invention relates to the technical field of user portrait analysis, and more specifically, to a user portrait analysis platform and method based on big data. Background Art

[0002] At present, in order to gain more accurate insights into users and ensure customer experience, enterprises need to collect and update user feature data to build user portraits. The current user portrait analysis platform collects user data regularly and updates user portraits based on the collected data. However, this user portrait update method is relatively simple and cannot be customized for each user. It is easy for some user portraits to change abnormally and the user portraits on the platform are not updated in time. Summary of the invention

[0003] In view of the deficiencies in the prior art, the object of the present invention is to provide a user portrait analysis platform and method based on big data.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] A user portrait analysis platform based on big data, including a portrait analysis module, a portrait update module, and an emergency analysis module;

[0006] The portrait analysis module is used to regularly perform user portrait analysis on users and obtain user portrait analysis logs;

[0007] The portrait update module determines whether to mark the user as a portrait special update user based on the comparison result between the portrait analysis variable value of the portrait analysis log and its threshold value, and sets the portrait update level of the portrait special update user;

[0008] After a special profile update user appears, the emergency analysis module determines whether there is an emergency analysis user, and determines whether to perform user profile analysis on the emergency analysis user.

[0009] Furthermore, user portrait analysis is performed on users regularly to obtain user portrait analysis logs. Specifically, based on a preset cycle, user portrait analysis is performed on users at each cycle node to generate user portrait analysis logs. The portrait analysis logs include user ID, portrait analysis variable value, and portrait analysis time.

[0010] Furthermore, the profile analysis variable value of the profile analysis log is obtained in the following way:

[0011] Obtain user data of the user in the current cycle, synchronously obtain user data of the user in the previous cycle, and mark them as user data sets [A, B], where A is the user data of the user in the current cycle, and B is the user data of the user in the previous cycle, construct feature comparison models of all user feature types, use user data sets [A, B] as input data of feature comparison models of each user feature type, obtain feature comparison values ​​of each user feature type, obtain feature comparison thresholds of each user feature type, compare feature comparison values ​​of corresponding user feature types with feature comparison thresholds, and when feature comparison values ​​are greater than or equal to feature comparison thresholds, mark the user feature type as an abnormal feature type, and generate abnormal feature records of the user feature type, which include user feature types and abnormal marking time. When feature comparison values ​​are less than feature comparison thresholds, no corresponding processing is performed;

[0012] Get all the abnormal feature types of the user, sum up the feature comparison values ​​of all the abnormal feature types and take the average to get the abnormal feature comparison mean, and mark it as Ht, build the knowledge graph of the user feature type, compare all the abnormal feature types in pairs, when two abnormal feature types are associated in the knowledge graph, increase the number of abnormal spreads by one, and calculate the difference between the feature comparison values ​​of the two abnormal feature types and take the absolute value to get the abnormal feature comparison difference, when the two abnormal feature types are not associated in the knowledge graph, no corresponding processing is performed, sum up all the abnormal spread times to get the total number of abnormal spreads, and mark it as Mk, sum up all the abnormal feature comparison differences to get the total abnormal feature comparison difference, and mark it as Fr, use the formula The image analysis variable value Gs is obtained, where a1 is the mean coefficient of the abnormal feature comparison, a2 is the total number of abnormal spread coefficients, and a3 is the total difference coefficient of the abnormal feature comparison.

[0013] Further, based on the comparison result between the portrait analysis variable value of the portrait analysis log and its threshold value, determine whether to mark the user as a portrait special update user, and set the portrait update level of the portrait special update user, specifically: set the portrait analysis threshold value, when the portrait analysis variable value is greater than or equal to the portrait analysis threshold value, mark the user as a portrait special update user, calculate the difference between the portrait analysis variable value of the portrait special update user and the portrait analysis threshold value, obtain the portrait update level value, and mark it as Zi; set the range of each portrait update level value Zi to correspond to a portrait update level, wherein the range of the portrait update level value Zi includes (0, Z1], (Z1, Z2], ..., (Zi-1, Zi], the portrait update level includes portrait update level 1, portrait update level 2, ..., portrait update level i-1, portrait update level i, the higher the portrait update level, the greater the degree of user portrait update of the portrait special update user, and synchronously generate the portrait special update log of the portrait special update user;

[0014] When the portrait analysis variable value is less than the portrait analysis threshold variable value, the portrait analysis log of the user is generated at a preset period.

