A push system and method based on tag recognition

Abstract

The present invention relates to the field of information push technology, specifically a push system based on tag identification and a method thereof, wherein the system includes a user data collection module, a tag customization module, a dynamic tag optimization module, an information matching push module, a group behavior analysis module, and a cross-domain synchronous update module. In the present invention, by capturing the real-time behavior of users and dynamically adjusting personalized tags, the relevance and accuracy of information push are significantly improved. By continuously monitoring user interactions, it is possible to respond to changes in user preferences in real time, automatically adjust push strategies, enhance user experience, and ensure that users always receive the information that best meets their current interests and needs. Through group behavior data analysis, it is possible to capture and predict industry trends, adjust information flows in advance, and optimize information distribution efficiency in a multi-user environment. Cross-domain tag synchronous updates further improve broad adaptability and data consistency, making the information push process smoother and more effective.

Description

Technical Field

[0001] The present invention relates to the technical field of information push, and in particular to a push system and method based on tag recognition. Background Art

[0002] The field of information push technology focuses on how to efficiently and in real time transmit data and information to users through the network, integrating big data analysis, machine learning, user behavior analysis and network communication technology to optimize the information screening, classification and distribution process. Information push technology is not only used in commercial advertising and social notifications, but is also increasingly used in key areas such as education, health monitoring, and emergency response systems. Through accurate analysis of user preferences and real-time data processing, it can provide users with more personalized and targeted information services.

[0003] Traditional information push technology relies on preset user analysis and static preference models, which results in the system being unable to update in real time when user needs change rapidly, and the pushed content often cannot accurately match the user's current actual needs. This lag is manifested in the user experience as irrelevance of information, resulting in a decrease in user engagement. Especially in highly dynamic environments such as the job market, static models cannot capture subtle changes in user preferences, which may cause users to miss critical recruitment opportunities. In addition, due to the lack of effective group behavior analysis and cross-domain data integration, existing systems often show consistency and timeliness issues when processing user data from different platforms, affecting the overall service quality and user satisfaction. These technical limitations are particularly prominent in application scenarios that require fast and accurate information push, such as emergency response and health monitoring, which may result in key information not being delivered to key users in a timely manner. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings in the prior art and to propose a push system and method based on tag recognition.

[0005] In order to achieve the above object, the present invention adopts the following technical solution: A push system based on tag recognition includes:

[0006] The user data collection module collects the personal information of users registered through their mobile phone numbers, simultaneously collects the user's behavior data on the platform, organizes the data and processes it in a standardized format to generate a user basic behavior data set;

[0007] The label customization module analyzes the basic behavior data set described by the user, captures the user's behavior pattern and frequent interaction content, determines the user's label customization options, and records the corresponding information to obtain the user's label preference set;

[0008] The dynamic tag optimization module performs data analysis based on the user tag preference set, identifies the interaction between tags and users, adjusts tag weights according to the interaction data, reorders tag priorities, and obtains optimized tag weight information;

[0009] The information matching and pushing module uses the optimized tag weight information to filter the recruitment and examination information associated with the user tag in the database, combines the user activity time record, pushes information through the instant messaging tool, and generates targeted information push records;

[0010] The group behavior analysis module combines the targeted information push records, collects the behavior data of different user groups, performs data comparison and analysis, identifies interest trends and summarizes group characteristics, and obtains group behavior analysis results;

[0011] The cross-domain synchronous update module conducts an overall review of the tags based on the group behavior analysis results, synchronously updates the tag library based on user feedback and behavior patterns on different platforms, determines the consistency and real-time nature of user tags on all platforms, and generates tag synchronization update records.

[0012] As a further solution of the present invention, the user basic behavior data set includes user personal information, behavior records and standardized data, the user tag preference set includes behavior pattern records, interactive content and user tag selection results, the optimized tag weight information includes weight adjustment records, priority sorting results and tag interaction analysis records, the targeted information push records include screening results, push time records and communication tool usage records, the group behavior analysis results include interest trend analysis results, group characteristics and group behavior data analysis records, and the tag synchronization update records include tag update records, platform synchronization results and real-time verification records.

[0013] As a further solution of the present invention, the user data acquisition module includes an information acquisition submodule, a data sorting submodule, and a data set generation submodule;

[0014] The information collection submodule collects user personal information, simultaneously records the user's click, reading and sharing behavior data on the platform, captures behavior data points to build behavior trajectories, and obtains user behavior information;

[0015] The data sorting submodule screens key data fields according to the user behavior information, performs format unification and error correction, performs deduplication and normalization processing on the data, and outputs standardized user data;

[0016] The data set generation submodule uses the standardized user data to sort the behavior data according to the time series, merges the personal information with the sorted behavior data, and generates a user basic behavior data set.

[0017] As a further solution of the present invention, the label customization module includes a behavior pattern analysis submodule, an interactive content identification submodule, and a label generation submodule;

[0018] The behavior pattern analysis submodule analyzes the indicators in the user basic behavior data set, including login frequency, active time period, and function usage preference, identifies the regular patterns and preference trends of user behavior, extracts the statistical characteristics of activities, and outputs a behavior pattern feature set;

[0019] The interactive content identification submodule analyzes and records the frequency of interaction between the user and the target content according to the behavioral pattern feature set, and deeply mines the content types and topics preferred by the user according to the interaction frequency to obtain an interactive preference set;

[0020] The label generation submodule uses the interaction preference set in combination with the behavior pattern feature set to construct a personalized label for each user, record the relevant information and attributes of the label, and perform data integration to generate a user label preference set.

