Advertisement crowd expansion method and device based on CatBoost model, equipment and medium

Abstract

The invention discloses an advertisement crowd expansion method and device based on a CatBoost model, equipment and a storage medium, and the method comprises the steps: obtaining user feature data of a target user group, and constructing a user feature data set according to the user feature data; after the user feature data set is preprocessed, a CatBoost model is obtained through training; dividing the user feature data set into a seed user feature data set and a non-seed user feature data set, and converting the seed user feature data set and the non-seed user feature data set into a seed user feature vector set and a non-seed user feature vector set through a CatBoost model; and on the basis of a preset similarity rule, obtaining corresponding non-seed users from the non-seed user feature vector set as an expansion group of the advertisement crowd. Compared with the prior art, the method has the following effects: the prediction model is constructed by using the CatBoost algorithm, and the potential target user group is obtained based on the similarity of the multi-dimensional features of the users, so that the delivery target group of the advertiser is more accurate, and the advertisement delivery effect is better.

Description

Technical Field

[0001] The present invention relates to the technical field of recommendation algorithms, and in particular to a method, device, equipment and medium for expanding advertising audiences based on a CatBoost model. Background Art

[0002] With the continuous development of information technology, the Internet has become an inseparable part of people's daily lives. In today's digital age, the core goal of the advertising recommendation system is to achieve precision marketing, that is, to show the right advertisement to the right user at the right time. In order to achieve this goal, advertisers not only pursue the exposure of advertisements, but also pay more attention to the deep conversion effect of advertisements, that is, the actual purchase behavior or other target behavior of users after clicking on the advertisement.

[0003] Currently, in order to achieve this goal, traditional advertising recommendation systems usually adopt white-box label-oriented methods, such as rule-based recommendations, collaborative filtering, and simple statistical models. Although these methods are effective in some cases, they often fail to meet expectations in terms of scalability and effectiveness, especially in scenarios that require large-scale personalized recommendations. Summary of the invention

[0004] The purpose of the present invention is to provide an advertising audience expansion method, device, equipment and storage medium based on the CatBoost model to solve the problem of low accuracy of advertising placement by advertisers in related technologies.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] In the first aspect, the present invention provides an advertising population expansion method based on a CatBoost model, comprising: obtaining user feature data of a target user group, and constructing a user feature data set based on the user feature data; preprocessing the user feature data set, and inputting the pre-set CatBoost model for training to obtain a trained CatBoost model; dividing the user feature data set into a seed user feature data set and a non-seed user feature data set; converting the seed user feature data set and the non-seed user feature data set into a seed user feature vector set and a non-seed user feature vector set through the trained CatBoost model; based on a preset similarity rule between the seed user feature vector set and the non-seed user feature vector set, obtaining corresponding non-seed users from the non-seed user feature vector set as an expansion group of the advertising population.

[0007] Preferably, the user characteristic data includes: user attribute characteristic data and user behavior characteristic data; the user attribute characteristic data includes: user region information, user language information and user device information; the user behavior characteristic data includes: user recent behavior information and user interest and hobby information.

[0008] Preferably, after converting the seed user feature data set and the non-seed user feature data set into a seed user feature vector set and a non-seed user feature vector set through the trained CatBoost model, the method further includes: storing the non-seed user feature vector set into a Milvus database.

[0009] Preferably, the preprocessing of the user feature data set includes: merging, cleaning, filtering the data in the user feature data set, and structuring, standardizing, and vectorizing the filtered data.

[0010] Preferably, the method further comprises: delivering corresponding advertisements to an expanded group of the advertising population.

[0011] Preferably, the similarity rule adopts the following calculation formula:

[0012]

[0013] Among them, A is the seed user feature vector, B is the non-seed user feature vector, A i is the component of the seed user feature vector, B i is the component of the non-seed user feature vector.

