User Portrait Generation Method

Abstract

The present invention discloses a method for generating a user portrait. The method includes: receiving a user portrait generation instruction, and obtaining a target user corresponding to the user portrait generation instruction; according to the target user, obtaining a standard portrait model corresponding to the target user, and based on a pre-stored portrait adjustment database, calling adjustment parameters to be adjusted that match the target user; using the called adjustment parameters to adjust the standard portrait model to generate a user portrait that matches the target user; achieving the purpose of adjusting the standard user portrait model according to the adjustment parameters to be adjusted that match the target user, and improving the accuracy of user portrait matching; further, providing important data support for subsequent big data push.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to a method for generating a user portrait. Background Art

[0002] In today's big data era, since big data technology can take all data resources of any system as the object and discover the correlation relationships shown between the data, big data technology has currently been widely applied in aspects such as the process optimization of the Internet, targeted messages and advertisement push, user personalized services and improvement, etc., providing strong data support for Internet services.

[0003] A user portrait, also known as a user role, is an effective tool for depicting target users and connecting user demands with design directions, and is an important application of big data technology. Therefore, user portraits have been widely applied in various fields. Currently, when using relevant user portraits, a unified standard user portrait model is usually directly called; since the user feature information corresponding to different target users is not exactly the same, directly calling a unified user portrait model as the user portrait corresponding to different target users is likely to result in the situation where the user portrait cannot fully match the target user, thereby causing inaccurate push of big data. Summary of the Invention

[0004] The present invention provides a method for generating a user portrait, aiming to adjust the corresponding standard user portrait model according to different target users so that the generated user portrait can match the corresponding target user.

[0005] The present invention provides a method for generating a user portrait, and the method for generating a user portrait includes:

[0006] Receiving a user portrait generation instruction, and obtaining the target user corresponding to the user portrait generation instruction;

[0007] According to the target user, obtaining the standard portrait model corresponding to the target user, and based on a pre-stored portrait adjustment database, calling the adjustment parameters to be adjusted that match the target user;

[0008] Using the called adjustment parameters to be adjusted to adjust the standard portrait model to generate a user portrait that matches the target user.

[0009] Further, the method for generating a user portrait further includes:

[0010] Based on the standard portrait model, obtaining the adjustment parameters to be adjusted corresponding to different target users, and storing the adjustment parameters to be adjusted to form the corresponding portrait adjustment database.

[0011] Furthermore, the obtaining of parameters to be adjusted corresponding to different target users based on the standard portrait model includes:

[0012] Acquire behavior information corresponding to the target user, and construct a feature vector corresponding to the behavior feature of the target user;

[0013] According to the constructed eigenvector, calculating the eigenvalue corresponding to the eigenvector;

[0014] Collect multiple feature information of the target user in different dimensions and establish a landmark tag corresponding to the target user;

[0015] According to the landmark tag, the parameter to be adjusted corresponding to the target user is obtained by utilizing the calculated characteristic value.

[0016] Furthermore, the acquiring behavior information corresponding to the target user and constructing a feature vector corresponding to the behavior information of the target user includes:

[0017] Acquire behavior information corresponding to the target user, and extract behavior features from the behavior information to obtain a corresponding behavior feature set;

[0018] Based on the behavior feature set, a feature vector corresponding to each behavior feature in the behavior feature set is constructed.

[0019] Furthermore, the step of calculating the eigenvalue corresponding to the eigenvector according to the constructed eigenvector comprises steps A1-A5:

[0020] Step A1: Based on the constructed n feature vectors H i , using formula (1), calculate the average eigenvector of the eigenvectors Then we have:

[0021]

[0022] In formula (1), H i represents the feature vector of the i-th behavior feature, and n represents the total number of feature vectors corresponding to the behavior feature; Denotes n eigenvectors H i The average eigenvector of

[0023] Step A2: Using formula (2), obtain the first eigenvalue corresponding to the eigenvector Then we have:

[0024]

[0025] In formula (2), represents the first eigenvalue;

[0026] Step A3: Using formula (3), calculate the second eigenvalue of each of the said eigenvectors, then we have:

[0027] |λ i E - H i | = 0 (3)

[0028] In formula (3), λ i represents the second eigenvalue of the i-th eigenvector; E is the identity matrix, and H i represents the eigenvector of the i-th behavioral feature;

[0029] Step A4: Using formula (4), calculate the average eigenvalue λ corresponding to the second eigenvalue, then we have:

[0030]

[0031] In formula (4), n represents the total number of the said eigenvectors;

[0032] Step A5: Take the calculated first eigenvalue and average eigenvalue as the eigenvalue corresponding to the said eigenvector.