[0021] As a further solution of the present invention, the dynamic tag optimization module includes a behavior data analysis submodule, a tag weight adjustment submodule, and a priority sorting submodule;

[0022] The behavior data analysis submodule identifies key data points of the interaction between tags and users based on the user tag preference set, performs cluster analysis on the data, reveals the user's preference pattern for tags, and generates tag interaction analysis records;

[0023] The tag weight adjustment submodule calculates the weight score of each tag based on the tag interaction analysis record, dynamically adjusts the weight of each tag in combination with the user interaction data, verifies the consistency of the weight with the user's actual preference, and outputs the adjusted tag weight;

[0024] The priority sorting submodule uses the adjusted tag weights to re-sort the tags, optimizes the display priority of user-preferred tags, and outputs optimized tag weight information.

[0025] As a further solution of the present invention, the weight score of each tag is according to the formula:

[0026] ;

[0027] Calculate, where represents the calculated label weight score, represents the intensity of the user's reaction to label j, represents the frequency of occurrence of tag j in the user's historical interactions, C represents the average interaction frequency of all tags, which is used to evaluate the performance of each tag relative to the average level, j represents the tag, and n represents the number of tags.

[0028] As a further solution of the present invention, the information matching and pushing module includes an information screening submodule, a user activity time matching submodule, and a pushing record generating submodule;

[0029] The information screening submodule screens the recruitment and examination information directly associated with the user tag through database query based on the optimized tag weight information, confirms that the content matches the user's major and interests, and generates a screening content record;

[0030] The user activity time matching submodule uses the screening content record and combines it with the user's online activity log to perform time series analysis to determine the user's most frequently active time period, and obtains the user activity time analysis record;

[0031] The push record generation submodule analyzes the user activity time records, selects the most active time period of the user to push the screened recruitment and examination information through the instant messaging tool, adjusts the timing and frequency of information sending, completes the push and records the push activities, and generates targeted information push records.

[0032] As a further solution of the present invention, the group behavior analysis module includes a group data collection submodule, a data comparison and analysis submodule, and a trend induction submodule;

[0033] The group data collection submodule records the user group's response to the information, including click, read and share behaviors, based on the targeted information push records, and constructs and outputs a user group interaction data set;

[0034] The data comparison and analysis submodule uses the user group interaction data set to compare the behavioral responses between different user groups, including the interaction type, frequency and persistence, reveal the group behavior characteristics, and output the group behavior difference analysis record;

[0035] The trend summarization submodule performs trend analysis to extract group interests and behavior patterns based on the group behavior difference analysis records, summarizes unified group characteristics and behavior trends, and outputs group behavior analysis results.

[0036] As a further solution of the present invention, the cross-domain synchronization update module includes a label inspection submodule, a behavior pattern synchronization submodule, and a label update submodule;

[0037] The tag review submodule reviews the tag library based on the group behavior analysis results, filters and marks tags that no longer reflect current user behavior, evaluates the current validity and relevance of each tag, and generates tag review records;

[0038] The behavior pattern synchronization submodule uses the label audit record to summarize and compare user behavior data on different platforms, analyze cross-platform behavior consistency, identify labels that need to be updated, and output synchronization demand analysis records;

[0039] The label update submodule updates and synchronizes the label library on all platforms according to the synchronization demand analysis record, dynamically adjusts the label classification and weight matching behavior pattern changes, verifies the label consistency and update immediacy between platforms, and generates label synchronization update records.

[0040] A push method based on tag identification includes the following steps:

[0041] S1: Collect user personal information, synchronously record the user's click, reading and sharing behavior data on the platform, filter key data fields, unify the format and correct errors, sort the behavior data according to time series, and merge personal information to generate a basic user behavior data set;

[0042] S2: Analyze the indicators in the user basic behavior data set, identify the regular patterns and preference trends of user behaviors, analyze and record the frequency of interaction between users and target content, build personalized tags for each user, and generate a user tag preference set;

[0043] S3: According to the user tag preference set, identify key data points of tag-user interaction, reveal the user's preference pattern for tags, calculate the weight score of each tag, dynamically adjust the weight of each tag based on the user interaction data, and output the adjusted tag weight;

[0044] S4: re-ordering the tags using the adjusted tag weights, filtering recruitment and examination information associated with the user tags in combination with the database, confirming that the content matches the user's major and interests, and generating a filtered content record;

[0045] S5: Using the screening content record and combining it with the user's online activity log, a time series analysis is performed to determine the user's most frequently active time period, and the screened recruitment and examination information is pushed through the instant messaging tool during the user's most active time period to generate a targeted information push record;

[0046] S6: Based on the targeted information push record, record the user group's reaction to the information, compare the behavioral responses between different user groups, reveal the group behavior characteristics, perform trend analysis to summarize unified group characteristics and behavior trends, and output the group behavior analysis results;

[0047] S7: Based on the group behavior analysis results, evaluate the current validity and relevance of each tag, summarize and compare user behavior data on different platforms, identify tags that need to be updated, dynamically adjust tag classification and weight matching behavior pattern changes, and generate tag synchronization update records.

