A digital marketing platform human-computer interaction method and system

Abstract

This invention discloses a human-computer interaction method and system for a digital marketing platform, relating to the field of big data analysis. The method includes: real-time collection of user behavior, environmental, and scenario data; analysis of semantics and sentiment using natural language processing; construction of dynamically updated multimodal user profiles and demand scenario profiles based on historical data; generation of compliant scenario-based interaction strategies based on the dual profiles and a strategy library; pushing content and monitoring feedback after real-time verification; dynamically adjusting parameters and switching interaction paths in case of anomalies; establishing an effect evaluation system to quantify strategy effectiveness; continuously optimizing strategy parameters, content, and paths based on data feedback; and retaining complete iteration records for backtracking. The advantages of this invention are: constructing a closed-loop human-computer interaction system with a deep integration of intelligence, dynamism, and compliance as its core; and efficiently improving the accuracy and conversion efficiency of marketing interactions through profile construction, strategy generation, and parameter optimization, adapting to the diverse needs of digital marketing.

Description

Technical Field

[0001] This invention relates to the field of big data analytics, and in particular to a human-computer interaction method and system for a digital marketing platform. Background Technology

[0002] With the rapid development of the digital economy, enterprises are increasingly demanding precise and intelligent marketing, while traditional marketing models face challenges due to inefficiency and insufficient user reach. Digital marketing platforms, by integrating big data, artificial intelligence, and automation technologies, have significantly improved the precision and efficiency of marketing campaigns; however, existing human-computer interaction methods still suffer from pain points such as complex operation and slow response.

[0003] Current digital marketing human-computer interaction methods on the market generally suffer from shortcomings such as fragmented processes, insufficient intelligence, and poor dynamic adaptability, falling far short of a complete closed-loop system. Most methods are stacked functions, with fragmented and delayed data collection. Systems like CDP and MA operate independently, creating data silos. They rely on manual coordination between stages, failing to build dynamically updated multimodal user and scenario profiles, and often rely on static tagging for operations, making it difficult to uncover potential needs. Strategy generation is rigid and lacks adaptive adjustment capabilities, relying heavily on manual configuration of push rules without real-time compliance and adaptability checks. During interaction, they cannot dynamically adjust parameters or handle anomalies based on user feedback, and the interaction methods are limited. Furthermore, performance evaluation is delayed and one-sided, lacking real-time quantitative analysis and iterative optimization mechanisms, making it difficult to backtrack to previous strategy versions. Some methods also suffer from security compliance and business process fragmentation issues, with incomplete privacy anonymization, reducing marketing accuracy and conversion efficiency while impacting user experience and trust. Summary of the Invention

[0004] To improve existing methods and systems, this paper provides a human-computer interaction method and system for digital marketing platforms. This method constructs a closed-loop human-computer interaction system throughout the entire process, with the deep integration of intelligence, dynamism and compliance as its core. Through precise profile construction, adaptive strategy generation, dynamic parameter tuning and optimization, and full-process security and compliance assurance, it effectively improves the accuracy of marketing interactions, conversion efficiency and user trust, and adapts to the diverse needs of digital marketing.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A human-computer interaction method for a digital marketing platform includes: The platform collects user interaction behavior, device environment and scene-related data in real time, analyzes the semantics and sentiment of user input through natural language processing, and combines behavioral analysis to uncover potential needs and build demand scenario profiles. Based on demand scenario profiles, and by integrating historical interaction data and real-time incremental data, a multimodal user profile is constructed. After each interaction is completed, the abnormal profile dynamic data is updated and the label accuracy is corrected in a loop. Based on user profiles and demand scenario profiles, the system calls a pre-set strategy library and combines it with real-time marketing rules and platform resources to generate scenario-based interaction strategies. The system verifies the compliance and adaptability of the strategies through strategy validation, removes unqualified strategies, and regenerates them. Based on the generated strategy, push interactive content and start the session, collect real-time user feedback, dynamically adjust interaction parameters, and automatically switch the interaction form or simplify the path when an interaction abnormality is detected, while providing alternative interaction strategies in parallel. Establish an interaction effect monitoring system to collect multi-dimensional evaluation index data during and after the interaction, identify the strengths and weaknesses of the interaction strategy, and quantify the contribution of the interaction behavior to the marketing conversion goal. Based on the evaluation results, dynamic profiles and changes in the scenario, the strategy is iteratively optimized. For the strategy parts with deficiencies, it is corrected by adjusting parameters, optimizing content and reconstructing paths. The content, basis and effect of each iteration are recorded, and strategy backtracking and version switching are supported. Security verification and privacy protection are performed synchronously throughout the entire process. Multi-factor authentication and data encryption are used to ensure the security of interactions. Sensitive profile data and interaction data are anonymized based on the scope of user authorization, and a privacy protection compliance report is generated.