[0033] Further, the collecting multiple feature information of the target user in different dimensions and establishing the landmark label corresponding to the target user includes:

[0034] Collect information of the target user in different dimensions to obtain multiple feature information corresponding to the target user in different dimensions;

[0035] Extract landmark information from the obtained multiple feature information, and establish a landmark label for each extracted landmark information to obtain the landmark label corresponding to the target user.

[0036] Further, the obtaining the parameter to be adjusted corresponding to the target user by using the calculated eigenvalue according to the landmark label includes:

[0037] Perform data processing on all the said landmark labels according to all the said landmark labels to generate the first portrait corresponding to the target user;

[0038] Perform correction processing on the first portrait according to the calculated first eigenvalue and average eigenvalue to obtain the second portrait corresponding to the target user;

[0039] Perform matching processing on the second portrait and the standard portrait model corresponding to the target user to obtain the difference information between the second portrait and the standard portrait model;

[0040] Use the difference information as the parameter to be adjusted for adjusting the standard portrait model.

[0041] Further, the generating, according to all the landmark tags, a first portrait corresponding to the target user by performing data processing on the landmark tags includes:

[0042] According to all the landmark tags, perform data processing on the landmark tags, and use formula (5) to calculate the comprehensive information value S of the first portrait, then:

[0043]

[0044] In formula (5), K represents the total number of the landmark tags; τ k1 represents the attribute value corresponding to the k1-th landmark tag; υ k1 represents the correction value corresponding to the k1-th landmark tag; y k1 represents the weight value corresponding to the k1-th landmark tag.

[0045] Further, the obtaining, according to the calculated first eigenvalue and average eigenvalue, a second portrait corresponding to the target user by performing correction processing on the first portrait includes steps B1 - B3:

[0046] Step B1: Perform correction processing on the first portrait to obtain a corrected portrait, and use formula (6) to calculate the comprehensive information value Y of the corrected portrait;

[0047] Among them, formula (6) is:

[0048]

[0049] In formula (6), S represents the comprehensive information value of the first portrait; Y represents the comprehensive information value of the corrected portrait after performing correction processing on the first portrait; λ0 represents a preset initial portrait eigenvalue, and its value range is [0, 1]; sim() represents a similarity function; |sim()| represents the eigenvalue of the similarity vector corresponding to the similarity function; represents the first eigenvalue; λ represents the average eigenvalue;

[0050] Step B2: According to the comprehensive information value Y of the corrected portrait, search for and obtain a pre-stored target tag set that matches the comprehensive information value Y of the corrected portrait;

[0051] Step B3: According to the obtained target tag set, obtain the second portrait corresponding to the target user and including the target tag set.

[0052] Furthermore, the user portrait generation method further includes:

[0053] According to the generated user portrait, the standard portrait model matching the target user is updated to the newly generated user portrait;

[0054] The step of updating the standard portrait model to the newly generated user portrait includes:

[0055] According to the generated user portrait, the standard portrait model matching the target user is replaced with the user portrait, and the replaced and updated user portrait is used as a new standard portrait model matching the target user, while retaining the corresponding update timestamp information;

[0056] Based on the replacement updated new standard portrait model and the update timestamp information, the matching relationship between the parameters to be adjusted that match the target user and the target user is released.

