Marketing mobile terminal security monitoring and dynamic response method and system

Abstract

The application discloses a marketing mobile terminal security monitoring and dynamic response method and system, belongs to the technical field of information monitoring and response, and aims to solve the problems of audience positioning ambiguity, data fragmentation, rigid delivery strategy and difficult control of security risks in traditional information reach; the method comprises the following steps: positioning potential audience characteristics and constructing a target population characteristic system, matching terminal labels to define the delivery range; collecting terminal user behavior data in the delivery range, extracting features to divide information categories, and constructing a multi-dimensional target information database; clustering features to generate differentiated delivery strategies, updating data and reclassifying, dividing sub-scenes to construct behavior baselines; identifying and removing abnormal terminals based on the baseline, constructing a category transfer path atlas, evaluating the strategy effect and judging whether to terminate delivery according to the standard; the application realizes the precision, security and dynamic adaptation of information reach delivery, and significantly improves the delivery efficiency and effect.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information monitoring and response, and more particularly to a marketing mobile terminal security monitoring and dynamic response method and system. BACKGROUND

[0002] In the practice of marketing mobile terminal deployment and management, the existing mode has many problems to be solved: often relying on single-dimensional user data such as basic registration information to divide the crowd, failing to integrate user consumption habits, scene needs and other characteristics, resulting in fuzzy target audience portrait, only wide-net type deployment, causing resource waste; the data generated in the process of target information deployment, such as terminal operation, user interaction and conversion, are scattered in different systems, the formats are not unified and lack of association to form data islands, which are difficult to be reused by the system, cannot provide comprehensive support for strategy optimization, and cannot realize terminal behavior tracking in the whole link, affecting the scientific nature of target information deployment decision; the same content push, frequency setting and interaction mode are used for all reached mobile terminals, ignoring the behavior preference differences of different terminal users, resulting in insufficient strategy adaptability, low conversion efficiency of core users and limited activation effect of potential users; due to the lack of effective definition of terminal behavior regularity, it is difficult to distinguish normal behavior fluctuation from abnormal state, and abnormal terminals cannot be identified and removed in time, which not only affects the accuracy of deployment data, but also may cause problems such as abuse of target information deployment resources and damage to user experience; at the same time, the evaluation of deployment effect focuses on the final conversion data, lacks tracking of the evolution process of terminal behavior, and cannot determine the actual impact of the strategy, and relies on manual experience to determine the termination time of deployment, lacks scientific standards, either stops too early and does not reach the target of target information deployment, or continues to deploy and causes resource redundancy, and cannot realize dynamic optimization and closed-loop management of deployment effect. Therefore, in order to overcome these limitations, the present application provides a marketing mobile terminal security monitoring and dynamic response method and system. SUMMARY

[0003] In view of the deficiencies in the prior art, the purpose of the present application is to provide a marketing mobile terminal security monitoring and dynamic response method and system to solve the problems of fuzzy audience positioning, fragmented data difficult to reuse, lack of targeted deployment strategy, difficulty in identifying abnormal terminals and difficulty in evaluating and dynamically adjusting the deployment effect in marketing mobile terminal deployment.

[0004] To achieve the above-mentioned purpose, the present application provides the following technical scheme:

[0005] The marketing mobile terminal security monitoring and dynamic response method comprises:

[0006] Locating the characteristics of the potential audience group of the target information to be deployed, constructing a target audience characteristic system, matching the terminal tags of the full marketing mobile terminals of the deployment platform, and delimiting the deployment range of the target information;

[0007] In the range of the launch, the user behavior data of each marketing mobile terminal for the target information is collected and the behavior value features are extracted, the information categories of each marketing mobile terminal are divided, and a multi-dimensional target information database is constructed;

[0008] The behavior value features of each marketing mobile terminal under each information category are clustered and the typical behavior value features are screened to match the launch parameters of the target information under each information category, to generate a differentiated launch response strategy, launch the target information, update the multi-dimensional target information database, divide the sub-scene under each information category, construct a scene-specific dataset, calculate the common range and fluctuation threshold of the behavior value features, and generate the behavior baseline of each information category;

[0009] Based on the behavior baseline, abnormal terminals are identified and excluded, the transfer path of the marketing mobile terminal information category is constructed based on the directed graph, the remaining marketing mobile terminal information category transfer path atlas is generated through path aggregation, and the effect evaluation of the differentiated launch response strategy is performed to identify the strategy execution anomaly and determine whether to terminate the target information launch in combination with the preset stop standard.

[0010] Specifically, the step of delimiting the launch range of the target information includes:

[0011] The potential audience group features of the target users of the target information to be launched are located, and each potential audience group feature is assigned a feature priority to construct a potential audience group feature set;

[0012] Based on the potential audience group feature set, a target population feature system is constructed, which includes a feature coding layer, a feature association layer, and a feature adaptation layer;

[0013] The feature coding layer is used for standardized coding of the potential audience group features; the feature association layer is used for establishing the association relationship and association coefficient between different potential audience group features; and the feature adaptation layer is used for matching and docking the target population feature system with the terminal label of the marketing mobile terminal;

[0014] The running data of the full marketing mobile terminal of the launch platform is collected, the running data set of the full marketing mobile terminal is constructed, and the target population feature system is subjected to format normalization processing, the quantized feature data of the feature coding layer and the association relationship and association coefficient of the feature association layer are extracted, and a standardized target population feature matrix is constructed;

[0015] The running data set of the full marketing mobile terminal is called, the terminal data dimensions associated with the potential audience group features are screened, a terminal feature matrix is formed and matched with the feature dimensions of the target population feature matrix, which is used to calculate the association and matching scores of each marketing mobile terminal and the target population, and a terminal label is constructed;

[0016] Set the correlation matching score threshold and the basic range, filter all marketing mobile terminals based on the terminal tags of the marketing mobile terminals, locate the target marketing mobile terminal, and determine the delivery range of the target information.

[0017] Specifically, the correlation matching score of each marketing mobile terminal and the target population is calculated, and the step of constructing the terminal tag comprises:

[0018] The feature dimensions in the target population feature matrix and the corresponding dimensions in the terminal feature matrix are taken as evaluation indexes, a judgment matrix is constructed by pairwise comparison, and the basic weight values of each feature dimension of the target population feature matrix and the terminal feature matrix are calculated;

[0019] The dynamic feature dimensions in the terminal feature matrix are determined, the real-time data of the dynamic features of the marketing mobile terminal are synchronized according to a preset period, and a real-time weight adjustment factor is generated based on the fluctuation amplitude of the dynamic features; the real-time weight adjustment factor and the basic weight values of each dimension are weighted and fused to generate a comprehensive weight system;

[0020] For the feature dimensions corresponding to each other in the target population feature matrix and the terminal feature matrix, the feature values are multiplied by the corresponding dimension comprehensive weight to obtain weighted feature values; the cosine value of the angle between the feature vector of a single marketing mobile terminal and the feature system vector of the target population is calculated as the correlation matching score of the single marketing mobile terminal and the target population;

[0021] The hardware identification data of each marketing mobile terminal is collected, the hardware identification data is encrypted, a hardware encrypted string is generated, and dynamic parameters in the real-time running data of the marketing mobile terminal are extracted to generate a dynamic verification factor, which is spliced with the hardware encrypted string to form a terminal unique identifier;

[0022] Based on the correlation matching score and the terminal unique identifier, the terminal tag is constructed by fusing the basic information of the marketing mobile terminal and the information is structured and packaged.