[0006] Preferably, the platform collects user interaction behavior, device environment, and scene-related data in real time, analyzes user input semantics and sentiment through natural language processing, and combines behavioral analysis to mine potential needs, constructing a demand scenario profile, specifically including: The digital marketing platform collects user interaction behavior data, device environment data, and scenario-related data in real time. The interaction behavior data includes page dwell time, operation path, click frequency, semantics of input content, and interaction interruption nodes. The device environment data includes terminal type, operating system, network status, and geographical location information. The scenario-related data includes the current time dimension, marketing activity cycle, and user group tags. Natural language processing technology is used to segment user input and semantic feedback, identify intent and determine sentiment, and combine behavioral data to obtain users’ potential marketing needs. Based on equipment environment data and scenario-related data, a requirement scenario profile is constructed. The explicit and potential requirements after analysis are prioritized and a requirement analysis report is generated.

[0007] Preferably, the step of constructing a multimodal user profile based on demand scenario profiling, integrating historical interaction data and real-time incremental data, and cyclically updating the abnormal profile dynamic data and correcting the label accuracy after each interaction specifically includes: Based on demand scenario profiles, and by integrating historical marketing interaction data and real-time incremental behavioral data, a multimodal user profile is constructed, which includes basic attribute dimensions, marketing preference dimensions, behavioral habit dimensions, conversion intention dimensions, and interaction tolerance dimensions. Based on the profile iteration mechanism, after each human-computer interaction cycle is completed, the dynamic dimension data in the profile is automatically updated, and the accuracy of the profile tags is corrected in a synchronous manner.

[0008] Preferably, the step of generating scenario-based interaction strategies based on user profiles and demand scenario profiles, by calling a preset strategy library and combining real-time marketing rules and platform resources, verifies the compliance and adaptability of the strategies through strategy verification, eliminates unqualified strategies, and regenerates specific strategies, including: Based on user profiles and demand scenario profiles, the system calls upon a pre-set marketing interaction strategy library and combines real-time marketing activity rules and platform resource compatibility to generate scenario-based interaction strategies. The interaction strategy includes interaction form selection, content presentation scheme, push timing planning and interaction path design. It adapts text, voice, graphics or multimodal interaction forms based on user interaction habits, customizes content and layout according to marketing preferences, plans push timing according to time period and scenario data, and the interaction path design responds to different interaction feedback through preset branch paths. The policy validation process verifies the compliance, resource suitability, and user experience suitability of the generated policies, removes non-compliant policies, and regenerates them.

[0009] Preferably, the step of pushing interactive content and initiating a session based on the generated strategy, collecting real-time user feedback, dynamically adjusting interaction parameters, automatically switching the interaction form or simplifying the path when an interaction anomaly is detected, and simultaneously providing alternative interaction strategies specifically includes: Based on the generated scenario-based interaction strategy, push the appropriate interactive content to the user terminal and start the interactive session; During the interaction process, real-time user interaction feedback data is collected, including operation response speed, content dwell time, interaction action feedback and semantic feedback information. The interaction parameters are dynamically adjusted based on preset adaptation rules, including content presentation rhythm, operation guidance method, feedback response speed and push frequency. When abnormal user interaction behavior is detected, the strategy adjustment mechanism is automatically triggered to switch the interaction form or simplify the interaction path, while pushing alternative interaction strategies.

[0010] Preferably, the establishment of an interaction effect monitoring system, which collects multi-dimensional evaluation index data during and after the interaction, identifies the strengths and weaknesses of the interaction strategy, and quantifies the contribution of the interaction behavior to the marketing conversion goal, specifically includes: Establish an interaction effect monitoring system to collect multi-dimensional evaluation index data in real time during the interaction execution process and after the interaction session ends. The evaluation indexes include interaction completion rate, content click rate, user dwell time, conversion behavior occurrence rate, interaction satisfaction score and negative feedback rate. By comprehensively analyzing the evaluation indicators through data visualization analysis, an interaction effect evaluation report is generated, identifying the strengths and weaknesses of the interaction strategy, and quantifying the contribution of the interaction behavior to the marketing conversion goal.

[0011] Preferably, the iterative optimization of the strategy based on evaluation results, dynamic profiles, and scene changes, and the correction of deficient strategy parts through parameter adjustment, content optimization, and path reconstruction, recording the content, basis, and effect of each iteration, and supporting strategy backtracking and version switching specifically includes: Based on the results of interaction effect evaluation, combined with the dynamic update data of user profiles and real-time changes in marketing scenarios, the strategy is iterated and optimized. Interaction strategy nodes that are rated as excellent are retained, while strategies with deficiencies are corrected through parameter adjustment, content optimization, and path reconstruction to generate optimized interaction strategies. The strategy generation model is trained based on iterative data, and the decision logic and adaptation algorithm of the model are optimized. By recording the content, basis and effect of each iteration, the strategy backtracking and version switching are supported.

[0012] Preferably, the step of synchronously performing security verification and privacy protection throughout the entire process, ensuring interaction security through multi-factor authentication and data encryption, and de-identifying sensitive profile data and interaction data based on the user's authorized scope to generate a privacy protection compliance report specifically includes: Throughout the entire interaction process, interactive security verification and privacy protection operations are performed simultaneously. The security verification includes user identity verification, interactive data transmission encryption, operation permission control and malicious interactive behavior identification. Multi-factor authentication technology is used to ensure the authenticity and validity of user identity, and encrypted transmission protocol is used to prevent interactive data from being stolen or tampered with. The privacy protection operation is based on the user's authorized scope, de-identifies sensitive profile data and interaction data, obtains the boundaries of data collection and use, avoids privacy data calls beyond the authorized scope, and generates a privacy protection compliance report by recording data usage trajectory.

