Advertising effect analysis method and system based on artificial intelligence

Abstract

The invention discloses an advertising effect analysis method and system based on artificial intelligence, and relates to the technical field of commercial advertising prediction, the advertising effect analysis method based on artificial intelligence can significantly improve the accuracy and effectiveness of advertisement putting, and through multi-modal biological feature collection, the advertising effect is improved. Data of brain waves and eye movement tracks when consumers watch advertisements are obtained in real time, the nerve excitement degree, the attention concentration degree and the emotional response of the consumers are deeply known, a nerve behavior decoding engine constructs a three-layer cognitive mapping model based on the data, the real-time transformation probability of the consumers is accurately predicted, and the conversion efficiency of the consumers is improved. The dynamic advertisement gene pool specifically adjusts visual, auditory and narrative elements of the advertisement according to the prediction result, personalized customization of the advertisement content is achieved, the accurate advertisement putting mode can better meet the requirements and interests of consumers, the matching degree of the advertisement and the consumers is improved, and therefore the conversion rate of the advertisement is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of commercial advertising prediction technology, specifically to an artificial intelligence-based method and system for analyzing advertising effectiveness. Background Technology

[0002] Advertising, as a crucial means for businesses to promote products and services and enhance brand awareness, faces an increasingly competitive environment. With the widespread adoption of the internet, mobile devices, and social media, advertising channels and forms are becoming increasingly diversified, such as video ads, social media ads, and search engine ads. Consumers are exposed to a massive amount of advertising information every day, making it crucial for ads to attract consumers' attention and drive purchasing behavior in a short period of time. Therefore, how to accurately evaluate the effectiveness of advertising, adjust advertising strategies in real time based on consumer feedback, and improve advertising conversion rates and return on investment has become a key issue that the advertising industry urgently needs to address.

[0003] Currently, traditional advertising effectiveness analysis methods mainly rely on macro data indicators such as click-through rate, conversion rate, and impressions. These methods can only reflect the overall effect of advertising to a certain extent, but they have obvious limitations: on the one hand, they cannot deeply understand the real psychological state and emotional response of consumers when watching advertisements, and it is difficult to accurately grasp consumers' interests and needs; on the other hand, existing technologies lack effective utilization of multimodal biometric data. When consumers watch advertisements, their brain waves, eye movements, and other biometric characteristics can directly reflect their neural excitability and attention concentration, but traditional methods have failed to incorporate these data into the advertising effectiveness analysis system, resulting in the inability to achieve real-time, accurate evaluation and dynamic optimization of advertising effectiveness.

[0004] In conclusion, existing methods for analyzing advertising effectiveness are inadequate in the face of an increasingly complex advertising market environment and cannot meet the advertising industry's needs for precise marketing and efficient promotion. Therefore, developing an artificial intelligence-based method and system for analyzing advertising effectiveness, which utilizes multimodal biometric data to deeply analyze consumer psychology and behavior, and enables real-time evaluation and dynamic adjustment of advertising effectiveness, has significant practical implications and application value. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies by providing an artificial intelligence-based method and system for analyzing advertising effectiveness. This system significantly improves the accuracy and effectiveness of advertising placement. Through multimodal biometric data collection, it acquires real-time data such as brainwaves and eye movements of consumers while viewing advertisements, gaining a deeper understanding of consumers' neural excitability, attention concentration, and emotional responses. A neurobehavioral decoding engine constructs a three-layer cognitive mapping model based on this data to accurately predict consumers' real-time conversion probability. A dynamic advertising gene library then adjusts the visual, auditory, and narrative elements of the advertisement based on the prediction results, achieving personalized customization of advertising content. This precise advertising placement method better meets consumers' needs and interests, improves the matching degree between advertisements and consumers, and thus effectively increases the conversion rate of advertisements.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: On the one hand, an artificial intelligence-based method for analyzing advertising effectiveness, the specific steps of which are as follows:

[0007] S100: Collect the user's EEG alpha and beta neurophysiological signals and pupil movement trajectory, and perform spatiotemporal alignment using timestamp t;

[0008] S200 constructs a three-layer cognitive mapping through a neurobehavioral decoding engine, including a physiological layer, a behavioral layer, and a cognitive layer, and calculates the physiological layer attention index (ACI), the behavioral layer consistency score (BCS), and the cognitive layer memory encoding strength (MES).