[0014] Preferably, the method of obtaining corresponding non-seed users from non-seed user vectors as an extended group of the advertising population according to a similarity rule between a seed user vector set and a non-seed user vector set comprises: selecting each seed user feature vector from the seed user feature vector data set in turn; cyclically sorting the non-seed user feature vectors in the non-seed user feature vector set according to the similarity with each seed user feature vector, and each time taking the non-seed users corresponding to the first N non-seed user feature vectors and putting them into a similar user set; wherein N is a natural number greater than 0; and performing deduplication processing on the similar user set, and using the deduplication-processed similar user set as an extended group of the advertising population.

[0015] In a second aspect, the present invention also provides an advertising population expansion device based on a CatBoost model, comprising: an acquisition module, used to acquire user feature data of a target user group, and construct a user feature data set based on the user feature data; a training module, used to pre-process the user feature data set, and input it into a preset CatBoost model for training to obtain a trained CatBoost model; a data partitioning module, used to partition the user feature data set into a seed user feature data set and a non-seed user feature data set; a data processing module, used to convert the seed user feature data set and the non-seed user feature data set into a seed user feature vector set and a non-seed user feature vector set through the trained CatBoost model; a data screening module, based on a preset similarity rule between the seed user feature vector set and the non-seed user feature vector set, to acquire corresponding non-seed users from the non-seed user feature vector set as an expansion group of the advertising population.

[0016] Preferably, the device further comprises: an expansion conversion module, for delivering corresponding advertisements to an expanded group of the advertising population.

[0017] A third aspect: The present invention also provides a computer electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of any one of the above-mentioned methods for expanding advertising audiences based on the CatBoost model are implemented.

[0018] Fourth aspect: The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of any one of the above-mentioned advertising audience expansion methods based on the CatBoost model.

[0019] The present invention provides a method, device, equipment and storage medium for expanding advertising population based on a CatBoost model. The method obtains user feature data of a target user group and constructs a user feature data set according to the user feature data; after preprocessing the user feature data set, the data set is input into a preset CatBoost model for training to obtain a trained CatBoost model; the user feature data set is divided into a seed user feature data set and a non-seed user feature data set; the seed user feature data set and the non-seed user feature data set are converted into a seed user feature vector set and a non-seed user feature vector set through the trained CatBoost model; based on the preset similarity rule between the seed user feature vector set and the non-seed user feature vector set, the corresponding non-seed user is obtained from the non-seed user feature vector set as an expansion group of the advertising population. Compared with the related art, the present invention has the following effects: the present invention constructs a prediction model by using the CatBoost algorithm, and obtains a potential target user group based on the similarity of the multi-dimensional features of the user, thereby making the advertiser's delivery target group more accurate and the advertising delivery effect better. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a flow chart of an advertising crowd expansion method based on a CatBoost model according to an embodiment of the present invention;

[0021] Figure 2 It is a structural schematic diagram of an advertising crowd expansion device based on a CatBoost model according to an embodiment of the present invention;

[0022] Figure 3 It is a structural schematic diagram of a computer electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] 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 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.

[0024] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. In contrast, when an element is referred to as being "directly on" another element, there is no intermediate element. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0025] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0026] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0027] The terms used in one or more embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit one or more embodiments of the present invention. The singular forms "a", "said" and "the" used in one or more embodiments of the present invention are also intended to include plural forms, unless the context clearly indicates other meanings.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein in the specification of the template are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the associated listed items.

[0029] It should be understood that, although the terms first, second, etc. may be used to describe various information in one or more embodiments of the present invention, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of the present invention, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "when..." or "when...".

[0030] Please refer to Figure 1 The present invention provides an advertising crowd expansion method based on a CatBoost model, which is applied to the field of advertising delivery and includes at least the following steps:

[0031] S10: Acquire user characteristic data of a target user group, and construct a user characteristic data set based on the user characteristic data.

[0032] Specifically, in this example, first, a massive data source is obtained from the big data platform through big data analysis technology as a potential target user group. Secondly, after obtaining the potential target user group, the user feature data of the target user group is further obtained, and based on the user feature data, a user feature data set is constructed.

[0033] In some optional embodiments, the user characteristic data includes: user attribute characteristic data and user behavior characteristic data; the user attribute characteristic data includes: user region information, user language information and user device information; the user behavior characteristic data includes: user recent behavior information and user interest and hobby information.