[0013] Furthermore, a human-computer interaction system for a digital marketing platform is proposed, including: Data acquisition module: Collects user interaction behavior, device environment and scene-related data in real time, analyzes semantics and sentiment through natural language processing, and builds dynamic demand scene profiles; User profile building module: It integrates historical and real-time data to generate multimodal user profiles and dynamically updates tag accuracy and abnormal data based on interactive loops; The strategy generation and verification module combines user profiles, scenario profiles, and the strategy library to generate scenario-based interaction strategies. It also performs compliance and adaptability checks to eliminate invalid strategies and optimize the generation logic. Interaction Execution and Optimization Module: Push strategy content and monitor user feedback in real time, dynamically adjust interaction parameters or switch paths, and trigger alternative solutions when anomalies occur; Performance evaluation and analysis module: Collects multi-dimensional indicators, quantifies the advantages and disadvantages of strategies, and generates visual reports to evaluate the contribution to marketing goals; Security and Privacy Management Module: Implements multi-factor authentication, data encryption, and sensitive information anonymization throughout the entire process, and generates compliance reports to ensure secure interaction and privacy protection; Iterative optimization module: Optimizes strategy parameters, content, and paths based on evaluation results, records iteration history, and supports version backtracking and model training; Processor: The processor is used to handle the calculation process of each formula and the construction calculation process of each model.

[0014] Compared with the prior art, the advantages of the present invention are: Achieving a deep integration of intelligence, dynamism, and compliance. Through multi-dimensional data collection and natural language processing technology, it accurately constructs demand scenario profiles and multimodal user profiles. Relying on interactive loops, it dynamically updates profile data and corrects tag accuracy, providing precise support for strategy generation. The strategy generation process combines a strategy library with real-time rules, ensuring the feasibility of the solution through compliance and adaptability verification. During interaction, it can dynamically adjust parameters based on user feedback and handle anomalies, improving interaction smoothness and user experience. Simultaneously, it quantifies marketing contributions through multi-dimensional performance monitoring, continuously improves strategies through an iterative optimization mechanism, and supports version rollback. Coupled with end-to-end security verification and privacy anonymization, it effectively improves the accuracy of marketing interactions, conversion efficiency, and user trust while ensuring data security and compliance, adapting to the diversified needs of digital marketing. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the method proposed in this invention; Figure 2 This is a schematic diagram illustrating the user demand perception and analysis proposed in this invention; Figure 3 This is a schematic diagram illustrating the user profile construction and iteration proposed in this invention; Figure 4 This is a schematic diagram illustrating the scenario-based interaction strategy proposed in this invention. Figure 5 This is a schematic diagram illustrating the multimodal interaction execution and adaptation proposed in this invention; Figure 6 This is a schematic diagram of the interaction effect monitoring and evaluation proposed in this invention; Figure 7 This is a schematic diagram illustrating the iterative optimization and refinement of the interaction strategy proposed in this invention; Figure 8 This is a schematic diagram illustrating the interactive security verification and privacy protection adaptation proposed in this invention. Detailed Implementation

[0016] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0017] A human-computer interaction system for a digital marketing platform, comprising: Data acquisition module: Collects user interaction behavior, device environment and scene-related data in real time, analyzes semantics and sentiment through natural language processing, and builds dynamic demand scene profiles; User profile building module: It integrates historical and real-time data to generate multimodal user profiles and dynamically updates tag accuracy and abnormal data based on interactive loops; The strategy generation and verification module combines user profiles, scenario profiles, and the strategy library to generate scenario-based interaction strategies. It also performs compliance and adaptability checks to eliminate invalid strategies and optimize the generation logic. Interaction Execution and Optimization Module: Push strategy content and monitor user feedback in real time, dynamically adjust interaction parameters or switch paths, and trigger alternative solutions when anomalies occur; Performance evaluation and analysis module: Collects multi-dimensional indicators, quantifies the advantages and disadvantages of strategies, and generates visual reports to evaluate the contribution to marketing goals; Security and Privacy Management Module: Implements multi-factor authentication, data encryption, and sensitive information anonymization throughout the entire process, and generates compliance reports to ensure secure interaction and privacy protection; Iterative optimization module: Optimizes strategy parameters, content, and paths based on evaluation results, records iteration history, and supports version backtracking and model training; Processor: The processor is used to handle the calculation process of each formula and the construction calculation process of each model.

