Crowd selection system and method based on semi-supervised random forest algorithm

Abstract

The invention relates to the technical field of crowd selection, in particular to a crowd selection system and method based on a semi-supervised random forest algorithm, which can provide optimized marketing strategies for enterprises, improve user experience and optimize business decisions. The method comprises the following steps: collecting historical crowd behavior data, wherein the historical crowd behavior data comprises behavior records, transaction history and social interaction; performing feature selection on the historical crowd behavior data, selecting features with high distinction degree, and obtaining a crowd behavior difference feature set; performing model training by using the crowd behavior difference feature set to obtain a crowd circling model; acquiring user activity data and activity scene environment data in real time, and preprocessing the user activity data and the activity scene environment data; inputting the user activity data and the activity scene environment data into a crowd selection model, and performing user grouping to obtain a crowd activity difference set; and performing crowd identification according to the crowd activity difference set, and selecting a target user group.

Description

Technical Field

[0001] This invention relates to the technical field of crowd selection, and in particular to a crowd selection system and method based on a semi-supervised random forest algorithm. Background Technology

[0002] In today's digital age, businesses and organizations increasingly rely on big data analytics to improve marketing efficiency, enhance user experience, and optimize business decisions. With the widespread adoption of the internet and the rise of social media, the amount of user-generated data is growing exponentially, including online behavior patterns, transaction records, and social interaction information. This data contains user preferences, habits, and potential needs, becoming a crucial foundation for refined operations and personalized services. However, effectively extracting valuable information from massive and complex data to accurately target specific user groups remains a significant challenge for businesses and researchers.

[0003] Traditional user segmentation methods are mostly based on supervised learning, requiring a large amount of labeled data as a training foundation. This is often difficult to meet in practical applications, especially when dealing with new users or new scenarios. Furthermore, the diversity and dynamism of user behavior necessitate models with high flexibility and adaptability. Therefore, exploring more efficient and adaptable user segmentation techniques has become an urgent need for industry development. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a semi-supervised random forest-based audience selection method that can optimize marketing strategies, improve user experience, and improve business decision-making for enterprises.

[0005] In a first aspect, the present invention provides a crowd selection method based on a semi-supervised random forest algorithm, the method comprising:

[0006] Collect historical population behavior data, which includes behavior records, transaction history, and social interactions;

[0007] Feature selection is performed on historical population behavior data to select high-discrimination features and obtain a set of population behavior difference features.

[0008] A crowd selection model is obtained by training a set of behavioral difference features of the population.

[0009] Real-time acquisition of user activity data and activity scenario environment data, and preprocessing of user activity data and activity scenario environment data;

[0010] User activity data and activity scenario environment data are input into the audience segmentation model to segment users and obtain a set of differences in audience activities;

[0011] Crowd identification is performed based on the set of differences in crowd activities to select the target user group.

[0012] Furthermore, channels for obtaining historical user behavior data include website logs, app tracking, transaction system records, social media interactions, and customer relationship management systems.

[0013] Furthermore, methods for feature selection from historical population behavior data include:

[0014] Preprocessing of historical population behavior data includes data cleaning, data normalization, and feature encoding;

[0015] Examine the processed historical population behavior data, including behavior records, transaction history, and social interaction information, to understand the data structure, scope, quality, and existing problems;

[0016] Through statistical analysis and visualization, patterns, outliers, and correlations in the data can be identified.

[0017] The random forest algorithm is used to train the preprocessed data to obtain the importance score of each feature;

[0018] Sort the features by importance and select the top-ranked set; the number of features selected is determined based on performance evaluation to determine the optimal feature subset.

[0019] The final selected features constitute the feature set of population behavior differences.

[0020] Furthermore, the method for constructing the crowd selection model includes:

[0021] The feature set of behavioral differences among the population is divided into a training set and a validation set. The training set is used to build the model, while the validation set is used to tune the model's hyperparameters and evaluate the model's performance.

[0022] For each tree in the training set, bootstrap sampling is used to randomly select data from all samples for training;

[0023] During the tree construction process, the best segmentation feature is selected from a randomly selected subset of features at each node to construct the decision tree;

[0024] A predetermined number of decision trees are generated, and the decision trees are merged to form a random forest, thus obtaining a population selection model.

[0025] Furthermore, methods for preprocessing user activity data and activity scenario environment data include data cleaning, standardization and normalization, feature encoding, time series processing, noise reduction, and feature selection.