[0048] Compared with the prior art, the advantages and positive effects of the present invention are:

[0049] In the present invention, by capturing real-time user behavior and dynamically adjusting personalized tags, the relevance and accuracy of information push can be significantly improved. By continuously monitoring user interactions, it is possible to respond instantly to changes in user preferences, automatically adjust push strategies, enhance user experience, and ensure that users always receive the information that best suits their current interests and needs. Through group behavior data analysis, it is possible to capture and predict industry trends, adjust information flows in advance, and optimize information distribution efficiency in a multi-user environment. The synchronous update of cross-domain tags further improves wide adaptability and data consistency, making the information push process smoother and more effective. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is a system flow chart of the present invention;

[0051] Figure 2 The flowchart of obtaining the basic behavior data set of users of the present invention;

[0052] Figure 3 A flowchart of obtaining a user tag preference set according to the present invention;

[0053] Figure 4 This is a flowchart for obtaining the optimized tag weight information of the present invention;

[0054] Figure 5 This is a flowchart for obtaining targeted information push records of the present invention;

[0055] Figure 6 A flowchart for obtaining group behavior analysis results of the present invention;

[0056] Figure 7 A flowchart for obtaining a synchronous update record of a tag of the present invention;

[0057] Figure 8 The figure is a flow chart of the steps of the method of the present invention. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0059] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0060] See also Figure 1 , a push system based on tag recognition includes:

[0061] The user data collection module collects the personal information of users registered through their mobile phone numbers, simultaneously collects the user's behavior data on the platform, organizes the data and processes it in a standardized format to generate a user basic behavior data set;

[0062] The label customization module analyzes the user's basic behavior data set, captures the user's behavior pattern and frequently interacted content, determines the user's label customization options, and records the corresponding information to obtain the user's label preference set;

[0063] The dynamic tag optimization module performs data analysis based on the user tag preference set, identifies the interaction between tags and users, adjusts tag weights according to the interaction data, reorders tag priorities, and obtains optimized tag weight information;

[0064] The information matching and pushing module uses the optimized tag weight information to filter the recruitment and examination information associated with the user tag in the database, combines the user's activity time record, pushes information through instant messaging tools, and generates targeted information push records;

[0065] The group behavior analysis module combines targeted information push records, collects behavioral data of different user groups, compares and analyzes data, identifies interest trends and summarizes group characteristics to obtain group behavior analysis results;

[0066] The cross-domain synchronous update module conducts an overall review of the tags based on the results of group behavior analysis, synchronously updates the tag library based on user feedback and behavior patterns on different platforms, determines the consistency and real-time nature of user tags on all platforms, and generates tag synchronization update records.

[0067] The user basic behavior data set includes user personal information, behavior records and standardized data; the user tag preference set includes behavior pattern records, interactive content and user tag selection results; the optimized tag weight information includes weight adjustment records, priority sorting results and tag interaction analysis records; the targeted information push records include screening results, push time records and communication tool usage records; the group behavior analysis results include interest trend analysis results, group characteristics and group behavior data analysis records; the tag synchronization update records include tag update records, platform synchronization results and real-time verification records.

[0068] See also Figure 2 ,The user data collection module includes an information collection submodule, a data collating submodule, and a data set generation submodule;

[0069] The information collection submodule collects user personal information, simultaneously records the user's click, reading and sharing behavior data on the platform, captures behavior data points to build behavior trajectories, and obtains user behavior information;

[0070] After the information collection submodule records the user's click, reading and sharing behavior data on the platform, it begins to capture various data points to construct a behavior trajectory. In the specific process, the system first identifies each user's activity, such as the page browsing timestamp, click link and its corresponding page element. The data is then captured and marked through a real-time processing flow, and then the original data points are clustered and analyzed to identify behavior patterns and repeated behavior sequences, such as continuous clicks on the same function or similar reading behaviors at similar times. After clustering, these data points form clear behavior paths, so that the same user behavior sequences can be effectively identified and linked in the subsequent data sorting process. After obtaining the user behavior information, this information is synchronously stored in the database for further analysis.

[0071] The data sorting submodule selects key data fields based on user behavior information, unifies the format and corrects errors, performs deduplication and normalization on the data, and outputs standardized user data;

[0072] After screening key data fields, the data sorting submodule performs the detailed process of format unification and error correction, including identifying inconsistencies in various data formats, such as date and time formats, decimal places of numerical values, or encoding methods of text data, and using a predefined rule base to verify and convert data formats to ensure that all data conform to unified standards. Error detection and correction are then performed, including data entry errors, outliers automatically generated by the system, or data corruption problems. Algorithms are used to compare the deviations of each field data from industry standards, and outliers are marked and corrected. The system then performs deduplication operations to detect duplicate records in the data set, and uses hash table technology to quickly identify and remove identical data items. Normalization processing is to scale or convert data by calculating the proportional relationship between fields to maintain data consistency and comparability. The final output of standardized user data includes a clean and consistent data set that has been verified, corrected, deduplicated, and normalized, providing a reliable basis for data analysis and decision-making.