[0018] See Figure 1 As shown, a human-computer interaction method for a digital marketing platform includes: Step 1: The platform collects user interaction behavior, device environment and scene-related data in real time, analyzes the semantics and sentiment of user input through natural language processing, and combines behavioral analysis to uncover potential needs and build a demand scenario profile; Step 2: Based on the demand scenario profile, integrate historical interaction data and real-time incremental data to build a multimodal user profile. After each interaction is completed, update the dynamic data of abnormal profiles and correct the accuracy of tags in a loop. Step 3: Based on user profiles and demand scenario profiles, call the preset strategy library and combine it with real-time marketing rules and platform resources to generate scenario-based interaction strategies. Verify the compliance and adaptability of the strategies through strategy validation, remove unqualified strategies and regenerate them. Step 4: Push interactive content and start a session based on the generated strategy, collect real-time user feedback, dynamically adjust interaction parameters, and automatically switch the interaction form or simplify the path when an interaction anomaly is detected, while providing alternative interaction strategies simultaneously. Step 5: Establish an interaction effect monitoring system, collect multi-dimensional evaluation index data during and after the interaction, identify the strengths and weaknesses of the interaction strategy, and quantify the contribution of the interaction behavior to the marketing conversion goal. Step Six: Iterate and optimize the strategy based on the evaluation results, dynamic profiles, and changes in the scenario. For any deficient parts of the strategy, correct them by adjusting parameters, optimizing content, and reconstructing paths. Record the content, basis, and effect of each iteration, and support strategy backtracking and version switching. Step 7: Perform security verification and privacy protection synchronously throughout the entire process. Ensure interaction security through multi-factor authentication and data encryption. Based on the scope of user authorization, de-identify sensitive profile data and interaction data, and generate a privacy protection compliance report.

[0019] See Figure 2 As shown, the platform collects user interaction behavior, device environment, and scene-related data in real time. It analyzes user input semantics and sentiment through natural language processing, combines this with behavioral analysis to uncover potential needs, and constructs a demand scenario profile, specifically including: The digital marketing platform collects user interaction behavior data, device environment data, and scenario-related data in real time. The interaction behavior data includes page dwell time, operation path, click frequency, semantics of input content, and interaction interruption nodes. The device environment data includes terminal type, operating system, network status, and geographical location information. The scenario-related data includes the current time dimension, marketing activity cycle, and user group tags. Natural language processing technology is used to segment user input and semantic feedback, identify intent and determine sentiment, and combine behavioral data to obtain users’ potential marketing needs. Based on equipment environment data and scenario-related data, a requirement scenario profile is constructed. The explicit and potential requirements after analysis are prioritized and a requirement analysis report is generated.

[0020] Specifically, the input content and semantic feedback undergo noise reduction preprocessing to filter out interjections, invalid symbols, and duplicate content. Then, a bidirectional matching word segmentation algorithm is used to complete accurate word segmentation. Combined with a marketing-specific corpus (covering brand terminology, marketing scenario vocabulary, and rights statements, etc.), intent recognition is achieved to accurately determine the user's core needs. Sentiment determination is achieved by combining semantic feature extraction with sentiment dictionary matching to classify three levels of sentiment: positive, neutral, and negative. The intensity of sentiment is quantified simultaneously and associated with specific statements and context. Based on user behavior data, the platform can uncover potential marketing needs behind these behaviors. For example, repeatedly browsing a certain type of marketing content without clicking to place an order indicates potential purchase needs; repeatedly checking the rules of an activity without participating indicates a need for details regarding benefits; and combining device environment data and scenario-related data to build multi-dimensional demand scenario profiles, deeply integrating time, region, activity stage, terminal status with user behavior and semantic feedback to form a scenario-based demand understanding. A weighted scoring method is used to prioritize the explicit and potential needs after analysis, with conversion potential accounting for 60% and urgency accounting for 40%. A structured needs analysis report is generated, which clarifies the needs type, core requirements, scenario characteristics, and priority ranking results.

[0021] See Figure 3 As shown, based on demand scenario profiles, historical interaction data and real-time incremental data are integrated to construct multimodal user profiles. Each time an interaction is completed, the abnormal profile dynamic data is updated cyclically, and the label accuracy is corrected. Specifically, this includes: Based on demand scenario profiles, and by integrating historical marketing interaction data and real-time incremental behavioral data, a multimodal user profile is constructed, which includes basic attribute dimensions, marketing preference dimensions, behavioral habit dimensions, conversion intention dimensions, and interaction tolerance dimensions. Based on the profile iteration mechanism, after each human-computer interaction cycle is completed, the dynamic dimension data in the profile is automatically updated, and the accuracy of the profile tags is corrected in a synchronous manner.

[0022] Specifically, a five-dimensional multimodal user profile is constructed, with each dimension achieving refined breakdown and quantitative labeling: the basic attribute dimension not only covers user identity information but also refines consumption capacity levels, supplementing core static data such as user occupation attributes and age stratification, with static data being verified quarterly; the marketing preference dimension accurately labels the types of marketing content preferred by users, the priority of receiving channels, and the acceptable push frequency threshold, while simultaneously recording the preference weights for different types of benefits such as discounts, full reductions, and new product experiences; the behavioral habit dimension determines users' commonly used interaction methods, high-frequency operation periods, page browsing preferences, and operation chain habits through behavioral sequence analysis; the conversion intention dimension adopts behavioral signal quantitative evaluation, combining data such as the number of interaction conversions in the past 30 days, the frequency of adding to cart but not paying, the depth of activity participation, and the response speed to marketing guidance to generate a conversion intention score of 0-10; the interaction tolerance dimension defines the user's acceptance threshold for marketing push density, pop-up frequency, and content length through historical negative feedback records, and labels sensitive interaction scenarios.