[0009] S300 performs logistic regression prediction on ACI, BCS, and MES to generate real-time conversion probability values;

[0010] S400. Dynamically adjust the gene combination of the advertising element gene unit according to the real-time conversion probability value. The advertising element gene unit includes visual genes, auditory genes, and narrative genes.

[0011] S500 uses a weighted average of the historical contribution of advertising element gene units to eliminate the visual, auditory, and narrative gene units with the lowest contribution and replace them with new genes, thus dynamically switching the advertising version.

[0012] Furthermore, the state dynamic parsing of the S200 three-layer cognitive mapping is as follows:

[0013] The Physiological Attention Index (ACI) is calculated by constructing an exponential decay function based on the ratio of alpha wave energy to baseline value and the duration of ad playback, which measures the user's level of attention to the ad at the physiological level.

[0014] Behavioral Consistency Score (BCS): This analysis quantifies the temporal correlation between pupil diameter changes and interactive behavior using dynamic time warping, with the window set to... It is used to measure the consistency between a user's physiological responses and behavioral actions;

[0015] The Memory Encoding Strength Assessment (MES) method detects sudden increases in wavelet coherence in the β-band, calculates the phase synchronization cumulative energy of the frequency band, and then assesses the degree to which users encode advertising content.

[0016] Furthermore, the physiological attention index (ACI) comprehensively considers the impact of real-time collected alpha wave energy and the duration of advertisement playback on attention, i.e. ,in, Indicates at time The real-time acquisition of alpha wave energy reflects the current intensity of alpha wave activity in the brain. It is the baseline energy of the alpha wave. For the duration of the advertisement, This is the attention decay time constant, with a default value of 20 seconds, used to adjust the rate at which attention decays over time.

[0017] Furthermore, the Behavioral Consistency Score (BCS) analysis assesses user interest and engagement with the advertisement. Changes in pupil diameter reflect changes in user emotion and attention, while mouse movement speed reflects user interaction behavior. ,in For time-shifting windows, cov Indicates the rate of change of pupil diameter With mouse movement speed In time shift The covariance under these conditions reflects the degree of linear correlation between the two. and These are the standard deviations of the pupil diameter change rate and mouse movement speed, respectively, used to normalize the covariance. arrive The maximum correlation coefficient is found within the time-shifted window to capture the best temporal association between the two.

[0018] Furthermore, the cognitive layer memory encoding strength MES uses wavelet coherence analysis to detect sudden increases in β-band phase synchronization, thereby assessing the degree to which users encode advertising content, i.e., the depth of the impression the advertisement leaves in the user's brain. The calculation formula is as follows: ,in, These are the wavelet transform coefficients for the β band. The square of the modulus of the β-band wavelet transform coefficients reflects the energy distribution of the signal at different times and frequencies. Its value at 30... up to 40 Integrate over the frequency range and divide by the frequency. To comprehensively consider the signal energy within this frequency range, the cognitive layer memory coding strength is obtained.

[0019] Furthermore, the index obtained by the S300 through the integrated three-layer cognitive mapping is obtained through... To predict the real-time conversion probability of users, the influence of physiological attention index, behavioral consistency score, and cognitive memory encoding strength on conversion is assigned weights of 0.7, 0.25, and 0.05, respectively, to calculate the probability value of a user converting while watching an advertisement. .

[0020] Furthermore, the S400 gene combination initiates the gene recombination process when any triggering condition is met:

[0021] Condition 1: When When the ACI is <0.65 and ACI <0.6, the combination of visual gene weighting of 70% and narrative gene weighting of 30% is executed for 3 seconds.

[0022] Condition 2: When a single test When the value is <0.65 and BCS <0.4, the combination of auditory gene weighting of 60% and visual gene weighting of 40% is executed.

[0023] Condition 3: When When MES < 0.3 and the value is in the range [0.65, 0.8), the combination of narrative gene weight of 80% and auditory gene weight of 20% is executed for 5 seconds.