[0073] The dataset generation submodule uses standardized user data to sort the behavior data in time series, merges personal information with the sorted behavior data, and generates a user basic behavior dataset;

[0074] After the dataset generation submodule uses standardized user data, the detailed process of time series sorting of behavior data involves first sorting the behavior data of each user according to the timestamp. The system rearranges the data by comparing the order of timestamps to ensure that the behavior data is recorded in the order of actual occurrence. During the sorting process, merge sort or quick sort is used to handle the sorting needs of large amounts of data. The subsequent step of merging personal information with the sorted behavior data includes associating static data such as personal basic information such as name, age, gender, etc. with dynamic behavior data. The system uses the user ID as the key value to associate and merge the static information table with the behavior log table. The generated user basic behavior dataset contains a complete user profile, which includes both static personal information and behavior trajectories arranged in chronological order, providing a basis for subsequent user behavior analysis and pattern recognition.

[0075] See also Figure 3 ,The label customization module includes a behavior pattern analysis submodule, an ,interactive content identification submodule, and a label generation submodule;

[0076] The behavior pattern analysis submodule analyzes the indicators in the user's basic behavior data set, including login frequency, active time period, and function usage preference, identifies the regular patterns and preference trends of user behavior, extracts the statistical characteristics of activities, and outputs the behavior pattern feature set;

[0077] The behavior pattern analysis submodule first identifies the regular patterns and preference trends of user behavior, including using data mining technology to analyze login frequency, active time periods and function usage preferences. The specific method is to collect user login records and activity time data, and determine the high-frequency login time periods and main active intervals through statistical analysis of the data. At the same time, the system tracks the frequency of users' use of different functions, and identifies the functions and their combination patterns commonly used by users through the frequent item set algorithm. Further, user behavior is divided into different groups through cluster analysis. Each group represents a behavior pattern, such as one group of users who log in in the morning and often use the messaging function, and another group of users who are active at night and prefer entertainment content. In addition, the user's behavior changes are analyzed, and the dynamic trend of behavior is captured through time series analysis technology to update and refine the identification of behavior patterns. The output behavior pattern feature set includes the user's main activity time, preferred functions and behavior trends, providing data support for subsequent personalized content recommendations and user interface optimization.

[0078] The interactive content identification submodule analyzes and records the frequency of interaction between users and target content based on the behavioral pattern feature set, and deeply mines the content types and topics preferred by users based on the interaction frequency to obtain the interactive preference set;

[0079] The interactive content identification submodule analyzes and records the frequency of interaction between users and target content in detail based on the behavioral pattern feature set. This process begins with an accurate count of the number of times users interact with the content, including clicks, browsing, and long-term stays. The interactive data is analyzed using data mining technology to identify the content types and topics with which users frequently interact. Specific technologies include association rule learning and topic modeling. For example, the Apriori algorithm is used to extract common content browsing patterns from a large amount of user behavior data, and the LDA topic model is applied to determine the topic areas where users are most interested, such as sports or technology news. This process is continuously iterated and optimized to gradually refine and refine users' content preferences. The final interactive preference set describes in detail the content types and topics preferred by users.

[0080] The label generation submodule uses the interaction preference set combined with the behavior pattern feature set to construct personalized labels for each user, record the relevant information and attributes of the label, and perform data integration to generate the user label preference set;

[0081] The label generation submodule uses the interaction preference set combined with the behavior pattern feature set to construct personalized labels for each user. This process involves comprehensive analysis of the content preferences in the interaction preference set and the user behavior data in the behavior pattern feature set. First, the system compares the user's behavior data with their interaction preferences through a series of algorithms to identify the most significant behavior characteristics and content preferences. Then, using this information, the submodule uses machine learning techniques, such as decision trees and support vector machines (SVMs), to train models to generate labels that describe user-specific preferences, such as "morning active users" and "technology news enthusiasts." During this process, the system also records and updates the relevant information and attributes of each label, such as the label generation time, effectiveness evaluation, and user feedback. Ultimately, the personalized labels are integrated into a user label preference set, which records each user's label data in detail, providing a data basis for marketing strategies and personalized services, ensuring the personalization of service content and the continuity of user experience.

[0082] See also Figure 4 ,The dynamic tag optimization module includes the behavior data analysis submodule, the tag weight adjustment submodule, and the priority sorting submodule;

[0083] The behavior data analysis submodule identifies the key data points of the interaction between tags and users based on the user tag preference set, performs cluster analysis on the data, reveals the user's preference pattern for tags, and generates tag interaction analysis records;

[0084] After the behavioral data analysis submodule identifies key data points based on the user's tag preference set, the process uses the K-means clustering algorithm to group the data points. First, the distance between each data point is calculated as the basis for clustering, and the user's preference pattern for tags is identified. For example, the user's click frequency, browsing time, and repeated interaction pattern for a certain tag are considered as one category. The system then reveals the user's preferences based on these clustering results, such as preference for news content or game tags. Subsequently, the generated tag interaction analysis records include the number of users in each cluster, the main interaction types and their frequencies. These records describe in detail the user group's interaction habits and preference strength for various tags, providing a basis for further tag weight adjustment, and finally outputting the adjusted tag weight information. This information includes the new weight score for each tag, reflecting the relationship between the user's actual preference and the previously set weight and the adjusted priority.

[0085] The tag weight adjustment submodule calculates the weight score of each tag based on the tag interaction analysis record, dynamically adjusts the weight of each tag based on the user interaction data, verifies the consistency between the weight and the user's actual preference, and outputs the adjusted tag weight;

[0086] The weight score of each tag is as follows:

[0087] ;

[0088] Calculate, where represents the calculated label weight score, Represents the user's reaction intensity to label j, including clicks, likes, comments and other interactive data, which is used to reflect the user's active interaction with the target label. represents the frequency of occurrence of tag j in the user's historical interactions, which is used to measure the popularity of the tag and the user's consistent preference for the tag. C represents the average interaction frequency of all tags, which is used to evaluate the performance of each tag relative to the average level. j represents the tag, and n represents the number of tags.