[0034] S20, after preprocessing the user feature data set, input it into a preset CatBoost model for training to obtain a trained CatBoost model.

[0035] Specifically, in this embodiment, it is first necessary to perform a preprocessing operation on the user feature data set to improve the data quality and the performance of the model.

[0036] In some optional embodiments, the preprocessing of the user feature data set includes: merging, cleaning, filtering the data in the user feature data set, and structuring, standardizing, and vectorizing the filtered data. For example, data cleaning may include: processing missing values, processing abnormal values, and processing duplicate values. Data standardization may include: minimum-maximum standardization and Z-score standardization.

[0037] In this embodiment, a prediction model is constructed based on the CatBoost algorithm. CatBoost is a gradient boosting decision tree model based on a symmetric binary tree. This structure helps to reduce the bias of the model and improve the generalization ability. The following are the characteristics of CatBoost:

[0038] (1) When CatBoost preprocesses the data set, it first randomly sorts the data set. For each value of the category feature of each sample, when it is converted to a numerical type, the average of the category label values ​​before the sample is taken. At the same time, the priority (prior value) and the weight coefficient of the priority (prior value) are added. Assume that σ=(σ1,

[0039] σ2,...,σn) is a random sorting sequence, then:

[0040]

[0041] Among them, P represents the priority, and α>0 represents the weight coefficient of the priority. Adding weight can reduce the noise caused by low-frequency features in the category features.

[0042] (2) CatBoost automatically processes categorical features without the need for One-Hot encoding. It uses a special algorithm to process categorical features, which helps reduce the number of features and improve model efficiency.

[0043] (3) CatBoost can automatically consider feature combinations, which can capture high-order dependencies between features and thus improve the predictive ability of the model;

[0044] (3) When constructing a new split in the current tree, CatBoost considers the combination of features in a greedy way. The first split does not consider the combination of categorical features, and the subsequent splits consider all feature combinations, and the combined features become numerical.

[0045] (4) CatBoost uses a technique called “target statistics” to deal with the class imbalance problem, which helps the model better learn the characteristics of minority classes.

[0046] It should be noted that, in this embodiment, a prediction model is constructed based on the CatBoost algorithm. CatBoost can process the training data set to generate prediction data. The training process includes:

[0047] (1) Different segmentations are constructed by enumerating and combining different features, and a symmetric binary tree is constructed using these segmentations. The prediction values ​​of the leaf nodes obtained are calculated.

[0048] (2) Calculate the scores of different trees and select the best split based on the scores. The scoring process includes encoding each leaf node into a binary vector with a length equal to the depth of the tree. First, all floating-point features, statistical information, and one-hot encoded features are binarized, and then the binary features are used to calculate the predicted value. The binary feature values ​​of all samples are stored in a continuous vector that can be constructed in parallel at 3 times the speed, which enables fast scoring. The score is determined by the information entropy of the leaf node, and its calculation formula is:

[0049]

[0050] The number of leaf nodes in the decision tree is T, and we number each leaf node t=1, 2, ... T. For a leaf node, the number of samples on it is Nt, and the number of samples of category is Ntk;

[0051] (3) For the current model’s prediction, calculate the residual (i.e., the actual label minus the current model’s prediction);

[0052] (4) Use the residual as the target to build a new decision tree. This tree attempts to fit the residual to reduce the loss. The loss function of CatBoost is:

[0053]

[0054] Among them, Ω(fk) is the regularization term of the kth tree, which usually includes the number of leaf nodes and the sum of the squares of the leaf node weights:

[0055]

[0056] (5) Repeat the above steps until the predetermined number of iterations is reached or other stopping conditions are met.

[0057] It should be noted that in order to reduce overfitting, the CatBoost model uses a method called "orderliness" during training. This method ensures that the training samples for each new tree are randomly arranged, which prevents the model from being too dependent on a specific data order during training.