[0023] See Figure 4 As shown, based on user profiles and demand scenario profiles, a pre-defined strategy library is invoked, combined with real-time marketing rules and platform resources, to generate scenario-based interaction strategies. The compliance and adaptability of these strategies are verified through strategy validation, and unqualified strategies are eliminated before a new strategy is generated. Based on user profiles and demand scenario profiles, the system calls upon a pre-set marketing interaction strategy library and combines real-time marketing activity rules and platform resource compatibility to generate scenario-based interaction strategies. The interaction strategy includes interaction form selection, content presentation scheme, push timing planning and interaction path design. It adapts text, voice, graphics or multimodal interaction forms based on user interaction habits, customizes content and layout according to marketing preferences, plans push timing according to time period and scenario data, and the interaction path design responds to different interaction feedback through preset branch paths. The policy validation process verifies the compliance, resource suitability, and user experience suitability of the generated policies, removes non-compliant policies, and regenerates them.

[0024] Specifically, the platform's pre-set marketing interaction strategy library stores strategy templates in three layers: interaction form, marketing scenario, and user group. When calling a strategy, the marketing preferences and behavioral habits in the user profile are used as the core search conditions. Combined with the activity stage and regional attributes of the demand scenario profile, the candidate strategy set is locked. Based on the candidate strategy set, multi-dimensional adaptation and optimization are performed: the interaction form selection strictly matches the user behavioral habit dimension data. If the user marks "frequently used voice commands" and the current network is stable, the combination of voice + graphical interface is given priority, and the voice recognition sensitivity and command wording are optimized simultaneously. If the user prefers touch screen operation, the graphical interface hierarchy is simplified and the recognition of core buttons is enhanced. The content presentation plan is customized based on marketing preferences. For users who prefer text and images, a layout of "core benefits + concise copy + high-definition images" is used to highlight key information such as discounts and event expiration dates. For users who prefer short videos, 15-30 second short content is generated, with the core benefits presented in the first 3 seconds to suit users' attention habits. The timing of push notifications is planned based on user activity time preferences and scenario-related data, avoiding sensitive periods marked by interaction tolerance. During the peak of the event, push notifications are prioritized during high-frequency user activity times, while during the pre-heating period, push notifications are delivered lightly during low-interference times. The interaction path design aims for the "shortest conversion link," with a main path preset based on user operation habits, while 2-3 branch paths are configured to handle different feedback. For example, if a user rejects the initial push, they are automatically redirected to an alternative benefit recommendation path, simplifying the secondary operation steps.

[0025] See Figure 5 As shown, based on the generated strategy, interactive content is pushed and a session is initiated. Real-time user feedback is collected, and interaction parameters are dynamically adjusted. When an interaction anomaly is detected, the interaction method is automatically switched or the path is simplified, and alternative interaction strategies are provided simultaneously. Specifically, these include: Based on the generated scenario-based interaction strategy, push the appropriate interactive content to the user terminal and start the interactive session; During the interaction process, real-time user interaction feedback data is collected, including operation response speed, content dwell time, interaction action feedback and semantic feedback information. The interaction parameters are dynamically adjusted based on preset adaptation rules, including content presentation rhythm, operation guidance method, feedback response speed and push frequency. When abnormal user interaction behavior is detected, the strategy adjustment mechanism is automatically triggered to switch the interaction form or simplify the interaction path, while pushing alternative interaction strategies.

[0026] Specifically, the interaction execution phase adopts a terminal adaptive push mechanism, which optimizes content rendering and presentation effects according to the interaction form and user terminal type specified by the strategy: In voice interaction mode, it automatically adapts to the sensitivity of the terminal's sound pickup device, adjusts the recognition model parameters based on the user's voice habits, reduces the impact of dialects and background noise on recognition accuracy, and generates voice broadcast scripts simultaneously, controlling the speech rate and tone to match the user's emotional inclination; In graphical interface interaction mode, it adapts to the screen size and resolution of mobile and PC terminals, automatically adjusts the control layout, font size and button spacing, enhances gesture operation adaptation for touch screen terminals, and optimizes keyboard shortcuts and mouse operation logic for PC terminals; In multimodal combination mode, it realizes the coordinated linkage of voice, text, and interface elements, such as simultaneously displaying key points of text and images while broadcasting voice, and triggering supplementary voice explanations when clicking on interface elements; The entire interaction process collects real-time user feedback data, constructing a three-dimensional feedback collection system: Operation response speed is precisely recorded, noting the time interval from content push completion to the user's first action, with thresholds distinguishing between rapid response, delayed response, and no response; content dwell time is refined to the dwell time of a single element, simultaneously recording user actions on the content; action feedback captures user operation details, including swipe amplitude, click force, and operation sequence, distinguishing between active and accidental operations; semantic feedback combines contextual analysis, extracting details such as tone words and rhythm of expression in addition to core semantics to help determine the user's interaction state.