[0089] : The intensity of the user's reaction to label j, obtained by counting the user's interaction data (clicks, likes, comments). Collect labels 100 clicks, 50 likes, and 30 comments in the past week, then:

[0090] ;

[0091] : The frequency of occurrence of tag j in the user's historical interactions, calculated by dividing the total number of interactions for this tag by the total number of interactions for all tags. The total number of interactions collected is 1000 times, and the tag The number of interactions is 200, then:

[0092] ;

[0093] C: The average interaction frequency of all tags, which is the arithmetic mean of the interaction frequencies of all tags. Suppose the interaction frequencies of three tags are 0.2, 0.15, and 0.1 respectively, then:

[0094] ;

[0095] Take collecting 3 tags as an example, the specific data is:

[0096] Label 1: .

[0097] Tag 2: .

[0098] Tag 3: .

[0099] Average interaction frequency .

[0100] Calculation process:

[0101] ;

[0102] The results show that the weight score of the label can be obtained to be approximately ,The score reflects the importance and attractiveness of the tag in user ,interaction (the larger score indicates high importance and attractiveness).,By quantifying user interaction and tag popularity, the tag weight can ,be effectively adjusted dynamically to better adapt to the actual preferences of users.

[0103] The priority sorting submodule uses the adjusted tag weights to reorder the tags, optimize the display priority of user-preferred tags, and output the optimized tag weight information;

[0104] The priority sorting submodule uses the adjusted tag weights to reorder the tags. It first receives the weight scores from the tag weight adjustment submodule, and then uses the scores to execute the sorting algorithm. Specifically, the quick sorting algorithm is used. This algorithm quickly determines the position of each tag by comparing the weight scores between tags, ensuring that tags with high weights are displayed higher on the user interface. After the sorting is completed, the sorting results are converted into optimized tag weight information, including the final ranking and relative position of each tag. The output optimized tag weight information is used to update the user interface so that the tags displayed on the interface can more accurately reflect the user's latest preferences. This process ensures that users can see the content that best suits their interests every time they visit the platform, thereby improving user satisfaction and interaction efficiency.

[0105] See also Figure 5 ,The information matching and pushing module includes an information screening submodule, a user activity time matching submodule, and a pushing record generating submodule;

[0106] The information screening submodule, based on the optimized tag weight information, screens the recruitment and examination information directly related to the user tag through database query, confirms that the content matches the user's major and interests, and generates a screening content record;

[0107] The information screening submodule uses the optimized tag weight information to screen recruitment and examination information directly related to the user's tags from the database. The detailed execution process includes constructing SQL query statements to perform precise matching on the user's professional and interest tags. Multiple parameters are considered when designing the query statements, such as tag weights, user professional fields, and interest preferences. The system filters recruitment and examination information by tags to ensure that only content that is highly relevant to the user's tags is selected. Then, a detailed screening content record is generated based on the screened content. The record includes the source of the information, relevant tags, matching degree, and a detailed description of the content, providing users with customized information that matches their profession and interests. The final generated screening content record includes all eligible recruitment and examination information, providing basic data for subsequent user activity time matching and information push.

[0108] The user activity time matching submodule uses the filtered content records and combines them with the user's online activity logs to perform time series analysis to determine the user's most frequently active time period, and obtain the user activity time analysis record;

[0109] The user activity time matching submodule uses the filtered content records and combines them with the user's online activity log to perform time series analysis. The execution process includes collecting the user's login time, activity duration and frequency, and using time series analysis tools such as the ARIMA model to analyze the user activity data to determine the most frequent active time period. During the analysis process, the module first sets the model parameters, such as seasonal factors and sliding window size, and then applies the parameters to the user activity log. The model generates a frequency distribution diagram of time points to identify the peak period of user activity, such as 9 to 11 in the morning or 8 to 10 in the evening. The generated user activity time analysis records include statistical charts and descriptions of the user's daily and weekly activity, providing precise time targets for optimizing the timing of information push. The final output user activity time analysis records list in detail the user's most active time periods, which are used to adjust subsequent information push plans.

[0110] The push record generation submodule analyzes the user activity time records, selects the most active time period of the user to push the selected recruitment and examination information through the instant messaging tool, adjusts the timing and frequency of information sending, completes the push and records the push activities, and generates targeted information push records;

[0111] The push record generation submodule analyzes the user activity time records and selects the time period when the user is most active to push the screened recruitment and examination information through instant messaging tools. The execution process includes setting the time trigger of the push system. The trigger is programmed according to the user activity time analysis record to ensure that the information is sent when the user is most likely to be online. The submodule adjusts the sending timing and frequency of the information and adopts a dynamic adjustment strategy to fine-tune the push frequency and content according to the user's feedback and degree of interaction. If the user is highly interactive with a certain push, the system will increase the push frequency of similar content, and vice versa. During this process, the system records the time of each push, the user's interactive response, and the details of the push content. After completing the push activity, the generated targeted information push record details the effect of each push and the user's feedback, providing data support for future push strategy optimization. The final output push record is a detailed activity log, including push time, content, user response, and adjustment measures.