[0023] Specifically, the step of constructing the multi-dimensional target information database comprises:

[0024] The user behavior data collection dimensions are set, and the user behavior data in each marketing mobile terminal is collected within the delivery range;

[0025] The collected user behavior data is preprocessed, including data cleaning, missing processing, data desensitization and data format normalization, to generate a standardized user behavior data set;

[0026] The standardized user behavior data set of each marketing mobile terminal is extracted to obtain behavior value features in each collection dimension, a quantization threshold of the behavior value features in each collection dimension is set, the behavior value features are quantitatively scored, a weight of each collection dimension is set and a terminal value total score is calculated, and each marketing mobile terminal is classified according to information categories based on the terminal value total score and the quantitatively scored behavior value features in each collection dimension;

[0027] The information categories include a core value behavior category terminal, an important auxiliary behavior category terminal and a general correlation behavior category terminal.

[0028] A multi-dimensional target information database is constructed by adopting a hierarchical storage combined with an association index architecture to store the standardized user behavior data set of the marketing mobile terminal in the delivery range, terminal labels and the information categories to which the terminal belongs.

[0029] Specifically, the user behavior data collection dimensions include an information receiving behavior dimension, an interactive operation behavior dimension, a conversion completion behavior dimension and a feedback submission behavior dimension.

[0030] The information receiving behavior dimension is used to associate the basic running state of the marketing mobile terminal and collect user behavior data generated in the process of passive or active information receiving by the user.

[0031] The interactive operation behavior dimension is used to collect user behavior data generated in the process of active operation intervention on the target information by the user.

[0032] The conversion completion behavior dimension is used to collect user behavior data generated in the process of completing the target information guide by the user.

[0033] The feedback submission behavior dimension is used to collect user behavior data generated in the process of active feedback on the target information by the user.

[0034] Specifically, the step of generating a differentiated delivery response strategy includes:

[0035] The behavior value features of the marketing mobile terminal of each information category are clustered to obtain feature clustering clusters of each information category, and the feature threshold ranges of the behavior value features are counted.

[0036] The contribution degrees of the behavior value features in the feature clustering clusters to the category attributes are calculated, and the behavior value features in the clustering clusters are selected as typical behavior value features according to the contribution degrees.

[0037] A behavior value feature and delivery parameter mapping table is established, a delivery parameter range corresponding to the value features is preset, the delivery parameter types and delivery parameter ranges corresponding to the typical behavior value features are extracted, and the corresponding delivery parameter values are selected in the preset delivery parameter range according to the feature threshold ranges of the typical behavior value features to construct a delivery list under each information type.

[0038] The delivery parameters in the delivery list are associated with the marketing mobile terminal, and a differentiated delivery response strategy of each information category is constructed.

[0039] Specifically, the step of generating the behavior baseline of each information category includes:

[0040] The user behavior data of each marketing mobile terminal under the differentiated delivery response strategy is collected, the multi-dimensional target information database is updated, the behavior value characteristics reflecting the behavior value under each collection dimension are extracted, the terminal value total score is calculated, and the information category division of each marketing mobile terminal is performed;

[0041] The user behavior data of the original data layer of the multi-dimensional target information database and the environmental data in the terminal tag are called, the scene characteristics are extracted, and the scene characteristics are cross combined to form the segmented scene under each information category;

[0042] The multi-dimensional target information database is called, the information category and the segmented scene identification are taken as the screening conditions, the behavior value characteristics of the user behavior data of all terminals in the corresponding scene are extracted, and the terminal tag is associated to form a scene exclusive data set;

[0043] For each scene exclusive data set, the quantile analysis method is used to calculate the common range and fluctuation threshold of the behavior value characteristics, a quantization benchmark under each segmented scene is formed, a behavior baseline of each segmented scene is generated, and a differentiated iteration period is set to update the behavior baseline.

[0044] Specifically, the step of generating the remaining marketing mobile terminal information category transfer path map includes:

[0045] The deviation degree of the real-time behavior value characteristics of the marketing mobile terminal and the corresponding behavior baseline is calculated, and if the deviation degree exceeds the preset deviation degree threshold, the behavior suspected abnormal terminal is marked;

[0046] The monitoring period is configured, and in the monitoring period, the user behavior data of the behavior suspected abnormal terminal is continuously collected, and if the ratio of the number of times of being marked as the behavior suspected abnormal terminal to the total number of judgment times is greater than the preset abnormality proportion threshold, the behavior suspected abnormal terminal is determined as the abnormal terminal;

[0047] The abnormal terminal is removed from the multi-dimensional target information database, and for the remaining marketing mobile terminal, the initial information category in the target information delivery process, the current information category at each time node, and the behavior value characteristics and behavior data collection time stamp are extracted;

[0048] When the information category of the marketing mobile terminal is transferred, the behavior value characteristics before and after the transfer are called, the change amplitude of each behavior value characteristic is calculated, and the key features driving the information category transfer are identified;

[0049] Information categories are treated as nodes in a directed graph, with node attributes including category identifier and collection timestamp; the transition relationship from the initial information category to the current information category on the marketing mobile terminal is treated as a directed edge, with edge attributes including transition start time, transition end time, and transition driving feature label.

[0050] The trajectory of changes in marketing mobile terminal information categories is linked to form a single terminal transfer path; and transfer paths with the same transfer direction are merged to generate a map of the remaining marketing mobile terminal information category transfer paths.

[0051] Specifically, the steps for identifying strategy execution anomalies and determining whether to terminate the delivery of target information based on preset stop criteria include:

[0052] Based on the marketing mobile terminal information category transfer path map, after the execution of the differentiated delivery response strategy under each information category, the transfer driving indicators are quantified by the proportion of terminals in the expected transfer direction of the differentiated delivery response strategy and the proportion of terminals whose transfer time meets the preset time.

[0053] The terminal value indicators are quantified by the ratio of the increase in the number of core value terminals after the launch to that before the launch, the increase in the average total value score of terminals after the launch to that before the launch, and the proportion of terminals with improved core conversion action completion rate among the terminals covered by the launch.

[0054] The transfer-driven indicators and terminal value indicators are standardized, and the comprehensive effect score of the differentiated delivery response strategy under each information category is calculated by weighted summation. The comprehensive effect score threshold and the qualified threshold of each dimension indicator are set to determine whether there is any abnormality in the differentiated delivery response strategy.

[0055] It also configures a delivery termination standard system, which consists of a tiered system of core target achievement standard thresholds and terminal value stability standard thresholds; calculates the quantitative indicators required for the termination standard, performs termination verification, and determines whether to terminate the delivery of target information.

[0056] The marketing mobile terminal security monitoring and dynamic response system includes a data construction module, a security monitoring module, an effectiveness evaluation module, and a campaign optimization module.

[0057] The data construction module generates terminal tags to define the delivery scope and builds a multi-dimensional target information database to provide data support for target information delivery; the security monitoring module identifies abnormal terminals by dynamically constructing user behavior baselines for each information category and subdivided scenario; the effect evaluation module evaluates the effect of differentiated delivery response strategies based on the terminal information category transfer path map and identifies strategy execution anomalies; and the delivery optimization module determines whether to terminate target information delivery based on preset stop criteria.