[0026] Furthermore, methods for obtaining the set of differences in population activity include:

[0027] The preprocessed real-time data and the updated feature set are used as inputs and fed into the crowd selection model;

[0028] Based on the characteristics of the input data, the model uses the ensemble learning mechanism of random forest to group users through voting by multiple decision trees;

[0029] Based on the model output, user segmentation results are generated, forming a set of differences in population activities.

[0030] Furthermore, methods for acquiring the target user group include:

[0031] The set of differences in crowd activity indicates that users are assigned to different groups, and each group represents a type of users with similar behavioral characteristics;

[0032] The characteristics of each user group are summarized and analyzed to understand the typical behavioral patterns, preferences and transaction habits of each group, and to identify the unique attributes and potential value of each group.

[0033] Build ideal user profiles for the marketing goals and business scenarios that enterprises are concerned with;

[0034] Prioritize each group to obtain a set of user group sequences;

[0035] Select the user group that ranks highest in the user group sequence set as the target user group.

[0036] On the other hand, this application also provides a crowd selection system based on a semi-supervised random forest algorithm, the system comprising:

[0037] The data collection module is used to collect historical population behavior data, which includes behavior records, transaction history, and social interactions.

[0038] The feature selection module is used to select features from crowd behavior data, select features with high discriminative power, and obtain a set of crowd behavior difference features.

[0039] The model training module uses a set of behavioral difference features of the population to train the model and obtain a population selection model.

[0040] The data preprocessing module is used to acquire user activity data and activity scene environment data in real time, and to preprocess the user activity data and activity scene environment data.

[0041] The user segmentation module is used to input user activity data and activity scenario environment data into the crowd selection model to segment users and obtain a set of differences in crowd activities.

[0042] The crowd identification module is used to identify target user groups based on the set of differences in crowd activities.

[0043] Thirdly, this application provides an electronic device including a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor. The transceiver, the memory, and the processor are connected via the bus, and the computer program, when executed by the processor, implements the steps of any of the methods described above.

[0044] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described above.

[0045] Compared with the prior art, the beneficial effects of the present invention are: the method can use unlabeled data for training, which greatly alleviates the need for a large amount of labeled data and reduces the threshold and cost of data preparation;

[0046] Due to the diversity and dynamic changes in user behavior, models need to be highly flexible. Semi-supervised random forests, by integrating multiple decision trees, not only improve the stability and accuracy of the model, but also adapt well to the complexity and variability of user behavior, making the model more adaptable when facing new situations.

[0047] Through a careful feature selection process, only features with high discriminative power for user grouping are retained, reducing noise interference and improving the model's learning efficiency and grouping accuracy; this helps to extract the most core and valuable information from massive amounts of data, making user profiles more accurate.

[0048] This method can acquire and process user activity data and activity scenario environment data in real time, ensuring the timeliness of user segmentation; the preprocessing step ensures data quality, providing a reliable foundation for subsequent model prediction; through the analysis of the set of differences in user activity, it can more accurately identify target user groups with specific behavioral patterns and needs, creating conditions for enterprises to provide refined operations and personalized services; the user segmentation model generated by this method not only serves user segmentation, but also provides deep insights into user needs, driving the intelligent upgrade of business decision-making.

[0049] In summary, the audience selection method based on the semi-supervised random forest algorithm provides enterprises with a powerful tool for optimizing marketing, improving user experience, and optimizing business decisions in the digital age, thanks to its high efficiency, flexibility, and accuracy. Attached Figure Description

[0050] Figure 1 This is a flowchart of the present invention;

[0051] Figure 2 This is a structural diagram of a crowd selection system based on a semi-supervised random forest algorithm. Detailed Implementation

[0052] As will be apparent to those skilled in the art from the description of this application, this application can be implemented as a method, apparatus, electronic device, and computer-readable storage medium. Therefore, this application can be specifically implemented in the following forms: entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software. Furthermore, in some embodiments, this application can also be implemented as a computer program product contained in one or more computer-readable storage media, which includes computer program code.

[0053] The aforementioned computer-readable storage medium may be any combination of one or more computer-readable storage media. Computer-readable storage media include: electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of computer-readable storage media include: portable computer disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, flash memory, optical fiber, optical disc read-only memory, optical storage devices, magnetic storage devices, or any combination thereof. In this application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0054] The acquisition, storage, use, and processing of data in this application all comply with relevant national laws and regulations.