[0027] See Figure 6 As shown, an interaction effect monitoring system is established to collect multi-dimensional evaluation index data during and after the interaction, identify the strengths and weaknesses of the interaction strategy, and quantify the contribution of the interaction behavior to the marketing conversion goal. Specifically, this includes: Establish an interaction effect monitoring system to collect multi-dimensional evaluation index data in real time during the interaction execution process and after the interaction session ends. The evaluation indexes include interaction completion rate, content click rate, user dwell time, conversion behavior occurrence rate, interaction satisfaction score and negative feedback rate. By comprehensively analyzing the evaluation indicators through data visualization analysis, an interaction effect evaluation report is generated, identifying the strengths and weaknesses of the interaction strategy, and quantifying the contribution of the interaction behavior to the marketing conversion goal.

[0028] Specifically, the interaction monitoring system uses user ID and session ID as the core to build a real-time data link, achieving precise binding of behavioral data with user profiles; it adopts a dual mode of "instantaneous collection + 10-minute aggregation" to cover the entire process of interaction behavior, and ensures data quality through verification algorithms. An evaluation system is established around eight dimensions: interaction completion rate, content click-through rate, dwell time, conversion rate, satisfaction, negative feedback rate, operation smoothness, and strategy adaptability. Each indicator is clearly defined and distinguishes between valid and invalid behaviors, achieving quantitative statistics and outlier removal. The analysis phase divides the metrics into three dimensions: process, result, and experience. Process metrics use trend charts to identify weak points in the interaction chain; result metrics use funnel analysis to identify the reasons for conversion and churn; experience metrics combine user profiles to uncover the root causes of negative feedback; for abnormal metrics, behavioral sequence backtracking is used to locate problem nodes, and user profiles are used to distinguish the causes of problems into four categories: strategy design, content adaptation, interaction experience, or technical failure. The system quantifies the contribution of interactive behaviors to marketing conversion goals. Through a correlation analysis model, it establishes the relationship between each evaluation indicator and the core marketing goal, calculates the contribution weight of different interactive strategy nodes to conversion, selects the core links that have a significant impact on conversion, and simultaneously marks inefficient or ineffective interactive nodes to form a complete interactive effect evaluation report.

[0029] See Figure 7 As shown, the strategy is iteratively optimized based on evaluation results, dynamic profiles, and scene changes. For any deficient parts of the strategy, corrections are made through parameter adjustments, content optimization, and path reconstruction. The content, basis, and effects of each iteration are recorded, supporting strategy backtracking and version switching. Specifically, this includes: Based on the results of interaction effect evaluation, combined with the dynamic update data of user profiles and real-time changes in marketing scenarios, the strategy is iterated and optimized. Interaction strategy nodes that are rated as excellent are retained, while strategies with deficiencies are corrected through parameter adjustment, content optimization, and path reconstruction to generate optimized interaction strategies. The strategy generation model is trained based on iterative data, and the decision logic and adaptation algorithm of the model are optimized. By recording the content, basis and effect of each iteration, the strategy backtracking and version switching are supported.

[0030] Specifically, a dual-drive mechanism of "metric triggering + scenario adaptation" is adopted. When core metrics fall below the threshold or the marketing scenario changes, iteration is automatically initiated, and priorities are determined based on dimensions such as user reach and conversion value. The implementation process is divided into three layers: First, excellent strategies are selected to form standardized templates and stored in the library; second, problematic strategies are categorized and corrected, including reconstructing interaction paths, updating content materials, and optimizing the experience process, and verified through small-scale testing; finally, personalized strategies are dynamically adjusted based on user profiles; after each iteration, the strategy generation model is upgraded, the decision-making algorithm is optimized through data feedback, and a 15% accuracy improvement threshold is set to ensure the model's effectiveness; a full lifecycle version management system is established to achieve strategy version traceability, rapid rollback, and categorized accumulation, and invalid strategies are cleaned up quarterly to maintain the timeliness of resources in the library, ultimately forming a closed-loop ecosystem of "execution-evaluation-optimization-accumulation"; the entire system achieves continuous self-evolution of strategies through standardized processes, focusing on ensuring the efficiency of strategy optimization for high-value scenarios and core user groups.

[0031] See Figure 8 As shown, security verification and privacy protection are performed synchronously throughout the entire process. Multi-factor authentication and data encryption ensure interaction security. Based on the user's authorized scope, sensitive profile data and interaction data are anonymized, generating a privacy compliance report that specifically includes: Throughout the entire interaction process, interactive security verification and privacy protection operations are performed simultaneously. The security verification includes user identity verification, interactive data transmission encryption, operation permission control and malicious interactive behavior identification. Multi-factor authentication technology is used to ensure the authenticity and validity of user identity, and encrypted transmission protocol is used to prevent interactive data from being stolen or tampered with. The privacy protection operation is based on the user's authorized scope, de-identifies sensitive profile data and interaction data, obtains the boundaries of data collection and use, avoids privacy data calls beyond the authorized scope, and generates a privacy protection compliance report by recording data usage trajectory.