[0058] The beneficial effects of this invention are:

[0059] The application effectively solves the problem of traditional audience positioning ambiguity by constructing a multi-dimensional target person feature system that matches the target information delivery appeal, generating terminal tags that dynamically adjust with the real-time state of the terminal, realizing precise matching of marketing mobile terminals and target people, and avoiding resource waste caused by indiscriminate delivery; by constructing a hierarchical multi-dimensional target information database integrating full-link user behavior data, terminal tags and information categories, breaking the data silos of traditional target information delivery data fragmentation, providing comprehensive and reusable data support for delivery strategy optimization, and improving the scientific nature of target information delivery decision-making; by implementing differentiated delivery strategies for different information categories and dynamically constructing user behavior baselines in sub-scenarios, the problem of insufficient adaptability of traditional one-size-fits-all delivery is solved, and relying on baseline comparison and continuous monitoring, abnormal terminals are accurately identified and removed, ensuring the safety of the delivery environment; by generating terminal information category transition path maps, quantitatively evaluating strategy effectiveness and combining hierarchical stopping standards to realize delivery closed-loop optimization, the limitations of traditional delivery effect evaluation and termination timing relying on human experience are solved, dynamic tracking of terminal behavior evolution and precise iteration of strategies are realized, and ultimately the precision, safety, dynamic adaptability and resource utilization efficiency of target information delivery mobile terminal delivery are comprehensively improved, significantly improving core user conversion efficiency and potential user activation effect. BRIEF DESCRIPTION OF DRAWINGS

[0060] Figure 1 The flowchart of the marketing mobile terminal safety monitoring and dynamic response method of the application;

[0061] Figure 2 The flowchart of the application for defining the delivery range of target information;

[0062] Figure 3 The flowchart of the application for constructing a multi-dimensional target information database;

[0063] Figure 4 The flowchart of the application for generating differentiated delivery response strategies;

[0064] Figure 5 The flowchart of the application for generating a remaining marketing mobile terminal information category transition path map. DETAILED DESCRIPTION

[0065] Please refer to Figure 1 The embodiment introduces a marketing mobile terminal safety monitoring and dynamic response method, including:

[0066] Step S1: According to the delivery appeal and value positioning of the target information, the potential audience group characteristics of the target information to be delivered are positioned, a target group feature system is constructed, which is used for matching with the terminal label of the full marketing mobile terminal running data set of the delivery platform, and the delivery range of the target information is delimited; wherein, the marketing mobile terminal is the core carrier of the information interaction between the enterprise and the user, and can send video, text, picture and other information to the target user to achieve the information promotion goal; the running data set refers to the full mobile terminal running data after desensitization collected by the delivery platform through the standardized data interface, which is used to provide multi-dimensional data support for the generation of the terminal label, and provide data benchmark for the subsequent terminal screening and delivery type matching; the terminal label refers to the terminal feature symbol with identification and time limit weight combined with the real-time running state of the marketing mobile terminal, which is generated by associating the target group feature system with the running data set of the marketing mobile terminal, and is used for cross-source data cross-validation combined with feature weight matching to associate and map the potential audience group characteristics and the terminal label.

[0067] In the embodiment, by locking the potential audience group characteristics, the problem of traditional audience positioning ambiguity is avoided; a unified standard is provided for terminal matching to reduce matching deviation; user privacy is protected when collecting data, and comprehensive and reliable data support is provided for subsequent work to solve the limitations and risks of traditional data collection; the generated terminal label can be dynamically adjusted with the real-time state of the terminal to break the adaptation deficiency of traditional static label; the matching of audience characteristics and terminal label is more accurate and reliable to provide a scientific basis for delimiting the delivery range; the delivery strategy is deeply matched with the terminal characteristics and audience demand to avoid resource waste caused by one-size-fits-all delivery, and the accuracy, safety and efficiency of delivery are comprehensively improved.

[0068] Please refer to Figure 2 , preferably, the step of delimiting the delivery range of the target information comprises:

[0069] The multi-source data fusion technology is adopted to integrate user registration information, third-party platform behavior data, and scenario interaction data, and a feature extraction algorithm combining deep learning and statistical analysis is used to mine and locate potential audience group features of target users of target information to be pushed from the integrated user data, wherein the potential audience group features specifically include user basic attributes, consumption behavior trajectories, and scene demand preferences; and based on a positioning of a value and a layering of a demand of the target information, a feature priority is assigned to each potential audience group feature, and for example, a demand target of the target information is disassembled into quantifiable demand indexes through disassembly of the demand target, importance of different demand indexes is sorted, and then according to a correlation degree between the demand indexes and the potential audience group features, such as a higher correlation degree between a brand exposure demand and the user basic attributes, a higher correlation degree between a user conversion demand and the consumption behavior trajectories, and a higher correlation degree between a scene penetration demand and the scene demand preferences, an analytic hierarchy process is used to assign a corresponding priority weight to each potential audience group feature, and finally a potential audience group feature set with a clear priority is formed, providing a precise feature sorting basis for subsequent target group feature system construction and terminal matching.

[0070] Based on the located potential audience group feature set, a target group feature system is constructed through structured modeling, and the target group feature system includes a feature coding layer, a feature correlation layer, and a feature adaptation layer; wherein the feature coding layer standardizes coding of various potential audience group features for feature quantization, the feature correlation layer establishes a correlation relationship and a correlation coefficient between different potential audience group features through a correlation rule, such as a correlation mode between the user basic attributes and the consumption behavior trajectories, and the feature adaptation layer reserves an adaptation interface for a marketing mobile terminal, for realizing matching and docking of the target group feature system and a marketing mobile terminal running data set and a terminal label.

[0071] Relying on a multi-source data access gateway deployed by a pushing platform, through a standardized data interface compatible with data protocols of different terminal manufacturers, running data of all marketing mobile terminals is collected, including: terminal system version, hardware configuration parameters, network environment attributes, historical application usage records, active time period distribution, and data transmission link features. In the collection stage, data collection is realized without awareness through a terminal-side lightweight burying point module; in the desensitization stage, a dynamic desensitization algorithm is adopted to select a corresponding desensitization strategy according to a data sensitivity level; and in the integration stage, a data cleaning engine is used to remove redundant data and repair missing data, and then a standardized engine is used to convert different format data into a unified data structure, forming a regular running data set of all marketing mobile terminals.

[0072] The constructed target population characteristic system data is subjected to format normalization processing, and quantified characteristic data of the characteristic coding layer, such as numerically coded user age and classification label coded consumption scene, and the association relationship and association coefficient of the characteristic association layer, such as the association rule weight of the user basic attribute and the consumption behavior, are extracted, and are arranged into a standardized target population characteristic matrix according to the structure of the characteristic name, the characteristic value and the association coefficient;

[0073] Meanwhile, the marketing mobile terminal running data set subjected to the standardization processing is called, and terminal data dimensions associated with the potential audience group characteristics, such as the terminal hardware configuration corresponding to the user consumption ability characteristics and the terminal active period corresponding to the user scene demand characteristics, are screened to form a terminal characteristic matrix. The characteristic dimensions of the target population characteristic matrix and the terminal characteristic matrix are matched, such as deleting irrelevant terminal redundant characteristics, supplementing the missing terminal corresponding dimensions in the population characteristics, and unifying the data format, and finally a target population characteristic matrix and a terminal characteristic matrix which can be directly used for mapping are generated.

[0074] Based on the delivery appeal and value positioning of the target information, the priority of the potential audience group characteristics is converted into a quantifiable weight system, an analytic hierarchy process is used to construct a weight evaluation model, the characteristic dimensions in the target population characteristic matrix and the corresponding dimensions in the terminal characteristic matrix are used as evaluation indexes, a judgment matrix is constructed by pairwise comparison, and the basic weight values of each characteristic dimension of the target population characteristic matrix and the terminal characteristic matrix are calculated, for example, if the user conversion is the core of the delivery appeal, the consumption behavior trajectory dimension weight is higher than the user basic attribute dimension.

[0075] On the basis of having calculated the basic weight values of each dimension of the target population characteristic matrix and the terminal characteristic matrix, according to the dynamic nature of the terminal characteristic dimensions and the association degree of the delivery scene adaptation requirements, the dynamic characteristic dimensions in the terminal characteristic matrix are demarcated, such as the network environment attribute, the data transmission link characteristic, the active period distribution, etc., through a lightweight data acquisition interface, the real-time data of the dynamic characteristics of the marketing mobile terminal is synchronized according to a preset period, such as the network stability parameter, the data transmission rate, the current active state, etc., and based on the fluctuation amplitude of the dynamic characteristics, the fluctuation quantized value is converted into a real-time weight adjustment factor according to the weight adjustment coefficient mapping rule, the real-time weight adjustment factor is weighted and fused with the basic weight values of each dimension, a comprehensive weight system is generated, the dynamic adaptation of the weight system to the real-time state of the terminal is realized, and a more accurate weight basis is provided for the subsequent feature mapping matching.