[0055] This application describes the provided methods, apparatus, and electronic devices using flowcharts and / or block diagrams.

[0056] It should be understood that each block of a flowchart and / or block diagram, as well as combinations of blocks in a flowchart and / or block diagram, can be implemented by computer-readable program instructions. These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine that, when executed by a computer or other programmable data processing apparatus, creates means for implementing the functions / operations specified in the blocks of the flowchart and / or block diagram.

[0057] These computer-readable program instructions may also be stored in a computer-readable storage medium that enables a computer or other programmable data processing device to function in a particular manner. In this way, the instructions stored in the computer-readable storage medium produce an instruction apparatus product that includes the functions / operations specified in the blocks of a flowchart and / or block diagram.

[0058] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus or other device to produce a computer-implemented process, such that the instructions that execute on the computer or other programmable data processing apparatus provide a process for implementing the functions / operations specified in the blocks of the flowchart and / or block diagram.

[0059] This application will now be described with reference to the accompanying drawings.

[0060] Example 1: As Figure 1 As shown, the present invention provides a crowd selection method based on a semi-supervised random forest algorithm, which specifically includes the following steps:

[0061] S1. Collect historical population behavior data, which includes behavior records, transaction history, and social interactions;

[0062] Behavior Records: These cover all user interactions across digital touchpoints such as websites, apps, and smart devices, including page browsing, clicks, search queries, and content viewing. This data is collected using event tracking technology and requires close collaboration with the product and technology teams to ensure the comprehensiveness and accuracy of the tracking code deployment.

[0063] Transaction history: Collects users' purchase records, order details, return and exchange information, shopping cart behavior, etc. This data comes from the enterprise's ERP system, CRM system or third-party payment platform interface;

[0064] Social interaction information: including posts, comments, shares, likes, follows, and fan relationships on social media; this type of data is obtained through social media API interfaces and must comply with the data usage policies of each platform and consider privacy protection measures;

[0065] The channels for obtaining historical population behavior data include:

[0066] Website logs and app tracking: Record user browsing behavior, clickstream, page dwell time, etc., reflecting users' online interests and preferences;

[0067] Transaction system records include users' purchase history, purchase frequency, average order value, etc., revealing users' spending power and purchasing habits;

[0068] Social media interaction: Capture user interaction information such as posts, comments, likes, and shares on social media to analyze their social preferences and influence;

[0069] Customer Relationship Management System: Integrates basic customer information, service records, feedback, etc., to provide comprehensive background information for user profiles;

[0070] Step S1 effectively collects and organizes historical population behavior data, laying a solid foundation for subsequent feature selection, model training, and real-time data processing. By analyzing and utilizing this data, companies can better understand and serve their user groups, thereby improving marketing efficiency, enhancing user experience, and optimizing business decisions.

[0071] S2. Select features from historical population behavior data, choose features with high discriminative power, and obtain a set of population behavior difference features.

[0072] Preprocessing historical population behavior data, including data cleaning, data normalization, and feature coding, ensures data quality and provides a good foundation for subsequent feature selection.

[0073] A comprehensive review of the processed historical population behavior data is conducted, including behavior records, transaction history, and social interaction information, to understand the basic structure, scope, quality, and potential problems of the data.

[0074] By using statistical analysis and visualization techniques, patterns, outliers, and correlations in the data can be identified, providing an intuitive basis for feature selection.

[0075] Feature selection typically tends to use embedded methods, particularly leveraging feature importance scores from random forests. Random forests assess feature importance by calculating the reduction in the average impurity of a feature across all decision trees. Features with high feature importance scores are considered key features for distinguishing user groups.

[0076] The random forest algorithm is used to train the preprocessed data to obtain the importance score of each feature;

[0077] The features are sorted according to their importance, and the top-ranked set of features is selected. The number of features selected is determined based on performance evaluation to determine the optimal feature subset, balancing the complexity of the model with its predictive power.

[0078] The final selected features form a set of behavioral difference features of the population, which will be the basis for building a population selection model;

[0079] Through the above steps, a set of highly discriminative features, namely the set of behavioral difference features, can be obtained. These features can not only effectively capture key information about user behavior, but also have sufficient flexibility to adapt to the diversity and dynamism of user behavior, laying a solid foundation for building an accurate audience selection model.