[0112] See also Figure 6 ,The group behavior analysis module includes a group data collection submodule, a data comparison and analysis submodule, and a trend induction submodule;

[0113] The group data collection submodule records the user group's response to the information, including click, read and share behaviors, based on the targeted information push records, and constructs and outputs the user group interaction data set;

[0114] The group data collection submodule first records the user group's response to the information. This process involves extracting the user's click, reading and sharing behaviors from the targeted information push records. The system collects data through database query and log analysis. Specifically, each type of interaction, such as clicking a link, reading an article, sharing to social media, etc., is recorded separately and marked with time and duration. The data is aggregated to form a user group interaction data set. This data set records in detail the way and frequency of different user groups respond to different information, which is used for subsequent behavioral comparison and trend analysis. Through the collected data, the data comparison and analysis submodule can perform its function, compare the behavioral responses between different user groups, and reveal the group behavior characteristics. The output group behavior difference analysis record describes in detail the differences in interaction type, frequency and duration between groups, providing basic data for further trend induction and behavior pattern extraction.

[0115] The data comparison and analysis submodule uses the user group interaction data set to compare the behavioral responses between different user groups, including the interaction type, frequency and duration, reveal the group behavior characteristics, and output the group behavior difference analysis record;

[0116] The data comparison analysis submodule uses the user group interaction data set to perform comparative analysis of behavioral responses. This process involves using statistical analysis methods and data visualization tools to compare behavioral differences between different user groups. First, the submodule extracts data on the type, frequency, and duration of interaction for each group, and uses statistical tests such as analysis of variance (ANOVA) and chi-square tests to evaluate the significance of these differences. Visual techniques such as heat maps and bar charts are then used to display the differences, making the comparison between behavioral patterns more intuitive and easy to understand. During the analysis, attention is paid to abnormal points in the behavioral patterns or particularly active responses in order to identify potential group behavior trends. The output group behavior difference analysis records list in detail the main differences between the groups, such as one group participating in sharing more frequently while another group is more inclined to read content. These records provide precise input data for the trend summary submodule for more in-depth trend and interest analysis.

[0117] The trend induction submodule analyzes the group behavior difference records, performs trend analysis to extract group interests and behavior patterns, summarizes unified group characteristics and behavior trends, and outputs group behavior analysis results;

[0118] The trend induction submodule performs trend analysis based on the group behavior difference analysis records and extracts group interests and behavior patterns. The execution process includes using time series analysis and cluster analysis techniques to summarize unified group characteristics and behavior trends. Specifically, the submodule identifies long-term and short-term trends by analyzing data changes at different time points. For example, moving average or exponential smoothing techniques are used to smooth data and highlight trend lines. In addition, hierarchical clustering or K-means clustering methods are used to group user behavior data to find groups with similar behavioral characteristics. The group behavior analysis results generated during the analysis process include clear descriptions of behavioral patterns and graphical displays of group interests, such as a group's continued attention to technology news or seasonal interest in entertainment content. The results provide a basis for the platform to optimize user experience and content recommendation strategies in a targeted manner. The final output group behavior analysis results list in detail the interests and behavior trends of each group, providing data support for future marketing and product development.

[0119] See also Figure 7 ,The cross-domain synchronization update module includes a label inspection submodule, a behavior pattern synchronization submodule, and a label update submodule;

[0120] The tag review submodule reviews the tag library based on the group behavior analysis results, filters and marks tags that no longer reflect current user behavior, evaluates the current validity and relevance of each tag, and generates tag review records;

[0121] The tag review submodule begins to review the tag library by analyzing the results of group behavior analysis. The execution process includes using data analysis tools, such as SQL queries and Python scripts, to extract tag usage frequency and interaction type data from group behavior data. The system evaluates the current effectiveness and relevance of each tag based on the extracted data. For example, it conducts statistical analysis on the number of clicks, reading depth, and sharing rate of each tag, and uses indicators to determine which tags no longer reflect current user behavior trends and mark these tags as inactive or outdated. In addition, the submodule also compares the historical data of the tags to identify tags that need to be adjusted or deleted. The final generated tag audit record contains the effectiveness score and recommended actions for each tag, such as retain, update, or delete, providing detailed data for the behavior pattern synchronization submodule for cross-platform behavior consistency analysis and tag updates.

[0122] The behavior pattern synchronization submodule uses label audit records to aggregate and compare user behavior data on different platforms, analyze cross-platform behavior consistency, identify labels that need to be updated, and output synchronization demand analysis records;

[0123] The behavior pattern synchronization submodule uses label audit records to summarize and compare user behavior data on different platforms. The execution process covers the integration of cross-platform data. Specific operations include importing user behavior and label usage data from the databases of each platform, and using data integration tools such as ETL (Extract, Transform, Load) to process these data to ensure unified format and data integrity. Then, the module applies statistical analysis methods, such as multi-dimensional variance analysis, to compare the differences in responses of users on different platforms to the same label. Through these analyses, it is identified which labels show behavioral consistency across platforms and which show behavioral differences due to regional or cultural differences. The output synchronization requirements analysis record lists in detail the labels that need to be updated and specific update suggestions, such as adding new labels, modifying the definition of existing labels, or adjusting label weights, which provides a basis for the label update submodule to synchronize and maintain the label library on all platforms.