[0057] The user portrait generation method of the present invention receives a user portrait generation instruction to obtain a target user corresponding to the user portrait generation instruction; according to the target user, a standard portrait model corresponding to the target user is obtained, and based on a pre-stored portrait adjustment database, parameters to be adjusted that match the target user are called; the standard portrait model is adjusted using the called parameters to be adjusted to generate a user portrait that matches the target user; the purpose of adjusting the standard user portrait model according to the parameters to be adjusted that match the target user is achieved, and the accuracy of user portrait matching is improved; further, important data support is provided for the subsequent push of big data.

[0058] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the contents pointed out in the written description, claims, and drawings.

[0059] The technical solution of the present invention is further described below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0061] Figure 1 It is a schematic diagram of the workflow of an implementation method of the user portrait generation method of the present invention.

[0062] Figure 2It is a schematic diagram of the work flow of another implementation manner of the user portrait generation method of the present invention.

[0063] Figure 3 It is a schematic diagram of the work flow of an implementation manner of obtaining the parameters to be adjusted corresponding to different target users according to the standard portrait model in the user portrait generation method of the present invention. Specific implementation manner

[0064] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0065] The present invention provides a user portrait generation method. By calling the parameters to be adjusted corresponding to different target users, and then using the parameters to be adjusted to adjust the standard user portrait model, a user portrait that can be well matched with the corresponding target user is obtained, thereby improving the accuracy of user portrait matching. Further, it provides important data support and push basis for big data push.

[0066] As Figure 1 shown, Figure 1 It is a schematic diagram of the work flow of an implementation manner of the user portrait generation method of the present invention; the user portrait generation method of the present invention can be implemented as steps S10-S30 described below.

[0067] Step S10: Receive a user portrait generation instruction, and obtain the target user corresponding to the user portrait generation instruction.

[0068] In the embodiment of the present invention, when the system receives a triggered user portrait generation instruction, it responds to the user portrait generation instruction and obtains the target user pointed to by the user portrait generation instruction. Among them, obtaining the target user includes but is not limited to: the user type of the target user, the positioning population corresponding to the target user, the behavior characteristic information of the target user, the preference information and behavior habit information of the target user, etc.

[0069] Step S20: According to the target user, obtain the standard portrait model corresponding to the target user, and based on the pre-stored portrait adjustment database, call the parameters to be adjusted that match the target user.

[0070] In the embodiments of the present invention, in the big data era, since the system has pre-set standard portrait models that match different types of target users for different types of target users, based on the obtained target user, the user type of the target user can be known; and then according to the user type, the standard portrait model pre-set by the system that matches the user type of the target user can be found and called through searching, that is, the standard portrait model corresponding to the target user is obtained.

[0071] Meanwhile, due to the different focus points, different user behavior characteristic information and other difference information of the target users of the same type, for a specific target user, the system has pre-stored a portrait adjustment database for different target users to adjust their matching standard user portrait models. In the portrait adjustment database pre-stored by the system, the to-be-adjusted parameters that match different target users are stored. According to the characteristic information corresponding to the target user, the to-be-adjusted parameters that match the target user can be found in the portrait adjustment database.

[0072] Step S30: Use the called to-be-adjusted parameters to adjust the standard portrait model to generate a user portrait that matches the target user.

[0073] According to the called to-be-adjusted parameters that match the target user, the standard portrait model corresponding to the target user is adjusted, so as to obtain a user portrait that matches the target user. According to the obtained user portrait, the system can further perform operations such as information push and solution customization for the user portrait.

[0074] Further, in one embodiment, after the user portrait generation method generates a user portrait that matches the target user, it further includes step S31.

[0075] Step S31: According to the generated user portrait, update the standard portrait model that matches the target user to the newly generated user portrait.

[0076] By directly updating the corresponding standard portrait model based on the newly generated user portrait, when the system obtains the corresponding user portrait for the same target user subsequently, the updated user portrait that matches the target user can be directly obtained through data searching.

[0077] Furthermore, the updated user portrait can also be used as a new standard portrait model that matches the target user. As time accumulates, when a user portrait generation instruction for the target user is received again after a certain period of time, the new standard portrait model can be adjusted based on the parameters to be adjusted corresponding to the newly generated user data of the target user, thereby generating a more accurate user portrait that better matches the real-time data of the current target user.