[0076] The generated target population feature matrix and the terminal feature matrix are taken as double input sources, introduced into a comprehensive weight system, and a similarity calculation threshold is set according to a delivery appeal; for the corresponding feature dimensions in the target population feature matrix and the terminal feature matrix, the feature values are multiplied by the corresponding dimension comprehensive weight to obtain weighted feature values; the cosine of the angle between a single marketing mobile terminal feature vector and a target population feature system vector is calculated by a cosine similarity formula, and taken as a correlation matching score of the single marketing mobile terminal and the target population. For features with multi-dimensional cross correlation, such as user scene demand preference for a commuting scene, the active time period distribution of the terminal during the morning rush hour, the network environment attribute of the mobile network and the historical application usage record of the navigation APP need to be synchronously associated. After the multi-dimensional cross correlation feature values are normalized, they are weighted and summed according to the comprehensive weight proportion, and the comprehensive matching score of the cross correlation feature is output, so as to ensure the matching accuracy in a complex feature correlation scene.

[0077] A terminal hardware information extraction interface is called to collect hardware identification data of each marketing mobile terminal, such as terminal device serial number, chip identification code, MAC address, etc. The hardware identification data is encrypted to generate a hardware encrypted string, avoiding leakage of original hardware identification information. At the same time, dynamic parameters in real-time running data of the marketing mobile terminal are extracted, such as the current network IP suffix, system startup timestamp, real-time monitoring module collection period number, etc. to generate a dynamic verification factor. The dynamic verification factor is spliced with the hardware encrypted string and encrypted again to form a terminal unique identifier, effectively avoiding the label conflict problem caused by repeated or fixed hardware identification. Based on the correlation matching score and the terminal unique identifier, the basic information of the marketing mobile terminal is fused to construct a terminal label and perform information structure packaging, including system version information, hardware configuration parameters, terminal initial adaptation network type, terminal device factory number, etc.

[0078] The correlation matching score in the terminal label of the marketing mobile terminal is taken as a core screening index, and the correlation matching score threshold is set according to the delivery appeal. After the full marketing mobile terminal is preliminarily screened, a basic range is set based on the technical adaptation standard of the marketing target information and the target population basic attribute portrait, which is used to clearly define the core boundary of terminal screening. Based on the basic information in the terminal label, the target marketing mobile terminal is screened to define the delivery range of the target information.

[0079] Step S2: based on the target information delivery range, through the construction of a standardized user behavior information collection system, collect user behavior data of the marketing mobile terminal for the target information, carry out information category stratification, and finally build a multi-dimensional target information database, providing data support for subsequent delivery strategy optimization, user demand insight and delivery effect iteration; wherein the user behavior information refers to various behavior data generated in the interaction process between the user in the delivery range and the target information through the marketing mobile terminal, covering information reception, interaction operation, conversion completion, feedback submission and other full-link behavior records; the multi-dimensional target information database refers to a structured data storage and management system integrated with user behavior information, terminal tag data and target information basic data, constructed according to a standardized structure, with the ability of real-time data updating, multi-dimensional retrieval and correlation analysis, solving the problem of fragmented user behavior data in traditional marketing, which is difficult to reuse, realizing the precise linkage of behavior data and delivery strategy, and improving the scientificity and timeliness of delivery decision.

[0080] Please refer to Figure 3 , preferably, the step of building a multi-dimensional target information database comprises:

[0081] Based on the delivery link of the target information, set the user behavior data collection dimension, and collect user behavior data in each marketing mobile terminal through lightweight terminal embedding; the user behavior data collection dimension includes information reception behavior dimension, interaction operation behavior dimension, conversion completion behavior dimension and feedback submission behavior dimension; the information reception behavior dimension refers to the association of the basic running state of the marketing mobile terminal, collects the relevant user behavior data generated in the process of the user passively or actively receiving the target information after the target information reaches the user terminal, forms the collection dimension, and reflects the reach effect and reception scene adaptability of the target information, including target information reception state, reception time, reception time period, reception terminal network environment, and reception terminal system state; the interaction operation behavior dimension refers to the interaction feedback mechanism between the marketing mobile terminal and the target information, collects the user behavior data generated in the process of the user actively operating and intervening the target information, forms the collection dimension, and reflects the interest and willingness of the user to the target information, including click operation, slide operation, stay operation, share operation, and collection operation; the conversion completion behavior dimension refers to the combination of the delivery core target of the target information, collects the user behavior data generated in the process of the user completing the key action guided by the target information, forms the collection dimension, and reflects the achievement effect of the delivery target and the user conversion willingness, including core conversion action completion state, conversion action completion time, conversion path node, conversion associated information, and post-conversion secondary behavior; the feedback submission behavior dimension refers to the collection of user behavior data generated by the user actively feeding back to the target information, forms the collection dimension, and reflects the subjective satisfaction and optimization direction of the user to the target information, including feedback type, feedback content, feedback submission time, feedback associated scene, and feedback processing state.

[0082] The collected user behavior data undergoes preprocessing, including data cleaning, missing data handling, data anonymization, and data format normalization, to generate a standardized user behavior dataset. Data cleaning involves cleaning redundant data based on the terminal's unique identifier, the timestamp of the behavior occurrence, and the dimensions of user behavior data collection, identifying and eliminating duplicate user behavior data, and filtering invalid data. Missing data handling involves constructing a behavior sequence prediction model, using historical patterns of similar behaviors on the same terminal and behavioral characteristics of similar terminals as training samples to complete missing fields. Data anonymization involves using dynamic anonymization algorithms, referring to data sensitivity level classification standards, to perform field replacement and encryption on highly sensitive data, and partial information masking on low-sensitivity data to ensure data compliance. Data format normalization involves using a standardization engine to convert heterogeneous behavior data output from different terminal manufacturers into a unified data structure, providing a high-quality data foundation for subsequent category classification.

[0083] For each marketing mobile terminal's standardized user behavior dataset, extract its behavioral value features under each collection dimension. For example, in the information reception dimension, extract the terminal's reception duration and reception scenario adaptability; in the interaction operation dimension, extract the terminal's active operation frequency and high-value operation ratio; in the conversion completion dimension, extract the terminal's core action completion rate and conversion path completeness; and in the feedback submission dimension, extract the terminal's positive feedback ratio and effective feedback quantity.

[0084] Set quantitative thresholds for behavioral value characteristics of each collection dimension, quantify and score the behavioral value characteristics, set weights for each collection dimension according to the campaign objectives, and calculate the total terminal value score. Based on the total terminal value score and the quantitative scores of each collection dimension, classify each marketing mobile terminal into information categories. Information categories include core value behavior terminals, important auxiliary behavior terminals, and general related behavior terminals. For example, marketing mobile terminals with a total terminal value score greater than 80 points and meeting any of the following core dimension conditions: a conversion completion dimension quantitative score greater than 85 points or an interaction operation dimension quantitative score greater than 90 points and an information reception dimension score greater than 85 points are classified as core value behavior terminals.