[0015] Furthermore, the portrait special update log includes the portrait update level and the portrait special update time.

[0016] Furthermore, after a special portrait update user appears, it is determined whether there is an emergency analysis user, and whether to perform user portrait analysis on the emergency analysis user. Specifically, when a user is marked as a special portrait update user, it is determined whether the special portrait update user has a preferred feature type. When the special portrait update user has a preferred feature type, the remaining users with the same preferred feature type as the special portrait update user are marked as emergency analysis users. The portrait change analysis value of the special portrait update user and the portrait change analysis value of the emergency analysis user are obtained, and the difference is calculated to obtain the portrait change analysis difference. A portrait change analysis threshold difference is set. When the portrait change analysis difference is greater than or equal to the portrait change analysis threshold difference, no corresponding processing is performed. When the portrait change analysis difference is less than the portrait change analysis threshold difference, user portrait analysis is synchronously performed on the emergency analysis user.

[0017] When the preferred feature type does not exist for the special profile update user, no corresponding processing is performed.

[0018] Furthermore, the preferred feature type is obtained by the following method: obtaining all abnormal feature records of the user before the current time of the system, obtaining the user feature type of the abnormal feature record, marking the abnormal feature records of the same user feature type as the same type of abnormal feature records, marking the total number of the same type of abnormal feature records as Ty, sorting all the same type of abnormal feature records in the order of the abnormal mark time, calculating the time difference between the abnormal mark times of two adjacent similar abnormal feature records after sorting, obtaining the abnormal mark interval, summing up all the abnormal mark intervals and taking the average, obtaining the average abnormal mark interval, and marking it as Bj, using the formula The prominent feature value Tc of the user feature type is obtained, where b1 is the total number coefficient of similar abnormal feature records, b2 is the average abnormal mark interval coefficient, and a prominent feature threshold is set. When the prominent feature value is greater than or equal to the prominent feature threshold, the user feature type is marked as a prominent feature type. When the prominent feature value is less than the prominent feature threshold, no corresponding processing is performed, and the prominent feature type with the largest prominent feature value is marked as the preferred feature type. When the user does not have a prominent feature type, it is determined that the user does not have a preferred feature type.

[0019] Furthermore, the portrait change analysis value is obtained in the following way: obtain all the portrait analysis logs and portrait special update logs of the user before the current time of the system, sum up the portrait analysis change values ​​of all the portrait analysis logs and take the average value to obtain the average portrait analysis change value, and mark it as Ak; sort all the portrait special update logs according to the order of the portrait special update time, calculate the difference between the portrait update levels of the two adjacent portrait special update logs after sorting, and obtain the portrait update level change value; sum up all the portrait update level change values ​​and take the average value to obtain the average portrait update level change value, and mark it as Fz, and use the formula The user's portrait change analysis value Es is obtained, where c1 is the average portrait analysis variation coefficient and c2 is the average portrait update level variation coefficient.

[0020] Furthermore, a user portrait analysis method based on big data includes the following steps:

[0021] Step 1: Perform user portrait analysis on users regularly and obtain user portrait analysis logs;

[0022] Step 2: Based on the comparison result of the profile analysis variable value of the profile analysis log and its threshold value, determine whether to mark the user as a profile special update user, and set the profile update level of the profile special update user;

[0023] Step 3: After the special profile update user appears, determine whether there is an emergency analysis user, and determine whether to perform user profile analysis on the emergency analysis user.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] 1. The method of the present invention can accurately determine whether there are abnormal changes in user portraits, and grade the abnormally changed user portraits, and take corresponding measures for the user portraits according to different evaluation levels;

[0026] 2. Set up a portrait analysis module and a portrait update module to conduct in-depth analysis of the characteristic data of user portraits on a regular basis to ensure accurate and reasonable processing of user portraits. Set up an emergency analysis module to promptly determine whether there are users with abnormal changes in potential user portraits after determining the users whose portraits need special updates, and promptly conduct portrait analysis on users with abnormal changes in potential user portraits to avoid the situation where the user portraits of some users are not updated in time after abnormal changes occur. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 This is a flowchart of a user portrait analysis method based on big data;