[0078] Furthermore, in one embodiment, the updating of the standard portrait model matching the target user to the newly generated user portrait according to the generated user portrait can be implemented by the following technical means:

[0079] According to the generated user portrait, the user portrait is used to replace the standard portrait model that matches the target user, and the replaced and updated user portrait is used as the new standard portrait model that matches the target user, while retaining the corresponding update timestamp information; based on the replaced new standard portrait model and the update timestamp information, the matching relationship between the parameters to be adjusted that match the target user and the target user is released.

[0080] In an embodiment of the present invention, for operations in which the standard portrait model carries update timestamp information, when it is necessary to generate a user portrait of the same target user again, the system can determine whether to directly use the standard portrait model based on the update timestamp information, thereby saving data processing time and improving the efficiency of obtaining user portraits while ensuring the accuracy of the user portraits.

[0081] The user portrait generation method of the present invention receives a user portrait generation instruction to obtain a target user corresponding to the user portrait generation instruction; according to the target user, a standard portrait model corresponding to the target user is obtained, and based on a pre-stored portrait adjustment database, parameters to be adjusted that match the target user are called; the standard portrait model is adjusted using the called parameters to be adjusted to generate a user portrait that matches the target user; the purpose of adjusting the standard user portrait model according to the parameters to be adjusted that match the target user is achieved, and the accuracy of user portrait matching is improved; further, important data support is provided for the subsequent push of big data.

[0082] based on Figure 1 The description of the embodiment is as follows: Figure 2 As shown, Figure 2 It is a workflow diagram of another implementation of the user portrait generation method of the present invention. Figure 2 The embodiment described in Figure 1Before the step "S10 in the embodiment: Receive a user profile generation instruction and obtain a target user corresponding to the user profile generation instruction", there is also a step S40.

[0083] Step S40: Based on the standard profile model, obtain adjustment parameters corresponding to different target users, and store the adjustment parameters to form the corresponding profile adjustment database.

[0084] In the embodiment of the present invention, step S40 can be run when the profile generation method of the present invention is first run, or according to system requirements, after a certain historical duration, when updating the profile adjustment database, step S40 is run.

[0085] In the embodiment of the present invention, for different standard profile models, obtain adjustment parameters corresponding to different target users that match the standard profile model, and store the above adjustment parameters to construct the corresponding profile adjustment database. Then, when it is necessary to generate a user profile of the target user subsequently, the adjustment parameters that match the target user and the corresponding standard profile model can be called, and the standard profile model is adjusted using the adjustment parameters, so that the generated user profile is more adapted to and more accurate for the target user.

[0086] Further, in one embodiment, as Figure 3 shown, Figure 3 is a schematic flowchart of a working process of an implementation manner of obtaining adjustment parameters corresponding to different target users according to a standard profile model in the user profile generation method of the present invention. Figure 3 In the embodiment, Figure 2 In step S40 in the embodiment, the obtaining of adjustment parameters corresponding to different target users according to the standard profile model can be implemented as steps S11 - S14 described below.

[0087] Step S11: Obtain the behavior information corresponding to the target user and construct a feature vector corresponding to the behavior characteristics of the target user.

[0088] In the embodiment of the present invention, the behavior information corresponding to the target user obtained by the system includes but is not limited to: behavior event information, behavior characteristic information, behavior habit information, and preference information corresponding to the behavior operation events triggered by the target user in relevant application scenarios. Then, according to the behavior information, a feature vector corresponding to the behavior characteristics of the target user is constructed.

[0089] In one embodiment, the system obtains the behavior information corresponding to the target user, extracts the behavior features in the obtained behavior information of the target user according to the obtained behavior information, and obtains a corresponding behavior feature set; furthermore, based on the behavior feature set, a feature vector corresponding to each behavior feature in the behavior feature set is constructed.

[0090] Step S12: Calculate the eigenvalue corresponding to the constructed feature vector.

[0091] Step S13: Collect multiple feature information of the target user in different dimensions, and establish a landmark label corresponding to the target user.