[0124] The tag update submodule analyzes the records based on synchronization requirements, updates and synchronizes the tag library on all platforms, dynamically adjusts the tag classification and weight matching behavior pattern changes, verifies the tag consistency and update immediacy between platforms, and generates tag synchronization update records;

[0125] The tag update submodule updates and synchronizes the tag libraries on all platforms according to the synchronization requirements analysis records. The execution process includes first determining the list of tags that need to be updated or synchronized, and then using the central management system to add, delete or modify tags on each platform through API or direct database operations. This includes dynamically adjusting tag classifications and weights to match changes in behavior patterns, for example, adding new tag categories or increasing the visibility weights of certain tags. Next, the module performs a series of consistency checks to ensure that the tag libraries of all platforms accurately reflect the latest tag status. The inspection process uses automated scripts to compare the consistency of tag data on each platform to ensure that there are no errors or delays. The final generated tag synchronization update record records the update details of each platform, such as the number of updated tags, the user groups affected, and the consistency status of the data after synchronization, providing the management team with a comprehensive view of tag management and synchronization effects.

[0126] See also Figure 8 , a push method based on tag identification, comprising the following steps:

[0127] S1: Collect user personal information, synchronously record the user's click, reading and sharing behavior data on the platform, filter key data fields, unify the format and correct errors, sort the behavior data according to time series, and merge personal information to generate a basic user behavior data set;

[0128] S2: Analyze the indicators in the user basic behavior data set, identify the regular patterns and preference trends of user behavior, analyze and record the frequency of user interaction with target content, build personalized tags for each user, and generate user tag preference sets;

[0129] S3: Based on the user tag preference set, identify the key data points of tag-user interaction, reveal the user's preference pattern for tags, calculate the weight score of each tag, dynamically adjust the weight of each tag based on user interaction data, and output the adjusted tag weight;

[0130] S4: using the adjusted tag weights, reordering the tags, filtering the recruitment and examination information associated with the user tags in combination with the database, confirming that the content matches the user's major and interests, and generating a filtered content record;

[0131] S5: Use the filtered content records and combine them with the user's online activity logs to perform time series analysis to determine the user's most active time period, select the user's most active time period to push the filtered recruitment and examination information through instant messaging tools, and generate targeted information push records;

[0132] S6: Based on the targeted information push records, record the user group's response to the information, compare the behavioral responses between different user groups, reveal the group behavior characteristics, conduct trend analysis to summarize the unified group characteristics and behavior trends, and output the group behavior analysis results;

[0133] S7: Based on the results of group behavior analysis, evaluate the current validity and relevance of each tag, summarize and compare user behavior data on different platforms, identify tags that need to be updated, dynamically adjust tag classification and weight matching behavior pattern changes, and generate tag synchronization update records.

[0134] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A push system based on tag recognition, characterized by: The system comprises: The user data collection module collects the personal information of users registered through their mobile phone numbers, simultaneously collects the user's behavior data on the platform, organizes the data and processes it in a standardized format to generate a user basic behavior data set; The label customization module analyzes the user basic behavior data set, captures the user behavior pattern and frequent interaction content, determines the user label customization options, and records the corresponding information to obtain the user label preference set; The dynamic tag optimization module performs data analysis based on the user tag preference set, identifies the interaction between tags and users, adjusts tag weights according to the interaction data, reorders tag priorities, and obtains optimized tag weight information; The information matching and pushing module uses the optimized tag weight information to filter the recruitment and examination information associated with the user tag in the database, combines the user activity time record, pushes information through the instant messaging tool, and generates targeted information push records; The group behavior analysis module combines the targeted information push records, collects the behavior data of different user groups, performs data comparison and analysis, identifies interest trends and summarizes group characteristics, and obtains group behavior analysis results; The cross-domain synchronization update module conducts an overall review of the tags based on the group behavior analysis results, synchronously updates the tag library based on user feedback and behavior patterns on different platforms, determines the consistency and real-time nature of user tags on all platforms, and generates tag synchronization update records; The dynamic tag optimization module includes a behavior data analysis submodule, a tag weight adjustment submodule, and a priority sorting submodule; The behavior data analysis submodule identifies key data points of the interaction between tags and users based on the user tag preference set, performs cluster analysis on the data, reveals the user's preference pattern for tags, and generates tag interaction analysis records; The tag weight adjustment submodule calculates the weight score of each tag based on the tag interaction analysis record, dynamically adjusts the weight of each tag in combination with the user interaction data, verifies the consistency of the weight with the user's actual preference, and outputs the adjusted tag weight; The priority sorting submodule uses the adjusted tag weights to re-sort the tags, optimize the display priority of user-preferred tags, and output optimized tag weight information; The weight score of each tag is calculated according to the formula: ； Calculate, where represents the calculated label weight score, represents the intensity of the user's reaction to label j, represents the frequency of occurrence of tag j in the user's historical interactions, C represents the average interaction frequency of all tags, j represents the tag, and n represents the number of tags.

2. The push system based on tag identification according to claim 1, characterized in that: The user basic behavior data set includes user personal information, behavior records and standardized data, the user tag preference set includes behavior pattern records, interactive content and user tag selection results, the optimized tag weight information includes weight adjustment records, priority sorting results and tag interaction analysis records, the targeted information push records include screening results, push time records and communication tool usage records, the group behavior analysis results include interest trend analysis results, group characteristics and group behavior data analysis records, and the tag synchronization update records include tag update records, platform synchronization results and real-time verification records.