[0028] Figure 2 A flowchart for determining whether to conduct user profiling analysis on emergency analysis users. DETAILED DESCRIPTION

[0029] Example 1

[0030] Reference Figure 1 , a user portrait analysis method based on big data, comprising the following steps:

[0031] Step 1: Perform user portrait analysis on users regularly and obtain user portrait analysis logs;

[0032] Step 2: Based on the comparison result of the profile analysis variable value of the profile analysis log and its threshold value, determine whether to mark the user as a profile special update user, and set the profile update level of the profile special update user;

[0033] Step 3: After the special profile update user appears, determine whether there is an emergency analysis user, and determine whether to perform user profile analysis on the emergency analysis user.

[0034] Example 2

[0035] Reference Figure 2 , a user portrait analysis platform based on big data, including a portrait analysis module, a portrait update module, and an emergency analysis module.

[0036] Portrait analysis module: Based on the preset cycle, at each cycle node, user portrait analysis is performed on the user to generate a user portrait analysis log. The portrait analysis log includes the user ID, portrait analysis variable value, and portrait analysis time (the time when the portrait analysis log is generated).

[0037] The profile analysis variable value of the profile analysis log is obtained in the following way:

[0038] The user data of the user in the current cycle (the user data is the corresponding data generated by the user under transaction behavior, interactive behavior, etc.) is obtained, and the user data of the user in the previous cycle is simultaneously obtained and marked as user data sets [A, B], where A is the user data of the user in the current cycle and B is the user data of the user in the previous cycle. A feature comparison model of all user feature types (user features are used to build user portraits, and user features include multiple types, including but not limited to interest preference feature types, psychological feature types, and social behavior feature types) is constructed. The user data set [A, B] is used as the input data of the feature comparison model of each user feature type to obtain a feature comparison value of each user feature type, obtain a feature comparison threshold of each user feature type, compare the feature comparison value of the corresponding user feature type with the feature comparison threshold, and when the feature comparison value is greater than or equal to the feature comparison threshold, mark the user feature type as a mutated feature type, and generate a mutated feature record of the user feature type. The mutated feature record includes the user feature type and the mutated marking time (the time when the user feature type is marked as a mutated feature type). When the feature comparison value is less than the feature comparison threshold, no corresponding processing is performed.

[0039] Get all the abnormal feature types of the user, sum up the feature comparison values ​​of all the abnormal feature types and take the average to get the abnormal feature comparison mean, and mark it as Ht, build the knowledge graph of the user feature type, compare all the abnormal feature types in pairs, when two abnormal feature types are associated in the knowledge graph, increase the number of abnormal spreads by one, and calculate the difference between the feature comparison values ​​of the two abnormal feature types and take the absolute value to get the abnormal feature comparison difference, when the two abnormal feature types are not associated in the knowledge graph, no corresponding processing is performed, sum up all the abnormal spread times to get the total number of abnormal spreads, and mark it as Mk, sum up all the abnormal feature comparison differences to get the total abnormal feature comparison difference, and mark it as Fr, use the formula The portrait analysis variable Gs is obtained, where a1 is the mean coefficient of the abnormal feature comparison, a2 is the total number of abnormal spread coefficient, and a3 is the total difference coefficient of the abnormal feature comparison. The value of a1 is 0.82, the value of a2 is 0.68, and the value of a3 is 0.47.

[0040] Portrait update module: Set the portrait analysis threshold value. The portrait analysis threshold value is the system setting threshold and can be modified according to actual needs;

[0041] When the portrait analysis variable value is greater than or equal to the portrait analysis threshold variable value, the user is marked as a portrait special update user, and the difference between the portrait analysis variable value of the portrait special update user and the portrait analysis threshold variable value is calculated to obtain a portrait update level value, and marked as Zi; the range of each portrait update level value Zi is set to correspond to a portrait update level, wherein the range of the portrait update level value Zi includes (0, Z1], (Z1, Z2], …, (Zi-1, Zi], and the portrait update level includes portrait update level 1, portrait update level 2, …, portrait update level i-1, portrait update level i, when When Zi∈(0, Z1], the portrait special update user is marked as portrait update level 1. The higher the portrait update level, the greater the degree of update of the user portrait of the portrait special update user. A portrait special update log is generated for the portrait special update user. The portrait special update log includes the portrait update level and the portrait special update time (i.e., the time when the portrait special update log is generated). Corresponding measures can be taken for the portrait special update user according to the portrait update level, such as shortening the user portrait update time of the portrait special update user, such as reconstructing the user portrait of the portrait special update user.