[0092] In one embodiment, when the system collects feature information, it can collect information about the target user in different dimensions, obtain multiple feature information corresponding to the target user in different dimensions; and extract landmark information from the obtained multiple feature information, and establish a landmark label for each extracted landmark information, to obtain the landmark label corresponding to the target user.

[0093] Step S14: According to the landmark label, use the calculated eigenvalue to obtain the parameter to be adjusted corresponding to the target user.

[0094] In step S12 of the embodiment of the present invention, the system calculates the eigenvalue corresponding to the constructed feature vector, including the first eigenvalue and the average eigenvalue.

[0095] In one embodiment, the system calculates the eigenvalue corresponding to the constructed feature vector, including steps A1 - A5:

[0096] Step A1: According to the n constructed feature vectors H i , use formula (1) to calculate the average feature vector of the feature vectors Then there is:

[0097]

[0098] In formula (1), H i represents the feature vector of the i-th behavior feature, and n represents the total number of feature vectors corresponding to the behavior features; represents the average feature vector of the n feature vectors H i ;

[0099] Step A2: Use formula (2) to obtain the first eigenvalue corresponding to the feature vector Then there is:

[0100]

[0101] In formula (2), represents the first eigenvalue;

[0102] Step A3: Using formula (3), calculate the second eigenvalue of each of the eigenvectors respectively, then there is:

[0103] |λ i E - H i | = 0 (3)

[0104] In formula (3), λ i represents the second eigenvalue of the i-th eigenvector; E is the identity matrix, and H i represents the eigenvector of the i-th behavioral feature;

[0105] Step A4: Using formula (4), calculate the average eigenvalue λ corresponding to the second eigenvalue, then there is:

[0106]

[0107] In formula (4), n represents the total number of the eigenvectors;

[0108] Step A5: Take the calculated first eigenvalue and the average eigenvalue as the eigenvalue corresponding to the eigenvector.

[0109] Furthermore, based on the calculated first eigenvalue and the average eigenvalue λ, according to the signature label, obtain the parameter to be adjusted corresponding to the target user, which can be implemented by the following technical means:

[0110] According to all the signature labels, perform data processing on the signature labels to generate the first portrait corresponding to the target user; according to the calculated first eigenvalue and average eigenvalue, perform correction processing on the first portrait to obtain the second portrait corresponding to the target user; perform matching processing on the second portrait and the standard portrait model corresponding to the target user to obtain the difference information between the second portrait and the standard portrait model; take the difference information as the parameter to be adjusted for adjusting the standard portrait model.

[0111] Furthermore, in one embodiment, the performing data processing on all the signature labels to generate the first portrait corresponding to the target user includes:

[0112] According to all the signature labels, perform data processing on the signature labels, and using formula (5), calculate the comprehensive information value S of the first portrait, then there is:

[0113]

[0114] In formula (5), K represents the total number of the said landmark tags; τ k1 represents the attribute value corresponding to the k1-th landmark tag. The attribute corresponding to the landmark tag can be the preset attribute value corresponding to the category to which the classification belongs under the classification to which the characteristic information corresponding to the tag belongs, which is a preset value, and the value range is [1, 10]; υ k1 represents the correction value corresponding to the k1-th landmark tag, and the value range is [1, 5]; y k1 represents the weight value corresponding to the k1-th landmark tag, and the value range is (0, 1).

[0115] Furthermore, in one embodiment, the step of correcting the first portrait according to the calculated first eigenvalue and average eigenvalue to obtain the second portrait corresponding to the target user includes steps B1 - B3:

[0116] Step B1: Obtain a corrected portrait by correcting the first portrait, and use formula (6) to calculate the comprehensive information value Y of the corrected portrait;

[0117] wherein, the formula (6) is:

[0118]

[0119] In formula (6), S represents the comprehensive information value of the first portrait; Y represents the comprehensive information value of the corrected portrait after correcting the first portrait; λ0 represents a preset initial portrait eigenvalue, and its value range is [0, 1]; sim() represents a similarity function; |sim()| represents the eigenvalue of the similarity vector corresponding to the similarity function; represents the first eigenvalue; λ represents the average eigenvalue;

[0120] Step B2: According to the comprehensive information value Y of the corrected portrait obtained after correcting the first portrait, search for and obtain a pre-stored target tag set that matches the comprehensive information value Y of the corrected portrait;

[0121] Step B3: According to the obtained target tag set, obtain the second portrait corresponding to the target user that includes the target tag set.