3. The push system based on tag identification according to claim 1, characterized in that: The user data collection module includes an information collection submodule, a data sorting submodule, and a data set generation submodule; The information collection submodule collects user personal information, simultaneously records the user's click, reading and sharing behavior data on the platform, captures behavior data points to build behavior trajectories, and obtains user behavior information; The data sorting submodule screens key data fields according to the user behavior information, performs format unification and error correction, performs deduplication and normalization processing on the data, and outputs standardized user data; The data set generation submodule uses the standardized user data to sort the behavior data according to the time series, merges the personal information with the sorted behavior data, and generates a user basic behavior data set.

4. The push system based on tag identification according to claim 1, characterized in that: The label customization module includes a behavior pattern analysis submodule, an interactive content identification submodule, and a label generation submodule; The behavior pattern analysis submodule analyzes the indicators in the user basic behavior data set, including login frequency, active time period, and function usage preference, identifies the regular patterns and preference trends of user behavior, extracts the statistical characteristics of activities, and outputs a behavior pattern feature set; The interactive content identification submodule analyzes and records the frequency of interaction between the user and the target content according to the behavioral pattern feature set, and deeply mines the content types and topics preferred by the user according to the interaction frequency to obtain an interactive preference set; The label generation submodule uses the interaction preference set in combination with the behavior pattern feature set to construct a personalized label for each user, record the relevant information and attributes of the label, and perform data integration to generate a user label preference set.

5. The push system based on tag identification according to claim 1, characterized in that: The information matching and pushing module includes an information screening submodule, a user activity time matching submodule, and a pushing record generating submodule; The information screening submodule screens the recruitment and examination information directly associated with the user tag through database query based on the optimized tag weight information, confirms that the content matches the user's major and interests, and generates a screening content record; The user activity time matching submodule uses the screening content record and combines it with the user's online activity log to perform time series analysis to determine the user's most frequently active time period, and obtains the user activity time analysis record; The push record generation submodule analyzes the user activity time records, selects the most active time period of the user to push the screened recruitment and examination information through the instant messaging tool, adjusts the timing and frequency of information sending, completes the push and records the push activities, and generates targeted information push records.

6. The push system based on tag identification according to claim 1, characterized in that: The group behavior analysis module includes a group data collection submodule, a data comparison and analysis submodule, and a trend induction submodule; The group data collection submodule records the user group's response to the information, including click, read and share behaviors, based on the targeted information push records, and constructs and outputs a user group interaction data set; The data comparison and analysis submodule uses the user group interaction data set to compare the behavioral responses between different user groups, including the interaction type, frequency and persistence, reveal the group behavior characteristics, and output the group behavior difference analysis record; The trend summarization submodule performs trend analysis to extract group interests and behavior patterns based on the group behavior difference analysis records, summarizes unified group characteristics and behavior trends, and outputs group behavior analysis results.

7. The push system based on tag identification according to claim 1, characterized in that: The cross-domain synchronization update module includes a label inspection submodule, a behavior mode synchronization submodule, and a label update submodule; The tag review submodule reviews the tag library based on the group behavior analysis results, filters and marks tags that no longer reflect current user behavior, evaluates the current validity and relevance of each tag, and generates tag review records; The behavior pattern synchronization submodule uses the label audit record to summarize and compare user behavior data on different platforms, analyze cross-platform behavior consistency, identify labels that need to be updated, and output synchronization demand analysis records; The label update submodule updates and synchronizes the label library on all platforms according to the synchronization demand analysis record, dynamically adjusts the label classification and weight matching behavior pattern changes, verifies the label consistency and update immediacy between platforms, and generates label synchronization update records.

8. A push method based on tag identification, characterized in that: The push system based on tag identification according to any one of claims 1 to 7 comprises the following steps: Collect user personal information, synchronously record the user's click, reading and sharing behavior data on the platform, screen key data fields, unify the format and correct errors, sort the behavior data according to time series, and merge personal information to generate a basic user behavior data set; Analyze the indicators in the user basic behavior data set, identify the regular patterns and preference trends of user behavior, analyze and record the frequency of interaction between users and target content, build personalized tags for each user, and generate a user tag preference set; According to the user tag preference set, identify key data points of tag-user interaction, reveal the user's preference pattern for tags, calculate the weight score of each tag, dynamically adjust the weight of each tag in combination with the user interaction data, and output the adjusted tag weight; Re-order the tags using the adjusted tag weights, filter the recruitment and examination information associated with the user tags in combination with the database, confirm that the content matches the user's major and interests, and generate a filter content record; Using the screening content records, combined with the user's online activity log, time series analysis is performed to determine the user's most frequently active time period, and the screened recruitment and examination information is pushed through instant messaging tools during the user's most active time period to generate targeted information push records; Based on the targeted information push records, record the user group's reaction to the information, compare the behavioral responses between different user groups, reveal the group behavior characteristics, perform trend analysis to summarize unified group characteristics and behavior trends, and output group behavior analysis results; Based on the group behavior analysis results, the current validity and relevance of each tag are evaluated, the user behavior data on different platforms are summarized and compared, the tags that need to be updated are identified, the tag classification and weight are dynamically adjusted to match the behavior pattern changes, and the tag synchronization update record is generated.

Images