[0085] The multi-dimensional target information database is constructed by adopting a layered storage combined with an associated index architecture, and stores a standardized user behavior data set of a marketing mobile terminal in a delivery range, a terminal label, and a belonging information category. A distributed file system is selected for an original data layer, and a standardized user behavior data set of each marketing mobile terminal in the delivery range, a terminal label, and a double-dimensional index are stored according to collection time and a terminal unique identifier; a relational database is adopted for a cleaning and classification layer, and basic information of a terminal label and an information category belonging to each marketing mobile terminal in the delivery range is integrated, cross-data association of the standardized user behavior data set, the terminal label, and the information category is realized through the terminal unique identifier, and a multi-table associated structure is constructed; a time series database is used for a feature application layer, and a behavior value feature reflecting a behavior value in the standardized user behavior data set of each marketing mobile terminal in the delivery range and an information category feature of a terminal belonging to the marketing mobile terminal are used as a basis.

[0086] Step S3: Based on the constructed multi-dimensional target information database, a differentiated delivery response strategy is executed for three types of terminals, i.e., a core value behavior category, an important auxiliary behavior category, and a general associated behavior category, in the delivery range, and the multi-dimensional target information database data is updated synchronously, a user behavior baseline under each information category is dynamically constructed, abnormal terminals are identified and removed through user behavior baseline comparison and safety checking, a remaining marketing mobile terminal information category transfer path map is generated, and dynamic data support is provided for subsequent delivery strategy iteration and resource precise allocation; wherein the user behavior baseline refers to a behavior rule standard with category commonality extracted based on real-time delivery behavior data and current behavior value feature data of all terminals under a corresponding category for each information category, including the core value behavior category, the important auxiliary behavior category, and the general associated behavior category; the abnormal terminal refers to a marketing mobile terminal with behavior data deviating from a baseline of a belonging category or with a running safety hidden danger or a transmission safety risk; and the terminal information category transfer path refers to a track of a terminal transferring from an original information category to other categories due to behavior feature changes in a target information continuous delivery process, reflecting a behavior preference evolution rule of a terminal user for the target information.

[0087] In the embodiment, by executing a differentiated response strategy for different information category terminals, resource waste caused by traditional one-size-fits-all delivery is avoided, and conversion efficiency of core value terminals and activation probability of general associated terminals are improved; a category behavior baseline is dynamically constructed, and a problem that a traditional static baseline is difficult to adapt to terminal behavior changes is solved; abnormal terminals are identified in combination with baseline and safety data double checking, and a delivery environment safety is ensured; a terminal information category transfer path is generated, dynamic tracking of terminal behavior evolution is realized, a basis for precisely adjusting a delivery strategy is provided, and dynamic adaptability and effect controllability of delivery are comprehensively improved.

[0088] Please refer to Figure 4 , preferably, the step of generating the differentiated delivery response strategy comprises:

[0089] The behavior value characteristics of each marketing mobile terminal under each information category are obtained, a clustering algorithm is used to cluster the behavior value characteristics of the marketing mobile terminal of each information category, the feature clustering cluster of each information category is obtained, and the feature threshold range of each behavior value characteristic is counted; then, the contribution of each behavior value characteristic in the feature clustering cluster to the category attribute is calculated by using a random forest feature importance algorithm, and the behavior value characteristics in the clustering cluster are filtered as the typical behavior value characteristics of the information category according to the contribution, such as the typical characteristics of the core value category terminal, which are high core action completion rate and high frequency of secondary behavior after conversion, to form a standardized feature table of information category, typical behavior value characteristics and feature threshold range.

[0090] A behavior value characteristic and delivery parameter mapping table is established to preset the delivery parameter range corresponding to different behavior value characteristics, the delivery parameters include content, frequency and interaction scenario, the delivery parameter type corresponding to each typical behavior value characteristic and the preset delivery parameter range are extracted in combination with the mapping table to form a preliminary delivery list; the typical behavior value characteristics of each information category are called, corresponding delivery parameter values are selected in the preset delivery parameter range according to the feature threshold range in which the typical behavior value characteristics are located, and a delivery list under each information type is constructed.

[0091] Each marketing mobile terminal information category and terminal label in the multi-dimensional target information database are called, the delivery parameters in the delivery list are associated with the marketing mobile terminal, and a differentiated delivery response strategy of each information category is constructed; the differentiated delivery response strategy refers to that, for the marketing mobile terminal of each information category, based on the delivery parameters in the delivery list, in combination with the system version, hardware configuration, network environment, active period and other data in the terminal label, an exclusive delivery operation logic configured for each marketing mobile terminal, including content push type adaptation, push frequency dynamic adjustment, interaction component precise matching, scene trigger condition setting and other dimensions, is used to ensure that the delivery strategy matches the terminal hardware carrying capacity, network adaptability and user behavior habits, to provide a standardized delivery scene for subsequent collection of behavior data reflecting the real value of the terminal, and to provide strategy support for the differentiated presentation of the behavior characteristics of different information category terminals;

[0092] Preferably, the step of generating the behavior baseline of each information category comprises:

[0093] User behavior data of each marketing mobile terminal under the differentiated delivery response strategy is collected, the multi-dimensional target information database is updated, the behavior value characteristics reflecting the behavior value of each collection dimension are extracted, the terminal value total score is calculated, and each marketing mobile terminal is divided into an information category;

[0094] The user behavior data of the original data layer of the multi-dimensional target information database and the environmental data in the terminal label are called to extract scene characteristics, including information touch mode, user interaction environment and interaction target orientation. For example, the information touch mode is divided into active trigger scene, passive push scene, social communication scene and application internal touch scene. The user interaction environment dimension is combined with terminal positioning data, network environment attributes and system running state to divide indoor stable network scene, outdoor mobile network scene, application foreground interaction scene and application background wake-up scene. The interaction target orientation dimension divides information acquisition scene, service reservation scene, consumption conversion scene and social communication scene.

[0095] The scene characteristics are cross combined to form a subdivided scene under each information category, and a scene classification table including information category, scene characteristic dimension and subdivided scene identifier is generated.

[0096] The multi-dimensional target information database is called, and the behavior value characteristics of the user behavior data of all terminals in the corresponding scene are extracted as screening conditions according to the information category and the subdivided scene identifier, and the terminal label is associated to form a scene exclusive data set.

[0097] For each scene exclusive data set, the quantile analysis method is used to calculate the common range and fluctuation threshold of the behavior value characteristics, form a quantization benchmark in each subdivided scene, and generate a behavior baseline for each subdivided scene. Different types of subdivided scenes are set with different iteration periods. In each iteration, the new user behavior data of the corresponding subdivided scene in the current iteration period is extracted, the behavior baseline is updated, and the baseline of each subdivided scene is synchronized with the actual behavior of the terminal to ensure the adaptation.

[0098] Please refer to Figure 5 , preferably, the steps of generating the remaining marketing mobile terminal information category transfer path atlas include:

[0099] According to the information category and the subdivided scene identifier, determine the baseline standard to which each marketing mobile terminal belongs, real-time data preprocessing of the user behavior data of each marketing mobile terminal, and extraction of the behavior value characteristics corresponding to the baseline; calculate the deviation degree of the real-time behavior value characteristics of the marketing mobile terminal and the corresponding behavior baseline; if the deviation degree exceeds the preset deviation degree threshold, it is marked as a behavior suspected abnormal terminal.

[0100] A monitoring cycle is configured to avoid misjudgments caused by single behavioral fluctuations. By continuously tracking and verifying the stability of terminal behavior anomalies, the accuracy of anomaly determination is ensured. The duration of the monitoring cycle is set based on the fluctuation pattern of terminal behavior characteristics and is synchronized with the cycle of terminal information category reclassification in the multi-dimensional target information database. During the monitoring cycle, user behavior data of terminals with suspected abnormal behavior are continuously collected. After each information type classification, it is determined whether the terminal is still marked as a terminal with suspected abnormal behavior. If the ratio of the number of times a terminal is marked as a terminal with suspected abnormal behavior to the total number of judgments is greater than the preset anomaly ratio threshold during the monitoring cycle, the terminal with suspected abnormal behavior is determined as an abnormal terminal and marked as an abnormal terminal.