[0122] In the embodiment of the present invention, since the comprehensive information value of the corrected portrait is obtained after the correction process, it can effectively ensure the accuracy of the data. At the same time, according to the comprehensive information value of the corrected portrait, the corresponding target tag set is obtained, and then the second portrait is effectively constructed; wherein, the target tag set in the embodiment of the present invention is similar to the landmark tags of the corresponding first portrait.

[0123] In the embodiments of the present invention, based on the comprehensive information value of the first portrait, a correction process is performed to obtain the comprehensive information value of the corrected portrait. Then, according to the comprehensive information value of the corrected portrait, a pre-stored target tag set that matches the comprehensive information value Y of the second portrait is obtained. According to the target tag set, the second portrait corresponding to the target user that includes the target tag set is obtained, in order to improve the fineness of the second portrait, making the generated user portrait more adaptable and accurate to the target user.

[0124] In the embodiments of the user portrait generation method of the present invention, the system obtains user behaviors and calculates relevant eigenvalue based on the feature vector. Secondly, the feature information of the target user is obtained from different dimensions and landmark tags are established, which is convenient for improving the accuracy of the user portrait and also helps to improve the recognition of the target user. In addition, by obtaining the matching between the second portrait and the standard portrait, difference information is obtained, which is convenient for obtaining the parameters to be adjusted, thereby facilitating the optimization of the standard portrait model in the system database and improving the generation efficiency of the user portrait.

[0125] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects.

[0126] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, and the combination of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices produce means for realizing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0127] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including an instruction device that realizes the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0128] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are performed on the computer or other programmable apparatus to produce a computer-implemented process, and thus the instructions executed on the computer or other programmable apparatus provide steps for realizing the functions specified in one process or a plurality of processes and / or blocks Figure 1 one process or a plurality of processes and / or blocks Figure 1 steps for realizing the functions specified in one block or a plurality of blocks.

[0129] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A method for generating a user profile, characterized in that, The user portrait generation method includes: Receiving a user portrait generation instruction, and obtaining a target user corresponding to the user portrait generation instruction; According to the target user, obtaining a standard portrait model corresponding to the target user, and based on a pre-stored portrait adjustment database, calling adjustment parameters to be adjusted that match the target user; Using the called adjustment parameters to be adjusted to adjust the standard portrait model, and generating a user portrait that matches the target user; Based on the standard portrait model, obtaining adjustment parameters to be adjusted corresponding to different target users, and storing the adjustment parameters to be adjusted to form the corresponding portrait adjustment database; Among them, the obtaining of adjustment parameters to be adjusted corresponding to different target users based on the standard portrait model includes: Obtaining behavior information corresponding to the target user, and constructing a feature vector corresponding to the behavior characteristics of the target user; According to the constructed feature vector, calculating a feature value corresponding to the feature vector, where the feature value includes a first feature value and an average feature value; Collecting multiple feature information of the target user in different dimensions, and establishing a landmark label corresponding to the target user; According to the landmark label, using the calculated feature value, obtaining an adjustment parameter to be adjusted corresponding to the target user; Among them, the obtaining of the adjustment parameter to be adjusted corresponding to the target user according to the landmark label and using the calculated feature value includes: According to all the landmark labels, performing data processing on the landmark labels to generate a first portrait corresponding to the target user; According to the calculated first feature value and average feature value, performing correction processing on the first portrait to obtain a second portrait corresponding to the target user; Performing a matching process on the second portrait and the standard portrait model corresponding to the target user to obtain difference information between the second portrait and the standard portrait model; Using the difference information as the adjustment parameter to be adjusted for adjusting the standard portrait model.