[0042] When the portrait analysis variable value is less than the portrait analysis threshold variable value, the portrait analysis log of the user is generated at a preset period.

[0043] Each user feature type is represented in the form of a knowledge graph, and the association in the knowledge graph represents the influence relationship of each user feature type. For example, changes in interest preferences (feature type) will cause changes in social behavior (feature type), so there is an influence relationship between interest preferences (feature type) and social behavior (feature type).

[0044] Taking the interest preference feature type as an example, the feature comparison model of the interest preference feature type is obtained by the following method: obtaining multiple groups of user data sets [A, B], the user data in the user data sets [A, B] can be real data or system preset data, performing interest preference feature extraction on the multiple groups of user data sets [A, B] to obtain multiple groups of interest preference feature sets [A, B], constructing a neural network model, using the interest preference feature set [A, B] as training data for the neural network model, assigning feature comparison values ​​to the training data, dividing the training data into a training set and a validation set according to a set ratio, performing iterative neural network training on the training set and the validation set, and obtaining a feature comparison model of the interest preference feature type, wherein the feature comparison value is greater than 0, the larger the feature comparison value is, the greater the change in the user's interest preference in the current cycle is compared with the previous cycle, and the smaller the feature comparison value is, the smaller the change in the user's interest preference in the current cycle is compared with the previous cycle.

[0045] The feature comparison models of other user feature types are obtained in the same way as the feature comparison models of interest preference feature types. Different user feature types correspond to a feature comparison threshold.

[0046] By setting up a portrait analysis module and a portrait update module, you can regularly conduct in-depth analysis of the feature data of the user portrait, accurately determine whether there are abnormal changes in the user portrait, and rate the user portraits with abnormal changes. You can take corresponding measures for the user portraits based on different rating levels to ensure accurate and reasonable processing of the user portraits.

[0047] Emergency analysis module: After a user is marked as a special portrait update user, determine whether the special portrait update user has a preferred feature type. When the special portrait update user has a preferred feature type, mark the remaining users with the same preferred feature type as the special portrait update user as emergency analysis users, obtain the portrait change analysis value of the special portrait update user and the portrait change analysis value of the emergency analysis user, perform difference calculation to obtain the portrait change analysis difference, set the portrait change analysis threshold difference, which is the system-set threshold and can be modified according to actual needs. When the portrait change analysis difference is greater than or equal to the portrait change analysis threshold difference, no corresponding processing is performed. When the portrait change analysis difference is less than the portrait change analysis threshold difference, perform user portrait analysis on the emergency analysis user simultaneously.

[0048] When the preferred feature type does not exist for the special profile update user, no corresponding processing is performed.

[0049] The preferred feature type is obtained in the following way: obtain all abnormal feature records of the user before the current time of the system, obtain the user feature type of the abnormal feature record, mark the abnormal feature records of the same user feature type as the same type of abnormal feature records, mark the total number of the same type of abnormal feature records as Ty, sort all the same type of abnormal feature records in the order of the abnormal mark time, calculate the time difference between the abnormal mark times of two adjacent similar abnormal feature records after sorting, and obtain the abnormal mark interval, sum up all the abnormal mark intervals and take the average to obtain the average abnormal mark interval, and mark it as Bj, and use the formula The prominent feature value Tc of the user feature type is obtained, where b1 is the total number coefficient of similar abnormal feature records, b2 is the average abnormal mark interval coefficient, c1 is 0.73, c2 is 0.56, and a prominent feature threshold is set. When the prominent feature value is greater than or equal to the prominent feature threshold, the user feature type is marked as a prominent feature type. When the prominent feature value is less than the prominent feature threshold, no corresponding processing is performed. The prominent feature type with the largest prominent feature value is marked as the preferred feature type. When the user does not have a prominent feature type, it is determined that the user does not have a preferred feature type.