[0101] Send a pause command to the abnormal terminal through the delivery platform interface to terminate the push of target information, add an abnormal status tag to the abnormal terminal, record information such as the time of abnormal identification and the time of removal operation, and store it in the abnormal terminal management table.

[0102] Abnormal terminals are removed from the multi-dimensional target information database. For the remaining marketing mobile terminals, the initial information category and the current information category at each time point in the target information delivery process are extracted and associated with behavioral value characteristics and behavioral data collection timestamps.

[0103] When the information category of a marketing mobile terminal is transferred, the behavioral value characteristics of the marketing mobile terminal before and after the transfer are called, and the change range of each behavioral value characteristic is calculated to identify the key characteristics driving the information category transfer.

[0104] A terminal information category transfer path model is constructed using a directed graph with time attributes. Three information categories—core value behavior, important auxiliary behavior, and general association—are used as nodes in the directed graph. The node attributes include category identifier and collection timestamp. The transfer relationship from the initial information category to the current information category of the marketing mobile terminal is used as the directed edge. The edge attributes include transfer start time, transfer end time, and transfer driving feature label.

[0105] For each remaining marketing mobile terminal, the trajectory of its information category change is connected sequentially according to time nodes to form a single terminal transfer path; through the path aggregation algorithm, transfer paths with the same transfer direction are merged, and the terminal proportion, average transfer duration, and transfer driving feature distribution of each type of transfer path are statistically analyzed to generate a transfer path map of information categories for remaining marketing mobile terminals.

[0106] Step S4: based on the remaining terminal information category transfer path map, the differentiated delivery response strategy effect evaluation is performed, the strategy execution abnormality is identified, the preset stop standard is combined to judge whether to terminate the target information delivery, and the efficiency of the delivery is ensured; wherein the strategy effect evaluation refers to the quantitative index analysis of the driving effect of the differentiated delivery response strategy on the terminal information category transfer and the target achievement degree; the strategy abnormality refers to the situation that the terminal transfer path after the strategy execution does not meet the expectation, resulting in the substandard delivery effect; the stop standard refers to the delivery termination judgment basis set based on the target achievement rate, terminal value distribution, resource input-output ratio, etc.

[0107] Preferably, the step of identifying the strategy execution abnormality and combining the preset stop standard to judge whether to terminate the target information delivery comprises:

[0108] Based on the marketing mobile terminal information category transfer path map, after the execution of the differentiated delivery response strategy under each information category, the terminal proportion in the expected transfer direction, the terminal proportion meeting the preset time length in the transfer time length, and the quantitative transfer driving index are obtained.

[0109] The terminal value index is quantified by the ratio of the increment of the number of core value terminals after the delivery to the number of core value terminals before the delivery, the improvement ratio of the average terminal value total score after the delivery to the average terminal value total score before the delivery, and the proportion of the terminal whose core conversion action completion rate is improved in the terminal covered by the delivery range.

[0110] The transfer driving index and the terminal value index are standardized, and the comprehensive effect score of the differentiated delivery response strategy under each information category is calculated by weighted summation. The comprehensive effect score threshold and the qualified threshold of each dimension index are set to determine whether the differentiated delivery response strategy is abnormal. If there is an abnormality, the abnormality reason is located through data backtracking. For example, the differentiated delivery response strategy execution log and the marketing mobile terminal user behavior data are called through the data backtracking algorithm to analyze whether the unreasonable delivery parameter setting or the execution process deviation is caused. The influence degree of the adaptation deviation on the effect is confirmed by comparing the strategy effects of different groups according to the hardware network grouping through terminal stratified sampling experiment. If the terminal proportion of the expected transfer path is lower than that of the unexpected path, and the terminal behavior value characteristics of the unexpected path do not meet the target category standard, the association rule mining algorithm is used to analyze the association between the terminal behavior characteristics of the unexpected path and the strategy parameters to determine whether the strategy guidance logic is missing.

[0111] A delivery stop standard system is configured, which is composed of core target standard threshold and terminal value stability standard threshold, and is configured based on the delivery appeal and terminal behavior value evolution law to determine whether the target information delivery is necessary to continue, avoid resource waste caused by invalid delivery, and ensure efficient achievement of the delivery target and stable deposition of the terminal value.

[0112] Based on the residual marketing mobile terminal information category transfer path map, combined with the multi-dimensional target information database, the required quantitative indicators for the stop standard are calculated, including the core target standard threshold corresponding to the core value class terminal proportion, the terminal average core conversion action completion rate; the terminal value stability standard and the corresponding continuous two information category redivision period within the core value class, important auxiliary class, general correlation class terminal proportion fluctuation range, the average value fluctuation range of the terminal typical behavior value characteristics of each information category;

[0113] In turn, the stop standard quantitative indicators and the launch stop standard system are checked, if the core value class terminal proportion and the terminal average core conversion action completion rate do not meet the corresponding standard requirements, it is determined that the target information needs to continue to be launched, and at the same time the optimization of the previously positioned strategy abnormal root cause is started; if the core target meets the standard but the core value class, important auxiliary class, general correlation class terminal proportion fluctuation range exceeds the standard within the continuous two information category redivision period, or the average value fluctuation range of the terminal typical behavior value characteristics of each information category does not meet the requirements, it is determined that the target information needs to continue to be launched to stabilize the terminal value, maintain the core parameters of the current differentiated launch response strategy, and only fine-tune the non-core parameters such as content format and interaction components according to the terminal adaptability; if both standards are met, it is determined to terminate the target information launch, and a launch stop instruction is sent to all marketing mobile terminals within the launch range through the launch platform interface, the target information push is terminated, and the launch closed loop is completed.

[0114] Please refer to Figure 5 The embodiment introduces a marketing mobile terminal security monitoring and dynamic response system, which includes a data construction module, a security monitoring module, an effect evaluation module and a launch optimization module.

[0115] The data construction module is used to generate terminal labels to define the launch range and construct a multi-dimensional target information database to provide data support for target information launch; the security monitoring module is used to identify abnormal terminals by dynamically constructing user behavior baselines of each information category and sub-scene; the effect evaluation module is based on the terminal information category transfer path map to execute differentiated launch response strategy effect evaluation and identify strategy execution abnormalities; the launch optimization module is used to determine whether to terminate the target information launch in combination with the preset stop standard.

[0116] The data construction module adopts multi-source data fusion and feature extraction algorithm to mine potential audience group characteristics, and constructs a target population characteristic system containing a feature coding layer, a correlation layer and an adaptation layer; through a standardized interface, full-quantity marketing mobile terminal operation data is collected, and after desensitization, cleaning and standardization processing, an operation data set is formed; based on the analytic hierarchy process and dynamic weight adjustment, a comprehensive weight system is constructed, the matching score of the terminal and the population is calculated combined with the cosine similarity, the terminal label with real-time weight is generated and the delivery range is delimited; a full-link user behavior collection system is constructed, terminal behavior data in the delivery range is collected and preprocessed, behavior value characteristics are extracted to divide terminal information categories; a multi-dimensional target information database is constructed using a hierarchical storage architecture, and standardized behavior data, terminal labels and terminal categories are integrated to provide data support for subsequent delivery execution.

[0117] The safety monitoring module constructs a mapping relationship between behavior value characteristics and delivery parameters based on terminal category and label data in the multi-dimensional target information database, and configures exclusive differentiated delivery response strategies for three types of terminals; terminal behavior data under the execution of differentiated delivery response strategies is collected to update the multi-dimensional target information database, typical behavior characteristics of categories are screened through clustering algorithm and random forest algorithm, and subdivided scene behavior baselines are generated combined with scene characteristics cross combination and iteratively updated according to differentiated periods; terminal behavior characteristics are compared with corresponding baselines in real time, deviation degrees are calculated to mark suspected abnormal terminals, abnormal terminals are determined combined with the proportion of abnormalities in the monitoring period, and after confirming the abnormality through safety verification, the delivery of abnormal terminals is suspended and the abnormal terminal data is removed from the database to ensure the safety of the delivery environment.