[0080] S3. Use the feature set of differences in crowd behavior to train the model and obtain the crowd selection model;

[0081] Semi-supervised learning is a machine learning method that lies between supervised and unsupervised learning. It aims to learn using a small amount of labeled data and a large amount of unlabeled data. This method is suitable for dealing with user segmentation problems because in real-world applications, obtaining large-scale labeled data is costly and difficult, while unlabeled data is usually easy to obtain and plentiful.

[0082] The behavioral difference feature set of the population is divided into a training set and a validation set. The training set contains partially labeled data and a large amount of unlabeled data. The training set is used to build the model, while the validation set is used to adjust the hyperparameters of the model and evaluate the performance of the model.

[0083] For each tree in the training set, bootstrap sampling is used to randomly select data from all samples for training;

[0084] During the tree construction process, the best segmentation feature is selected from a randomly selected subset of features at each node to construct the decision tree;

[0085] Repeat the above process to generate a predetermined number of decision trees, forming a random forest;

[0086] Use cross-validation to evaluate the model's generalization ability and ensure that the model performs well on unseen data;

[0087] Adjust the hyperparameters of the random forest model to improve its performance and robustness;

[0088] After training is completed, a crowd selection model is obtained;

[0089] By employing effective feature selection and the application of a semi-supervised random forest algorithm, step S3 ensures that the model maintains good efficiency and accuracy when processing large-scale, high-dimensional data, thereby meeting the user segmentation needs of enterprises in the digital age.

[0090] S4. Acquire user activity data and activity scene environment data in real time, and preprocess the user activity data and activity scene environment data;

[0091] User activity data includes the user's latest online behavior, such as website access records, search queries, click behavior, browsing time, etc. This data reflects the user's current interests and activity level and is a key basis for judging the user's current status.

[0092] Activity scenario environment data: Activity scenario environment data refers to the environmental information when users engage in activities at specific times and locations; including the user's geographical location, the weather conditions at the time, trending topics on social media, etc.; this information can help to better understand the background and motivations behind user behavior;

[0093] Methods for preprocessing user activity data and activity scenario environment data include:

[0094] Data cleaning involves removing invalid, incomplete, or erroneous data records, such as outlier detection and handling, filling in missing values, and removing duplicate records.

[0095] Standardization and normalization are used for numerical features to ensure that features at different scales have equal importance in the model and to prevent certain features from dominating the model learning process due to their large numerical range.

[0096] Feature encoding converts categorical variables into numerical forms. Common methods include one-hot encoding, label encoding, or ordinal encoding, so that the model can understand and process this information.

[0097] Time series processing involves performing time series analysis on time-related data, such as sliding window aggregation and periodic pattern recognition, to extract time series features.

[0098] Noise reduction and feature selection: Apply statistical or model-based methods to further screen features, remove noisy features that are not closely related to user grouping, and retain the most discriminative feature set;

[0099] Through preprocessing steps, the raw data is transformed into a format that the model can directly process, improving data quality and model training efficiency, thereby enabling more accurate user segmentation tasks.

[0100] S5. Input user activity data and activity scenario environment data into the crowd selection model to segment users and obtain a set of differences in crowd activities;

[0101] The preprocessed real-time data and the updated feature set are fed into the previously trained semi-supervised random forest model. Due to the characteristics of semi-supervised learning, the model can not only process labeled training data, but also learn from a large amount of unlabeled data, which makes the model more adaptable when dealing with new users or new scenarios.

[0102] The model uses the ensemble learning mechanism of random forest to group users based on the features of the input data, through voting or probability averaging of multiple decision trees; each decision tree divides users based on different combinations of features, and the final grouping result is the output of the combined results of all decision trees.

[0103] Based on the model output, user segmentation results are generated, forming a set of differences in user activity, where each group represents a user group with similar behavioral patterns and characteristics; the differences between these groups reflect the diversity of user behavior, preferences, or needs.

[0104] The S5 process enables rapid and dynamic identification of user groups through real-time data processing, dynamic feature updates, model application, and real-time feedback mechanisms, meeting the real-time needs of enterprises for user segmentation and personalized services in a complex and ever-changing internet environment.