[0058] S30: Divide the user feature data set into a seed user feature data set and a non-seed user feature data set, wherein the seed users are users who have purchased the advertiser's products, and the non-seed users are users who have not purchased the advertiser's products.

[0059] Specifically, since the user feature data set is data obtained from massive data sources, it is necessary to divide the user feature data set. The purpose of the division is to distinguish between the advertiser's existing users and potential users. Therefore, in this embodiment, the user feature data set is divided into a seed user feature data set and a non-seed user feature data set, wherein the seed user is a user who has purchased the advertiser's product, and the non-seed user is a user who has not purchased the advertiser's product.

[0060] S40, converting the seed user feature data set and the non-seed user feature data set into a seed user feature vector set and a non-seed user feature vector set through the trained CatBoost model.

[0061] Specifically, the seed user feature data set and the non-seed user feature data set are converted into a seed user feature vector set and a non-seed user feature vector set through a CatBoost model.

[0062] It should be noted that after the conversion is completed, the user feature vectors in the non-seed user feature vector set can be stored in the Milvus database. Milvus is a vector database designed for vector query and retrieval. It can index trillions of vector data and is very suitable for storing and searching user vectors.

[0063] S50: Based on a preset similarity rule between the seed user feature vector set and the non-seed user feature vector set, corresponding non-seed users are obtained from the non-seed user feature vector set as an expansion group of the advertising population.

[0064] Specifically, it can be understood that since the corresponding seed users in the seed user feature vector set are users who have purchased the advertiser's products, the corresponding non-seed users can be obtained from the non-seed user feature vector set as an expansion group of the advertising population based on the similarity between the seed user feature vector and the non-seed user feature vector.

[0065] The present invention provides an advertising population expansion method based on a CatBoost model. The method obtains user feature data of a target user group, and constructs a user feature data set according to the user feature data; after preprocessing the user feature data set, the data set is input into a preset CatBoost model for training to obtain a trained CatBoost model; the user feature data set is divided into a seed user feature data set and a non-seed user feature data set; the seed user feature data set and the non-seed user feature data set are converted into a seed user feature vector set and a non-seed user feature vector set through the trained CatBoost model; based on the preset similarity rule between the seed user feature vector set and the non-seed user feature vector set, the corresponding non-seed user is obtained from the non-seed user feature vector set as an expansion group of the advertising population. Compared with the related art, the present invention has the following effects: the present invention constructs a prediction model by using the CatBoost algorithm, and obtains a potential target user group based on the similarity of the multi-dimensional features of the user, thereby making the advertiser's delivery target group more accurate and the advertising delivery effect better.

[0066] In some optional embodiments, the method further includes: delivering corresponding advertisements to an expanded group of the advertising population.

[0067] Specifically, after obtaining the corresponding expansion groups of the advertising population, the corresponding advertisements of the advertisers can be delivered to these expansion groups so that the expansion groups can complete the conversion, thereby enabling the advertisers to generate corresponding profits.

[0068] In some optional embodiments, the similarity rule is calculated using the following formula:

[0069]

[0070] Among them, A is the seed user feature vector, B is the non-seed user feature vector, A i is the component of the seed user feature vector, B i is the component of the non-seed user feature vector.

[0071] Specifically, in this embodiment, for the similarity between the seed user feature vector and the non-seed user feature vector, cosine similarity is used, which measures the cosine value of the angle between the two vectors, wherein the number of dimensions of the two vectors should be consistent, and the final cosine similarity calculation result is between -1 and 1. When the cosine similarity is equal to 1, it indicates that the directions of the two vectors in the multidimensional space are exactly the same, that is, they are completely similar; when the cosine similarity is equal to 0, it indicates that there is no linear relationship between the two vectors, and they are unrelated; when the cosine similarity is equal to -1, it indicates that the directions of the two vectors in the multidimensional space are exactly opposite, that is, they are completely different.

[0072] In some optional embodiments, step S50, obtaining corresponding non-seed users from non-seed user vectors as an expanded group of the advertising population according to a similarity rule between the seed user vector set and the non-seed user vector set, includes:

[0073] S51 . Select each seed user feature vector from the seed user feature vector data set in turn.