[0118] The effect evaluation module extracts the initial category, the current category at each time node and the behavior value characteristics of the remaining normal terminals, calculates the feature change amplitude when the category shifts to identify the driving factors; a directed graph with time attributes is used to construct a shift path model, taking three types of terminals as nodes and shift relationship as directed edges, linking single-terminal category change trajectories to form links, and generating a shift path atlas containing terminal proportion, average duration and driving characteristics through aggregation algorithm; based on the atlas, shift driving indicators and terminal value indicators are quantified, and after standardized processing, the comprehensive effect score of the strategy is calculated by weighted calculation, whether the strategy has abnormal effects such as unmet standards, adaptation deviation and path deviation is determined combined with threshold values, and the abnormal root cause is located through data backtracking, sampling experiment and correlation mining.

[0119] The delivery optimization module constructs a delivery stop standard system based on the core target standard and the stability of terminal value based on the delivery appeal and the terminal value law, calculates the core target indicators and the stability of terminal value; according to the priority check standard, if the core target is not met, the abnormal root cause optimization strategy is combined, if the target is met but the value is not stable, the adaptive parameters are fine-tuned, and if both standards are met, the delivery is terminated and a report is generated, realizing the efficiency of delivery and the rationality of resource utilization.

[0120] Working principle and its effect:

[0121] The present application relies on data construction, safety monitoring, effect evaluation, and linkage of optimization to realize precise, safe, and dynamic management of marketing mobile terminal delivery, solve the pain points of traditional delivery audience positioning ambiguity, strategy adaptation deficiency, abnormal difficulty control, and effect difficulty adjustment, and improve delivery efficiency and safety.

[0122] By integrating user multi-dimensional data to construct a target population feature system, collecting and standardizing terminal operation data, combining dynamic weight to generate terminal labels to define delivery range, and collecting user full-link behavior data, dividing terminal information categories, and constructing a multi-dimensional target information database, the process solves the problem of traditional audience positioning ambiguity, improves matching accuracy, breaks data silos, and avoids resource waste. Different categories of terminals are configured with differentiated delivery strategies, and a segmented scene behavior baseline is dynamically constructed. By comparing behavior and baseline deviation and continuously monitoring and verifying, abnormal terminals are accurately identified and removed. This mechanism solves the problem of insufficient one-size-fits-all strategy adaptation, improves conversion and activation effect, and avoids the risk of data distortion and resource abuse caused by abnormal terminals. A terminal information category transfer path map is constructed, strategy effect is quantified, and abnormal root causes are located. Combined with stop standard verification of delivery status, the strategy is dynamically optimized or terminated. This link solves the limitations of traditional effect evaluation and reliance on manual adjustment, realizes full-link tracking of terminal behavior and accurate analysis of strategies, and avoids resource redundancy.

[0123] In summary, the present application uses data and dynamic labels to consolidate the foundation of precise delivery, uses differentiated strategies and baseline monitoring to ensure process safety, uses quantitative evaluation and layered standards to optimize the delivery closed loop, and ultimately realizes efficient use of delivery resources, precise reach of terminals, and controllable delivery safety, providing a reliable solution for marketing mobile terminal management.

[0124] The above is only the preferred embodiment of the present application, and the protection scope of the present application is not limited to the above-mentioned embodiments. Any technical solution falling within the scope of the present application is within the protection scope of the present application. It should be noted that for ordinary technical personnel in the technical field, some improvements and refinements without departing from the principles of the present application are also considered within the protection scope of the present application.

Claims

1. A method for marketing mobile terminal security monitoring and dynamic response, characterized by, The method comprises the following steps: Positioning the characteristics of the potential audience group of the target information to be delivered, constructing a target group feature system, matching with the terminal tags of the full marketing mobile terminals of the delivery platform, and delimiting the delivery range of the target information; Within the delivery range, collecting user behavior data of each marketing mobile terminal for the target information and extracting behavior value features, classifying each marketing mobile terminal by information category, and constructing a multi-dimensional target information database; Clustering and filtering typical behavior value features of each marketing mobile terminal under each information category to match the delivery parameters of the target information under each information category, generating a differentiated delivery response strategy, and delivering the target information; Updating the multi-dimensional target information database; Dividing the sub-scenarios under each information category, constructing a scenario-specific dataset, calculating the common range and fluctuation threshold of the behavior value features, and generating the behavior baseline of each information category; Based on the behavior baseline, identifying and removing abnormal terminals, constructing the transfer path of the marketing mobile terminal information category based on a directed graph, generating a remaining marketing mobile terminal information category transfer path atlas through path aggregation, and evaluating the effect of the differentiated delivery response strategy, identifying strategy execution abnormalities, and determining whether to terminate the delivery of the target information according to the preset stop standard; The step of generating a differentiated delivery response strategy comprises: Clustering the behavior value features of the marketing mobile terminals of each information category to obtain feature clustering clusters of each information category, and counting the feature threshold range of each behavior value feature; Calculate the contribution of each behavior value feature in the feature clustering cluster to the category attribute, and filter the behavior value features in the clustering cluster as typical behavior value features according to the contribution; Establish a mapping table of behavior value features and delivery parameters, preset the delivery parameter range corresponding to the value features, extract the delivery parameter type and delivery parameter range corresponding to each typical behavior value feature, and select the corresponding delivery parameter value within the preset delivery parameter range according to the feature threshold range of the typical behavior value feature, and construct the delivery list under each information type; Associate the delivery parameters in the delivery list with the marketing mobile terminals to construct the differentiated delivery response strategy of each information category; The step of identifying strategy execution abnormalities and determining whether to terminate the delivery of the target information according to the preset stop standard comprises: Based on the marketing mobile terminal information category transfer path atlas, after the execution of the differentiated delivery response strategy under each information category, the terminal proportion in the expected transfer direction of the differentiated delivery response strategy and the terminal proportion that meets the preset time length in the transfer time length are quantified to obtain the transfer driving index; The ratio of the number of core value terminals after the delivery of the differentiated delivery response strategy to the number before the delivery, the improvement ratio of the average value total score of the terminals after the delivery to the value total score before the delivery, and the proportion of the terminals whose core conversion action completion rate is improved to the delivery range coverage terminals are quantified to obtain the terminal value index. The transfer driving indicators and the terminal value indicators are standardized, and a comprehensive effect score of the differentiated delivery response strategy under each information category is calculated by weighted summation. A comprehensive effect score threshold and a qualified threshold of each dimension indicator are set to determine whether the differentiated delivery response strategy is abnormal. A delivery stop standard system is configured, which is composed of core target compliance standard thresholds and terminal value stability standard thresholds. Quantitative indicators required for the stop standard are calculated, and a stop check is performed to determine whether to terminate the target information delivery.