[0105] S6. Identify the target user group based on the set of differences in crowd activities;

[0106] The set of differences in crowd activity indicates that users are assigned to different groups, and each group represents a type of users with similar behavioral characteristics;

[0107] The characteristics of each user group are summarized and analyzed to understand the typical behavioral patterns, preferences, and transaction habits of each group, and to identify the unique attributes and potential value of each group; this includes analyzing factors such as the activity level, purchasing power, content preferences, and social influence of each group.

[0108] Utilize statistical methods or machine learning models to evaluate the impact weight of each feature on user value or responsiveness, and identify several features that are crucial for target positioning;

[0109] Based on the above analysis, we can build ideal user profiles for specific marketing goals or business scenarios that enterprises are concerned with.

[0110] Prioritize each group based on its match with the target user profile and its potential to achieve business goals;

[0111] Establish a monitoring mechanism to track behavioral changes and market feedback among various target user groups, ensuring the timeliness and accuracy of user profiles;

[0112] Based on feedback, flexibly adjust the definition of the target user group, or develop differentiated marketing strategies and content for different groups to improve conversion rates and user satisfaction;

[0113] The identified target user group is the direct audience of the marketing campaign, and personalized content, channel and timing strategies are used to reach them;

[0114] Step S6 is not only the application stage of user segmentation results, but also a key link in the strategy formulation and execution of the entire audience selection process. By deeply analyzing the characteristics of user groups, it builds and optimizes target user profiles, ultimately guiding enterprises to achieve precise marketing and service optimization, and achieve predetermined business goals.

[0115] Example 2: Figure 2 As shown, the present invention provides a crowd selection system based on a semi-supervised random forest algorithm, which specifically includes the following modules;

[0116] The data collection module is used to collect historical population behavior data, which includes behavior records, transaction history, and social interactions.

[0117] The feature selection module is used to select features from crowd behavior data, select features with high discriminative power, and obtain a set of crowd behavior difference features.

[0118] The model training module uses a set of behavioral difference features of the population to train the model and obtain a population selection model.

[0119] The data preprocessing module is used to acquire user activity data and activity scene environment data in real time, and to preprocess the user activity data and activity scene environment data.

[0120] The user segmentation module is used to input user activity data and activity scenario environment data into the crowd selection model to segment users and obtain a set of differences in crowd activities.

[0121] The crowd identification module is used to identify target user groups based on the set of differences in crowd activities.

[0122] This system employs a semi-supervised learning method, which enables learning from a large amount of unlabeled data on the basis of limited labeled data, reducing the reliance on expensive and hard-to-obtain labeled data;

[0123] The semi-supervised random forest algorithm, by integrating multiple decision trees, not only enhances the stability of the model, but also improves its adaptability to the diversity and dynamic changes of user behavior, enabling the model to respond more flexibly to changes in the market and user behavior.

[0124] The data collection module and feature selection module in the system work together to ensure that the most valuable features are quickly selected from massive amounts of data, improving analysis efficiency and clustering accuracy.

[0125] The data preprocessing module can process user activity data and scene environment data in real time, ensuring that the system can respond quickly to market dynamics and adjust the segmentation strategy in a timely manner.

[0126] The combination of user segmentation and audience identification modules enables the accurate identification of target user groups with specific preferences and needs from complex user behaviors, providing a solid foundation for refined operations and personalized services, thereby improving marketing efficiency and user experience. Through this system, business decisions such as product development and marketing strategies can be optimized based on accurate user segmentation results, further enhancing the company's competitiveness and market responsiveness.

[0127] In summary, this system effectively solves the problem of user segmentation in a big data environment through its innovative semi-supervised learning method and efficient processing flow, providing strong support for data-driven decision-making for enterprises.

[0128] The various variations and specific embodiments of the crowd selection method based on the semi-supervised random forest algorithm in the aforementioned Embodiment 1 are also applicable to the crowd selection system based on the semi-supervised random forest algorithm in this embodiment. Through the foregoing detailed description of the crowd selection method based on the semi-supervised random forest algorithm, those skilled in the art can clearly understand the implementation method of the crowd selection system based on the semi-supervised random forest algorithm in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0129] In addition, this application also provides an electronic device, including a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor. The transceiver, the memory, and the processor are respectively connected via the bus. When the computer program is executed by the processor, it implements the various processes of the above-described method embodiment for controlling output data and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0130] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A population selection method based on a semi-supervised random forest algorithm, characterized in that, The method includes: Collect historical population behavior data, which includes behavior records, transaction history, and social interactions; Feature selection is performed on historical population behavior data to select high-discrimination features and obtain a set of population behavior difference features. A crowd selection model is obtained by training a set of behavioral difference features of the population. Real-time acquisition of user activity data and activity scenario environment data, and preprocessing of user activity data and activity scenario environment data; User activity data and activity scenario environment data are input into the audience segmentation model to segment users and obtain a set of differences in audience activities; Crowd identification is performed based on the set of differences in crowd activities to select the target user group.