[0074] S52, cyclically sorting the non-seed user feature vectors in the non-seed user feature vector set according to the similarity with each seed user feature vector, and each time taking the non-seed users corresponding to the first N non-seed user feature vectors and putting them into the similar user set; wherein N is a natural number greater than 0.

[0075] S53: performing deduplication processing on the similar user set, and using the deduplication-processed similar user set as an expansion group of the advertising population.

[0076] Specifically, first, each seed user feature vector is selected from the seed user feature vector dataset in turn;

[0077] Secondly, after each seed user characteristic vector is selected, the non-seed user feature vectors in the non-seed user feature vector data set are sorted according to the similarity with the seed user feature vector, and the non-seed users corresponding to the first N non-seed user vectors are selected and put into the acquaintance user set, where N is a natural number greater than 0, and the value of N can be set according to actual conditions. For example, the value of N can be 3, 5, 7, etc.

[0078] Finally, after all seed user feature vectors are sorted by similarity and non-seed users are selected and put into the similar user set, the acquaintance user set is deduplicated, and the deduplicated similar user set is used as the advertising audience expansion group corresponding to the advertiser.

[0079] See also Figure 2The embodiment of the present invention further provides an advertising crowd expansion device 200 based on a CatBoost model, comprising:

[0080] The acquisition module 201 is used to acquire user characteristic data of a target user group and construct a user characteristic data set according to the user characteristic data;

[0081] The training module 202 is used to pre-process the user feature data set and input it into a preset CatBoost model for training to obtain a trained CatBoost model;

[0082] A data division module 203 is used to divide the user feature data set into a seed user feature data set and a non-seed user feature data set;

[0083] A data processing module 204 is used to convert the seed user feature data set and the non-seed user feature data set into a seed user feature vector set and a non-seed user feature vector set through the trained CatBoost model;

[0084] The data screening module 205 obtains corresponding non-seed users from the non-seed user feature vector set as an expansion group of the advertising population based on a preset similarity rule between the seed user feature vector set and the non-seed user feature vector set.

[0085] In some optional embodiments, the advertising population expansion device 200 further includes: an expansion conversion module 206, which is used to deliver corresponding advertisements to the expansion group of the advertising population.

[0086] See also Figure 3 The embodiment of the present invention further provides a computer electronic device 400, including a memory 403 and a processor 402, wherein the memory 403 stores a computer program, and when the processor executes the computer program, the steps of the advertising crowd expansion method based on the CatBoost model described in any one of the above are implemented.

[0087] Specifically, the electronic device 400 includes: a transceiver 401, a bus interface and a processor 402, wherein the processor 402 is used to obtain user feature data of a target user group, and construct a user feature data set based on the user feature data; after preprocessing the user feature data set, input it into a preset CatBoost model for training to obtain a trained CatBoost model; divide the user feature data set into a seed user feature data set and a non-seed user feature data set, wherein the seed user is a user who has purchased the advertiser's product, and the non-seed user is a user who has not purchased the advertiser's product; convert the seed user feature data set and the non-seed user feature data set into a seed user feature vector set and a non-seed user feature vector set through the trained CatBoost model; based on a preset similarity rule between the seed user feature vector set and the non-seed user feature vector set, obtain corresponding non-seed users from the non-seed user feature vector set as an expansion group of the advertising population.

[0088] In the embodiment of the present invention, the electronic device 400 further includes a memory 403. Figure 3 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically linking various circuits of one or more processors represented by processor 402 and memory represented by memory 403. The bus architecture may also link various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface. The transceiver 401 may be a plurality of components, i.e., including a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium. The processor 402 is responsible for managing the bus architecture and general processing, and the memory 403 may store data used by the processor 402 when performing operations.

[0089] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of any one of the above-mentioned methods for expanding the advertising audience based on the CatBoost model are implemented.

[0090] In this embodiment, the computer readable storage medium may be a non-volatile storage medium or a volatile storage medium. For example, the computer storage medium may include but is not limited to: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0091] In all examples shown and described herein, any specific values ​​should be interpreted as merely exemplary and not as limiting, and thus other examples of the exemplary embodiments may have different values.