[0050] The portrait change analysis value is obtained in the following way: obtain all the portrait analysis logs and portrait special update logs of the user before the current time of the system, sum up the portrait analysis change values ​​of all the portrait analysis logs and take the average value to obtain the average portrait analysis change value, and mark it as Ak; sort all the portrait special update logs according to the order of the portrait special update time, calculate the difference between the portrait update levels of the two adjacent portrait special update logs after sorting, and obtain the portrait update level change value; sum up all the portrait update level change values ​​and take the average value to obtain the average portrait update level change value, and mark it as Fz, and use the formula The user's portrait change analysis value Es is obtained, where c1 is the average portrait analysis variation coefficient, c2 is the average portrait update level variation coefficient, the value of c1 is 0.46, and the value of c2 is 0.52.

[0051] By setting up an emergency analysis module, after determining the users with special portrait updates, it can be determined in a timely manner whether there are users with abnormal changes in potential user portraits, and portrait analysis can be performed in a timely manner on users with abnormal changes in potential user portraits to avoid the situation where the user portraits of some users are not updated in a timely manner after abnormal changes occur.

[0052] 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 in the formula are set by technicians in this field according to actual conditions.

[0053] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0054] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0055] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0056] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0057] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0058] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks or optical disks.

[0059] 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 analysis platform based on big data, characterized in that: Includes portrait analysis module, portrait update module, and emergency analysis module; The portrait analysis module is used to regularly perform user portrait analysis on users and obtain user portrait analysis logs; The portrait update module determines whether to mark the user as a portrait special update user based on the comparison result between the portrait analysis variable value of the portrait analysis log and its threshold value, and sets the portrait update level of the portrait special update user; The emergency analysis module determines whether there is an emergency analysis user after a special profile update user appears, and determines whether to perform user profile analysis on the emergency analysis user; After a special portrait update user appears, determine whether there is an emergency analysis user, and determine whether to perform user portrait analysis on the emergency analysis user. Specifically, when a user is marked as a special portrait update user, determine whether the special portrait update user has a preferred feature type. When the special portrait update user has a preferred feature type, mark the remaining users with the same preferred feature type as the special portrait update user as emergency analysis users, perform a difference calculation between the portrait change analysis value of the special portrait update user and the portrait change analysis value of the emergency analysis user to obtain a portrait change analysis difference, set a portrait change analysis threshold difference, and when the portrait change analysis difference is greater than or equal to the portrait change analysis threshold difference, do not perform corresponding processing. When the portrait change analysis difference is less than the portrait change analysis threshold difference, synchronously perform user portrait analysis on the emergency analysis user. When the user does not have the preferred feature type for the special portrait update, no corresponding processing is performed; The preferred feature type is obtained in the following way: obtain all abnormal feature records of the user before the current time of the system, obtain the user feature type of the abnormal feature record, mark the abnormal feature records of the same user feature type as the same type of abnormal feature records, mark the total number of the same type of abnormal feature records as Ty, sort all the same type of abnormal feature records in the order of the abnormal mark time, calculate the time difference between the abnormal mark times of two adjacent similar abnormal feature records after sorting, and obtain the abnormal mark interval, sum up all the abnormal mark intervals and take the average to obtain the average abnormal mark interval, and mark it as Bj, and use the formula Obtain the prominent feature value Tc of the user feature type, where b1 is the total number coefficient of similar abnormal feature records, b2 is the average abnormal mark interval coefficient, set the prominent feature threshold, when the prominent feature value is greater than or equal to the prominent feature threshold, mark the user feature type as a prominent feature type, when the prominent feature value is less than the prominent feature threshold, do not do the corresponding processing, mark the prominent feature type with the largest prominent feature value as the preferred feature type, when the user does not have a prominent feature type, it is determined that the user does not have a preferred feature type; The portrait change analysis value is obtained in the following way: obtain all the portrait analysis logs and portrait special update logs of the user before the current time of the system, sum up the portrait analysis change values ​​of all the portrait analysis logs and take the average value to obtain the average portrait analysis change value, and mark it as Ak; sort all the portrait special update logs according to the order of the portrait special update time, calculate the difference between the portrait update levels of the two adjacent portrait special update logs after sorting, and obtain the portrait update level change value; sum up all the portrait update level change values ​​and take the average value to obtain the average portrait update level change value, and mark it as Fz, and use the formula The user's portrait change analysis value Es is obtained, where c1 is the average portrait analysis variation coefficient and c2 is the average portrait update level variation coefficient.