2. The marketing mobile terminal security monitoring and dynamic response method of claim 1, wherein, The step of demarcating the delivery range of the target information comprises: locating potential audience group characteristics of target users of the target information to be delivered, and assigning a characteristic priority to each potential audience group characteristic to construct a potential audience group characteristic set; based on the potential audience group characteristic set, a target population characteristic system is constructed, which includes a characteristic coding layer, a characteristic correlation layer, and a characteristic adaptation layer; the characteristic coding layer is used for standardizing coding of the potential audience group characteristics; the characteristic correlation layer is used for establishing correlation relationships and correlation coefficients between different potential audience group characteristics; and the characteristic adaptation layer is used for matching and docking the target population characteristic system with the terminal labels of the marketing mobile terminals; running data of all marketing mobile terminals of the delivery platform are collected to construct a running data set of all marketing mobile terminals, and the target population characteristic system is subjected to format normalization processing to extract quantitative characteristic data of the characteristic coding layer and correlation relationships and correlation coefficients of the characteristic correlation layer, thereby constructing a standardized target population characteristic matrix; the running data set of all marketing mobile terminals is called to filter terminal data dimensions associated with the potential audience group characteristics, form a terminal characteristic matrix, and match the characteristic dimensions of the target population characteristic matrix, which are used to calculate the correlation matching scores of each marketing mobile terminal and the target population, and construct terminal labels; a correlation matching score threshold and a basic range are set, and all marketing mobile terminals are filtered based on the terminal labels of the marketing mobile terminals to locate target marketing mobile terminals and demarcate the delivery range of the target information.

3. The marketing mobile terminal security monitoring and dynamic response method of claim 2, wherein, The step of calculating the correlation matching scores of each marketing mobile terminal and the target population to construct the terminal labels comprises: the characteristic dimensions in the target population characteristic matrix and the corresponding dimensions in the terminal characteristic matrix are taken as evaluation indicators to construct a judgment matrix through pairwise comparison, and the basic weight values of each characteristic dimension of the target population characteristic matrix and the terminal characteristic matrix are calculated; dynamic characteristic dimensions in the terminal characteristic matrix are demarcated, real-time data of dynamic characteristics of the marketing mobile terminals are synchronized according to a preset period, and a real-time weight adjustment factor is generated based on the fluctuation amplitude of the dynamic characteristics. The real-time weight adjustment factor is weighted and fused with the basic weight values of each dimension to generate a comprehensive weight system; for the characteristic dimensions corresponding to each other in the target population characteristic matrix and the terminal characteristic matrix, the characteristic values are multiplied by the corresponding dimension comprehensive weights to obtain weighted characteristic values; and the included angle cosine value of a single marketing mobile terminal characteristic vector and a target population characteristic system vector is calculated as the correlation matching score of the single marketing mobile terminal and the target population. The hardware identification data of each marketing mobile terminal is collected, the hardware identification data is encrypted, a hardware encryption string is generated, dynamic parameters in real-time running data of the marketing mobile terminal are extracted, a dynamic check factor is generated, and the hardware encryption string is spliced with the dynamic check factor to form a terminal unique identification; Based on the association matching score and the terminal unique identification, a terminal tag is constructed by fusing the basic information of the marketing mobile terminal and is information-structured and packaged.

4. The marketing mobile terminal security monitoring and dynamic response method of claim 1, wherein, The step of constructing the multi-dimensional target information database comprises: Setting a user behavior data collection dimension, collecting user behavior data in each marketing mobile terminal within the delivery range; The collected user behavior data is pre-processed, including data cleaning, missing data processing, data desensitization, and data format normalization, to generate a standardized user behavior data set; For each marketing mobile terminal, the behavior value characteristics of the standardized user behavior data set in each collection dimension are extracted, the quantization threshold of each collection dimension behavior value characteristic is set, the behavior value characteristics are quantitatively scored, the weight of each collection dimension is set and the terminal value total score is calculated, and based on the terminal value total score and the quantization score of each collection dimension, each marketing mobile terminal is classified by information category; The information category includes core value behavior category terminal, important auxiliary behavior category terminal, and general correlation behavior category terminal; A multi-dimensional target information database is constructed by adopting a hierarchical storage combined with an association index architecture, and the standardized user behavior data set of the marketing mobile terminal within the delivery range, the terminal tag, and the information category to which it belongs are stored.

5. The marketing mobile terminal security monitoring and dynamic response method of claim 4, wherein, The user behavior data collection dimension comprises an information receiving behavior dimension, an interactive operation behavior dimension, a conversion completion behavior dimension, and a feedback submission behavior dimension; The information receiving behavior dimension is used to associate the basic running state of the marketing mobile terminal and collect user behavior data generated in the process of user passive or active reception of target information; The interactive operation behavior dimension is used to collect user behavior data generated in the process of user active operation intervention on target information; The conversion completion behavior dimension is used to collect user behavior data generated in the process of user completion of target information guidance; The feedback submission behavior dimension is used to collect user behavior data generated by user active feedback on target information.

6. The marketing mobile terminal security monitoring and dynamic response method of claim 1, wherein, The step of generating the behavior baseline of each information category comprises: Collecting user behavior data of each marketing mobile terminal under a differentiated delivery response strategy, updating the multi-dimensional target information database, extracting behavior value characteristics reflecting the behavior value in each collection dimension, calculating the terminal value total score, and classifying each marketing mobile terminal by information category; The user behavior data and environmental data in the terminal tag in the original data layer of the multi-dimensional target information database are called, scene characteristics are extracted, and the scene characteristics are cross-combined to form a subdivided scene under each information category; The multi-dimensional target information database is called, the information category and the subdivided scene are identified as screening conditions, the behavior value characteristics of the user behavior data of all terminals in the corresponding scene are extracted, and the terminal tag is associated to form a scene-specific data set; For each scene-specific data set, the quantile analysis method is used to calculate the common range and fluctuation threshold of the behavior value characteristics, form the quantitative benchmark in each sub-scene, generate the behavior baseline of each sub-scene, and set a differentiated iteration period to update the behavior baseline.

7. The marketing mobile terminal security monitoring and dynamic response method of claim 1, wherein, The step of generating the residual marketing mobile terminal information category transition path atlas comprises: Calculate the deviation degree of the real-time behavior value characteristics of the marketing mobile terminal and the corresponding behavior baseline, and if the deviation degree exceeds the preset deviation degree threshold, mark it as a behavior suspected abnormal terminal; Configure a monitoring period, continuously collect user behavior data of the behavior suspected abnormal terminal within the monitoring period, and if the ratio of the number of times marked as a behavior suspected abnormal terminal to the total number of judgment times is greater than the preset abnormal proportion threshold, the behavior suspected abnormal terminal is determined as an abnormal terminal; Remove the abnormal terminal from the multi-dimensional target information database, and for the remaining marketing mobile terminal, extract its initial information category, current information category at each time node, and associate the behavior value characteristics and behavior data collection time stamp during the target information delivery process; When the information category of the marketing mobile terminal is transferred, the behavior value characteristics before and after the transfer are called, the change amplitude of each behavior value characteristic is calculated, and the key features driving the information category transfer are identified; The information category is the node of the directed graph, and the node attribute includes the category identifier and the collection time stamp; the transfer relationship of the marketing mobile terminal from the initial information category to the current information category is taken as the directed edge, and the attribute of the edge includes the transfer start time, the transfer end time and the transfer driving feature label; Serially connect the information category change track of the marketing mobile terminal to form the transfer path of a single terminal; and combine the transfer paths of the same transfer direction to generate the information category transition path atlas of the remaining marketing mobile terminal.

8. A marketing mobile terminal security monitoring and dynamic response system for implementing the marketing mobile terminal security monitoring and dynamic response method according to any one of claims 1 to 7, characterized by, It comprises a data construction module, a safety monitoring module, an effect evaluation module and a delivery optimization module. The data construction module is used to generate terminal label to define the delivery range and construct a multi-dimensional target information database to provide data support for target information delivery; the safety monitoring module is used to identify abnormal terminals by dynamically constructing user behavior baselines of each information category and sub-scene; the effect evaluation module is used to evaluate the effect of the differentiated delivery response strategy based on the terminal information category transition path atlas and identify strategy execution abnormalities; The delivery optimization module is used to determine whether to terminate the target information delivery in combination with the preset stop standard.

Images