[0032] Specifically, the security verification system adopts a "real-time monitoring + hierarchical prevention and control" model, covering all aspects of the interaction; the identity legitimacy verification implements a "multi-factor hierarchical verification" mechanism: when a user logs in for the first time, triple verification is completed by combining account password, terminal characteristics, and dynamic verification code; during the session, terminal information and login status are automatically verified every 30 minutes. If risk scenarios such as terminal change, login from a different location, or abnormal IP are detected, a second verification is immediately triggered. Only after the verification is passed can the interaction continue; if it fails, the session is locked and a risk warning is pushed. Data transmission encryption adopts an end-to-end encryption scheme. After the interactive data is collected from the user terminal, it is immediately encrypted at the transport layer through the TLS1.3 protocol. Core sensitive data is additionally encrypted with symmetric encryption to generate a unique encryption key that can only be decrypted by authorized nodes. During the transmission process, a data integrity verification mechanism is enabled simultaneously to verify whether the data has been tampered with by comparing the check code. If data loss or tampering occurs, a retransmission mechanism is immediately triggered and an exception log is recorded.

[0033] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0034] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0035] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A digital marketing platform human interaction method, characterized in that, include: The platform collects user interaction behavior, device environment and scene-related data in real time, analyzes the semantics and sentiment of user input through natural language processing, and combines behavioral analysis to uncover potential needs and build demand scenario profiles. Based on demand scenario profiles, and by integrating historical interaction data and real-time incremental data, a multimodal user profile is constructed. After each interaction is completed, the abnormal profile dynamic data is updated and the label accuracy is corrected in a loop. Based on user profiles and demand scenario profiles, the system calls a pre-set strategy library and combines it with real-time marketing rules and platform resources to generate scenario-based interaction strategies. The system verifies the compliance and adaptability of the strategies through strategy validation, removes unqualified strategies, and regenerates them. Based on the generated strategy, push interactive content and start the session, collect real-time user feedback, dynamically adjust interaction parameters, and automatically switch the interaction form or simplify the path when an interaction abnormality is detected, while providing alternative interaction strategies in parallel. Establish an interaction effect monitoring system to collect multi-dimensional evaluation index data during and after the interaction, identify the strengths and weaknesses of the interaction strategy, and quantify the contribution of the interaction behavior to the marketing conversion goal. Based on the evaluation results, dynamic profiles and changes in the scenario, the strategy is iteratively optimized. For the strategy parts with deficiencies, it is corrected by adjusting parameters, optimizing content and reconstructing paths. The content, basis and effect of each iteration are recorded, and strategy backtracking and version switching are supported. Security verification and privacy protection are performed synchronously throughout the entire process. Multi-factor authentication and data encryption are used to ensure the security of interactions. Sensitive profile data and interaction data are anonymized based on the scope of user authorization, and a privacy protection compliance report is generated.

2. The digital marketing platform human interaction method of claim 1, wherein, The platform collects user interaction behavior, device environment, and scene-related data in real time. It analyzes user input semantics and sentiment through natural language processing, combines behavioral analysis to uncover potential needs, and constructs a demand scenario profile, specifically including: The digital marketing platform collects user interaction behavior data, device environment data, and scenario-related data in real time. The interaction behavior data includes page dwell time, operation path, click frequency, semantics of input content, and interaction interruption nodes. The device environment data includes terminal type, operating system, network status, and geographical location information. The scenario-related data includes the current time dimension, marketing activity cycle, and user group tags. Natural language processing technology is used to segment user input and semantic feedback, identify intent and determine sentiment, and combine behavioral data to obtain users’ potential marketing needs. Based on equipment environment data and scenario-related data, a requirement scenario profile is constructed. The explicit and potential requirements after analysis are prioritized and a requirement analysis report is generated.

3. The digital marketing platform human interaction method of claim 1, wherein, The process of constructing a multimodal user profile based on demand scenario profiling, integrating historical interaction data and real-time incremental data, and cyclically updating abnormal profile dynamic data and correcting tag accuracy after each interaction specifically includes: Based on demand scenario profiles, and by integrating historical marketing interaction data and real-time incremental behavioral data, a multimodal user profile is constructed, which includes basic attribute dimensions, marketing preference dimensions, behavioral habit dimensions, conversion intention dimensions, and interaction tolerance dimensions. Based on the profile iteration mechanism, after each human-computer interaction cycle is completed, the dynamic dimension data in the profile is automatically updated, and the accuracy of the profile tags is corrected in a synchronous manner.

4. The digital marketing platform human interaction method of claim 1, wherein, The process involves using user profiles and demand scenario profiles to call a pre-set strategy library and combine it with real-time marketing rules and platform resources to generate scenario-based interaction strategies. The strategy validation process verifies the compliance and adaptability of the strategies, eliminates unqualified strategies, and regenerates specific strategies, including: Based on user profiles and demand scenario profiles, the system calls upon a pre-set marketing interaction strategy library and combines real-time marketing activity rules and platform resource compatibility to generate scenario-based interaction strategies. The interaction strategy includes interaction form selection, content presentation scheme, push timing planning and interaction path design. It adapts text, voice, graphics or multimodal interaction forms based on user interaction habits, customizes content and layout according to marketing preferences, plans push timing according to time period and scenario data, and the interaction path design responds to different interaction feedback through preset branch paths. The policy validation process verifies the compliance, resource suitability, and user experience suitability of the generated policies, removes non-compliant policies, and regenerates them.