2. A user portrait analysis platform based on big data according to claim 1, characterized in that: Regularly conduct user portrait analysis on users and obtain user portrait analysis logs. Specifically: based on the preset cycle, at each cycle node, conduct user portrait analysis on users and generate user portrait analysis logs. The portrait analysis logs include user ID, portrait analysis variable value, and portrait analysis time.

3. A user portrait analysis platform based on big data according to claim 2, characterized in that: The profile analysis variable value of the profile analysis log is obtained in the following way: Obtain user data of the user in the current cycle, synchronously obtain user data of the user in the previous cycle, and mark them as user data sets [A, B], where A is the user data of the user in the current cycle, and B is the user data of the user in the previous cycle, construct feature comparison models of all user feature types, use user data sets [A, B] as input data of feature comparison models of each user feature type, obtain feature comparison values ​​of each user feature type, obtain feature comparison thresholds of each user feature type, compare feature comparison values ​​of corresponding user feature types with feature comparison thresholds, and when feature comparison values ​​are greater than or equal to feature comparison thresholds, mark the user feature type as an abnormal feature type, and generate abnormal feature records of the user feature type, which include user feature types and abnormal marking time. When feature comparison values ​​are less than feature comparison thresholds, no corresponding processing is performed; Get all the abnormal feature types of the user, sum up the feature comparison values ​​of all the abnormal feature types and take the average to get the abnormal feature comparison mean, and mark it as Ht, build the knowledge graph of the user feature type, compare all the abnormal feature types in pairs, when two abnormal feature types are associated in the knowledge graph, increase the number of abnormal spreads by one, and calculate the difference between the feature comparison values ​​of the two abnormal feature types and take the absolute value to get the abnormal feature comparison difference, when the two abnormal feature types are not associated in the knowledge graph, no corresponding processing is performed, sum up all the abnormal spread times to get the total number of abnormal spreads, and mark it as Mk, sum up all the abnormal feature comparison differences to get the total abnormal feature comparison difference, and mark it as Fr, use the formula The image analysis variable value Gs is obtained, where a1 is the mean coefficient of the abnormal feature comparison, a2 is the total number of abnormal spread coefficients, and a3 is the total difference coefficient of the abnormal feature comparison.

4. The user portrait analysis platform based on big data according to claim 1, characterized in that: Based on the comparison result of the portrait analysis variable value and the threshold value of the portrait analysis log, determine whether to mark the user as a portrait special update user, and set the portrait update level of the portrait special update user, specifically: set the portrait analysis threshold value, when the portrait analysis variable value is greater than or equal to the portrait analysis threshold value, mark the user as a portrait special update user, calculate the difference between the portrait analysis variable value of the portrait special update user and the portrait analysis threshold value, obtain the portrait update level value, and mark it as Zi; Set the range of each portrait update level value Zi to correspond to a portrait update level, wherein the range of the portrait update level value Zi includes (0, Zi1], (Z1, Zi2], ..., (Zi-1, Zi], and the portrait update level includes portrait update level 1, portrait update level 2, ..., portrait update level i-1, and portrait update level i. The higher the portrait update level, the greater the degree of update of the user portrait of the portrait special update user, and the portrait special update log of the portrait special update user is generated synchronously; When the portrait analysis variable value is less than the portrait analysis threshold variable value, the portrait analysis log of the user is generated at a preset period.

5. A user portrait analysis platform based on big data according to claim 4, characterized in that: The portrait special update log includes the portrait update level and the portrait special update time.

6. A user portrait analysis method based on big data, applied to the user portrait analysis platform based on big data described in claim 1, characterized in that: The steps include: Step 1: Perform user portrait analysis on users regularly and obtain user portrait analysis logs; Step 2: Based on the comparison result of the profile analysis variable value of the profile analysis log and its threshold value, determine whether to mark the user as a profile special update user, and set the profile update level of the profile special update user; Step 3: After the special profile update user appears, determine whether there is an emergency analysis user, and determine whether to perform user profile analysis on the emergency analysis user.

Images