[0092] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0093] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely schematic. For example, the flowcharts and structure diagrams in the accompanying drawings show the possible architecture, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in an alternative implementation, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the structure diagram and / or the flow diagram, and the combination of boxes in the structure diagram and / or the flow diagram, can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0094] In addition, the functional modules or units in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0095] If the functions are implemented in the form of software function modules 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, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a terminal device (which can be a smart phone, 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.

[0096] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the 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.

Claims

1. An advertising crowd expansion method based on CatBoost model, characterized in that: include: Acquire user characteristic data of the target user group, and construct a user characteristic data set based on the user characteristic data; After preprocessing the user feature data set, the data set is input into a preset CatBoost model for training to obtain a trained CatBoost model; Dividing the user feature data set into a seed user feature data set and a non-seed user feature data set, wherein the seed user is a user who has purchased an advertiser's product, and the non-seed user is a user who has not purchased an advertiser's product; Converting the seed user feature data set and the non-seed user feature data set into a seed user feature vector set and a non-seed user feature vector set through the trained CatBoost model; Based on a preset similarity rule between the seed user feature vector set and the non-seed user feature vector set, corresponding non-seed users are obtained from the non-seed user feature vector set as an expansion group of the advertising population.

2. The advertising crowd expansion method according to claim 1, characterized in that: The user characteristic data includes: user attribute characteristic data and user behavior characteristic data; The user attribute characteristic data includes: user location information, user language information and user device information; The user behavior feature data includes: user's recent behavior information and user's interest and hobby information.

3. The method for expanding the advertising audience according to claim 1, characterized in that: The preprocessing of the user feature data set includes: merging, cleaning, and filtering the data in the user feature data set, and structuring, standardizing, and vectorizing the filtered data.

4. The advertising crowd expansion method according to claim 1, characterized in that: The method further includes: delivering corresponding advertisements to an expanded group of the advertising population.

5. The method for expanding the advertising audience according to claim 1, characterized in that: The similarity rule adopts the following calculation formula: Among them, A is the seed user feature vector, B is the non-seed user feature vector, A i is the component of the seed user feature vector, B i is the component of the non-seed user feature vector.

6. The method for expanding the advertising audience according to claim 1, characterized in that: The step of obtaining corresponding non-seed users from non-seed user vectors as an expansion group of the advertising population according to the similarity rule between the seed user vector set and the non-seed user vector set includes: Selecting each seed user feature vector from the seed user feature vector data set in turn; The non-seed user feature vectors in the non-seed user feature vector set are cyclically sorted according to the similarity with each seed user feature vector, and the non-seed users corresponding to the first N non-seed user feature vectors are taken each time and put into the similar user set; wherein N is a natural number greater than 0; The similar user set is deduplicated, and the deduplicated similar user set is used as an expanded group of advertising audience.

7. An advertising crowd expansion device based on the CatBoost model, characterized in that: include: An acquisition module, used to acquire user characteristic data of a target user group and construct a user characteristic data set based on the user characteristic data; A training module, used for preprocessing the user feature data set and inputting it into a preset CatBoost model for training to obtain a trained CatBoost model; A data division module, used for dividing the user feature data set into a seed user feature data set and a non-seed user feature data set; A data processing module, used for converting the seed user feature data set and the non-seed user feature data set into a seed user feature vector set and a non-seed user feature vector set through the trained CatBoost model; The data screening module obtains corresponding non-seed users from the non-seed user feature vector set as an expansion group of the advertising population based on a preset similarity rule between the seed user feature vector set and the non-seed user feature vector set.

8. The advertising crowd expansion device according to claim 7, further comprising: The expansion conversion module is used to deliver corresponding advertisements to the expansion groups of the advertising population.

9. A computer electronic device, characterized in that: The invention comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the advertising audience expansion method based on the CatBoost model described in any one of claims 1 to 6 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the advertising audience expansion method based on the CatBoost model described in any one of claims 1 to 6 are implemented.

Images