5. The digital marketing platform human interaction method of claim 1, wherein, The process of pushing interactive content and initiating a session based on the generated strategy, collecting real-time user feedback, dynamically adjusting interaction parameters, automatically switching the interaction format or simplifying the path when an interaction anomaly is detected, and simultaneously providing alternative interaction strategies specifically includes: Based on the generated scenario-based interaction strategy, push the appropriate interactive content to the user terminal and start the interactive session; During the interaction process, real-time user interaction feedback data is collected, including operation response speed, content dwell time, interaction action feedback and semantic feedback information. The interaction parameters are dynamically adjusted based on preset adaptation rules, including content presentation rhythm, operation guidance method, feedback response speed and push frequency. When abnormal user interaction behavior is detected, the strategy adjustment mechanism is automatically triggered to switch the interaction form or simplify the interaction path, while pushing alternative interaction strategies.

6. The human-computer interaction method for a digital marketing platform according to claim 1, characterized in that, The establishment of an interaction effect monitoring system, which collects multi-dimensional evaluation index data during and after the interaction, identifies the strengths and weaknesses of the interaction strategy, and quantifies the contribution of the interaction behavior to marketing conversion goals, specifically includes: Establish an interaction effect monitoring system to collect multi-dimensional evaluation index data in real time during the interaction execution process and after the interaction session ends. The evaluation indexes include interaction completion rate, content click rate, user dwell time, conversion behavior occurrence rate, interaction satisfaction score and negative feedback rate. By comprehensively analyzing the evaluation indicators through data visualization analysis, an interaction effect evaluation report is generated, identifying the strengths and weaknesses of the interaction strategy, and quantifying the contribution of the interaction behavior to the marketing conversion goal.

7. The human-computer interaction method for a digital marketing platform according to claim 1, characterized in that, The iterative optimization of strategies based on evaluation results, dynamic profiles, and scene changes involves correcting deficient strategies through parameter adjustments, content optimization, and path reconstruction. Each iteration's content, basis, and effects are recorded, and strategy backtracking and version switching are supported. Specifically, this includes: Based on the results of interaction effect evaluation, combined with the dynamic update data of user profiles and real-time changes in marketing scenarios, the strategy is iterated and optimized. Interaction strategy nodes that are rated as excellent are retained, while strategies with deficiencies are corrected through parameter adjustment, content optimization, and path reconstruction to generate optimized interaction strategies. The strategy generation model is trained based on iterative data, and the decision logic and adaptation algorithm of the model are optimized. By recording the content, basis and effect of each iteration, the strategy backtracking and version switching are supported.

8. The human-computer interaction method for a digital marketing platform according to claim 1, characterized in that, The process of synchronously performing security checks and privacy protection throughout the entire process, ensuring interaction security through multi-factor authentication and data encryption, and de-identifying sensitive profile data and interaction data based on user authorization scope to generate a privacy protection compliance report specifically includes: Throughout the entire interaction process, interactive security verification and privacy protection operations are performed simultaneously. The security verification includes user identity verification, interactive data transmission encryption, operation permission control and malicious interactive behavior identification. Multi-factor authentication technology is used to ensure the authenticity and validity of user identity, and encrypted transmission protocol is used to prevent interactive data from being stolen or tampered with. The privacy protection operation is based on the user's authorized scope, de-identifies sensitive profile data and interaction data, obtains the boundaries of data collection and use, avoids privacy data calls beyond the authorized scope, and generates a privacy protection compliance report by recording data usage trajectory.

9. A human-computer interaction system for a digital marketing platform, used to implement the human-computer interaction method for a digital marketing platform as described in any one of claims 1-8, characterized in that, include: Data acquisition module: Collects user interaction behavior, device environment and scene-related data in real time, analyzes semantics and sentiment through natural language processing, and builds dynamic demand scene profiles; User profile building module: It integrates historical and real-time data to generate multimodal user profiles and dynamically updates tag accuracy and abnormal data based on interactive loops; The strategy generation and verification module combines user profiles, scenario profiles, and the strategy library to generate scenario-based interaction strategies. It also removes invalid strategies and optimizes the generation logic through compliance and adaptability verification. Interaction Execution and Optimization Module: Push strategy content and monitor user feedback in real time, dynamically adjust interaction parameters or switch paths, and trigger alternative solutions when anomalies occur; Performance evaluation and analysis module: Collects multi-dimensional indicators, quantifies the advantages and disadvantages of strategies, and generates visual reports to evaluate the contribution to marketing goals; Security and Privacy Management Module: Implements multi-factor authentication, data encryption, and sensitive information anonymization throughout the entire process, and generates compliance reports to ensure secure interaction and privacy protection; Iterative optimization module: Optimizes strategy parameters, content, and paths based on evaluation results, records iteration history, and supports version backtracking and model training; Processor: The processor is used to handle the calculation process of each formula and the construction calculation process of each model.

Images