2. The user portrait generation method according to claim 1, wherein The obtaining of the behavior information corresponding to the target user and constructing a feature vector corresponding to the behavior characteristics of the target user includes: Obtaining the behavior information corresponding to the target user, and extracting the behavior characteristics in the behavior information to obtain a corresponding behavior characteristic set; Based on the behavior characteristic set, constructing a feature vector corresponding to each behavior characteristic in the behavior characteristic set.

3. The user portrait generation method according to claim 1, wherein, The calculating of the feature value corresponding to the feature vector according to the constructed feature vector includes steps A1 - A5: Step A1. According to the n feature vectors corresponding to the behavior characteristics constructed , using formula (1), calculate the average feature vector of the n feature vectors , then there is: （1） In formula (1), represents the feature vector of the i-th behavioral feature, and n represents the total number of feature vectors corresponding to the behavioral features; represents the feature vectors of n behavioral features average feature vector; Step A2: Using formula (2), obtain the first eigenvalue corresponding to the eigenvector , then we have: （2） In Formula (2), represents the first eigenvalue; Step A3: Using formula (3), respectively calculating a second feature value of each feature vector, then there is: （3） In formula (3), represents the second eigenvalue of the eigenvector of the i-th behavioral feature ; E is the identity matrix, representing the eigenvector of the i-th behavioral feature; Step A4. Using formula (4), calculate the average eigenvalue corresponding to the second eigenvalue , then we have: （4） In formula (4), n represents the total number of feature vectors corresponding to the behavior characteristics; Step A5: Using the calculated first feature value and average feature value as the feature value corresponding to the feature vector.

4. The user portrait generation method according to claim 1, wherein The collecting of multiple feature information of the target user in different dimensions and establishing a landmark label corresponding to the target user includes: Performing information collection on the target user in different dimensions, and obtaining multiple feature information corresponding to the target user in different dimensions; The acquired multiple feature information are subjected to landmark information extraction, and a landmark label is established for each landmark information extracted to obtain a landmark label corresponding to the target user.

5. The user portrait generation method according to claim 3, characterized in that, The step of performing data processing on the iconic tags according to all the iconic tags to generate a first portrait corresponding to the target user includes: According to all the landmark tags, data processing is performed on the landmark tags, and the comprehensive information value S of the first portrait is calculated using formula (5), and then: （5） In formula (5), K represents the total number of the said landmark tags; represents the attribute value corresponding to the k1-th landmark tag, which is a preset value; represents the correction value corresponding to the k1-th landmark tag; represents the weight value corresponding to the k1-th landmark tag, which is a preset value.

6. The user portrait generation method according to claim 5, wherein, The first portrait is corrected according to the calculated first eigenvalue and average eigenvalue to obtain a second portrait corresponding to the target user, including steps B1-B3: Step B1, performing correction processing on the first portrait to obtain a corrected portrait, and using formula (6) to calculate the comprehensive information value Y of the corrected portrait; （6） In Formula (6), S represents the comprehensive information value of the first image; Y represents the comprehensive information value of the corrected image after the correction process is performed on the first image; represents a preset initial image feature value, whose value range is [0, 1]; sim() represents a similarity function; |sim()| represents the eigenvalue of the similarity vector corresponding to the similarity function; represents the first eigenvalue; represents the average eigenvalue; The n represents the total number of feature vectors corresponding to the behavior features; Step B2: according to the comprehensive information value Y of the corrected portrait, searching and acquiring a pre-stored target tag set matching the comprehensive information value Y of the corrected portrait; Step B3: Obtain the second portrait corresponding to the target user based on the acquired target tag set.

7. The user portrait generation method according to any one of claims 1 to 6, characterized in that The method further comprises: According to the generated user portrait, the standard portrait model matching the target user is updated to the newly generated user portrait; Wherein, updating the standard portrait model matching the target user to the newly generated user portrait includes: According to the generated user portrait, the standard portrait model matching the target user is replaced with the user portrait, and the user portrait is used as a new standard portrait model matching the target user, while retaining the corresponding update timestamp information; Based on the new standard portrait model and the update timestamp information, the matching relationship between the parameters to be adjusted that match the target user and the target user is released.

Images