2. The crowd selection method based on a semi-supervised random forest algorithm as described in claim 1, characterized in that, Historical user behavior data can be obtained from sources including website logs, app tracking, transaction system records, social media interactions, and customer relationship management systems.

3. The crowd selection method based on a semi-supervised random forest algorithm as described in claim 1, characterized in that, Methods for feature selection from historical population behavior data include: Preprocessing of historical population behavior data includes data cleaning, data normalization, and feature encoding; Examine the processed historical population behavior data, including behavior records, transaction history, and social interaction information, to understand the data structure, scope, quality, and existing problems; Through statistical analysis and visualization, patterns, outliers, and correlations in the data can be identified. The random forest algorithm is used to train the preprocessed data to obtain the importance score of each feature; Sort the features by importance and select the top-ranked set; the number of features selected is determined based on performance evaluation to determine the optimal feature subset. The final selected features constitute the feature set of population behavior differences.

4. The crowd selection method based on a semi-supervised random forest algorithm as described in claim 1, characterized in that, The method for constructing the crowd selection model includes: The feature set of behavioral differences among the population is divided into a training set and a validation set. The training set is used to build the model, while the validation set is used to tune the model's hyperparameters and evaluate the model's performance. For each tree in the training set, bootstrap sampling is used to randomly select data from all samples for training; During the tree construction process, the best segmentation feature is selected from a randomly selected subset of features at each node to construct the decision tree; A predetermined number of decision trees are generated, and the decision trees are merged to form a random forest, thus obtaining a population selection model.

5. The crowd selection method based on a semi-supervised random forest algorithm as described in claim 1, characterized in that, Methods for preprocessing user activity data and activity scenario environment data include data cleaning, standardization and normalization, feature encoding, time series processing, noise reduction, and feature selection.

6. The crowd selection method based on a semi-supervised random forest algorithm as described in claim 1, characterized in that, Methods for obtaining the set of differences in crowd activity include: The preprocessed real-time data and the updated feature set are used as inputs and fed into the crowd selection model; Based on the characteristics of the input data, the model uses the ensemble learning mechanism of random forest to group users through voting by multiple decision trees; Based on the model output, user segmentation results are generated, forming a set of differences in population activities.

7. The crowd selection method based on a semi-supervised random forest algorithm as described in claim 1, characterized in that, Methods for acquiring the target user group include: The set of differences in crowd activity indicates that users are assigned to different groups, and each group represents a type of users with similar behavioral characteristics; The characteristics of each user group are summarized and analyzed to understand the typical behavioral patterns, preferences and transaction habits of each group, and to identify the unique attributes and potential value of each group. Build ideal user profiles for the marketing goals and business scenarios that enterprises are concerned with; Prioritize each group to obtain a set of user group sequences; Select the user group that ranks highest in the user group sequence set as the target user group.

8. A crowd selection system based on a semi-supervised random forest algorithm, characterized in that, The system includes: The data collection module is used to collect historical population behavior data, which includes behavior records, transaction history, and social interactions. The feature selection module is used to select features from crowd behavior data, select features with high discriminative power, and obtain a set of crowd behavior difference features. The model training module uses a set of behavioral difference features of the population to train the model and obtain a population selection model. The data preprocessing module is used to acquire user activity data and activity scene environment data in real time, and to preprocess the user activity data and activity scene environment data. The user segmentation module is used to input user activity data and activity scenario environment data into the crowd selection model to segment users and obtain a set of differences in crowd activities. The crowd identification module is used to identify target user groups based on the set of differences in crowd activities.

9. An electronic device for crowd selection based on a semi-supervised random forest algorithm, comprising a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the transceiver, the memory, and the processor are connected via the bus, characterized in that, When the computer program is executed by the processor, it implements the steps of the method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-7.

Images