[0024] Condition 4: When When the value is >0.8, a combination of 50% weight for the visual gene and 50% weight for the narrative gene is triggered.

[0025] Furthermore, the S500 historical contribution decay weighted value is used to dynamically evaluate the historical contribution of each advertising gene unit, preferentially replacing inefficient genes through a time decay mechanism. The gene unit with the lowest contribution score is moved to the adjustment pool. Based on triggering conditions, a matching gene type is selected from the pool for replacement. The contribution score of gene unit g is given, where w is the time decay factor and w=0.98, and k is the gene activation number and k≥1. The conversion rate increase at the kth activation and , This represents the real-time conversion probability before gene activation. denoted as , where is the real-time conversion probability after gene activation, and n is the total number of activations.

[0026] On the other hand, an AI-based advertising effectiveness analysis system comprises: a multimodal biometrics acquisition module, a neural behavior decoding module, a dynamic advertising gene bank module, a real-time rendering module, and a gene bank optimization module.

[0027] The multimodal biometric acquisition module is used to collect EEG signals and pupil movement data of users when they watch advertisements in real time, and to preprocess the collected data and transmit the processed data to the neurobehavioral decoding module.

[0028] The neurobehavioral decoding module receives data preprocessed by the multimodal biometric acquisition module, constructs a three-layer cognitive mapping model, and calculates the physiological layer attention index, the behavioral layer consistency score, and the cognitive layer memory encoding strength.

[0029] The dynamic advertising gene library module predicts the real-time conversion probability based on the physiological layer attention index, the behavioral layer consistency score, and the cognitive layer memory encoding strength, and sends the real-time conversion probability and indicator data to the real-time rendering module.

[0030] The real-time rendering module adjusts the advertising content in a targeted manner based on the real-time conversion probability and indicators output by the neurobehavioral decoding module, combined with the advertising gene unit.

[0031] The gene pool optimization module eliminates genes with low contribution based on the historical contribution decay weighting value of advertising gene units, and autonomously evolves the advertising gene pool rule base.

[0032] Compared with existing technologies, this AI-based advertising effectiveness analysis method and system has the following advantages:

[0033] I. The AI-based advertising effectiveness analysis method of this invention can significantly improve the accuracy and effectiveness of advertising placement. Through multimodal biometric collection, it acquires real-time data on consumers' brainwaves and eye movements when watching advertisements, gaining a deeper understanding of consumers' neural excitability, attention concentration, and emotional responses. The neurobehavioral decoding engine constructs a three-layer cognitive mapping model based on this data to accurately predict consumers' real-time conversion probability. The dynamic advertising gene library then adjusts the visual, auditory, and narrative elements of the advertisement based on the prediction results, achieving personalized customization of advertising content. This precise advertising placement method can better meet consumers' needs and interests, improve the matching degree between advertisements and consumers, and thus effectively improve the conversion rate of advertisements.

[0034] Second, the AI-based advertising effectiveness analysis system of this invention enables real-time monitoring and dynamic optimization of advertising effectiveness. The real-time rendering engine quickly adjusts the advertising version based on real-time consumer feedback to ensure that the advertisement is always presented to consumers in the most attractive form. At the same time, it comprehensively records and analyzes various data during the advertising process, providing strong data support for the formulation and optimization of advertising strategies. Through long-term data accumulation and analysis, it can gain a deep understanding of consumer behavior patterns and preferences, and continuously optimize advertising content and placement strategies.

[0035] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0037] Figure 1 This is a flowchart illustrating the operation of an AI-based advertising effectiveness analysis method.

[0038] Figure 2 This is a flowchart illustrating the steps of an AI-based advertising effectiveness analysis system. Detailed Implementation

[0039] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0040] Example 1

[0041] like Figure 2 As shown, this embodiment focuses on the application of AI-based advertising effectiveness analysis methods in actual advertising scenarios. By elaborating on the entire process from multimodal biometric data collection to constructing cognitive mapping analysis, and then adjusting advertising genes and dynamically switching advertising versions based on the analysis results, it demonstrates how this method can improve the accuracy and effectiveness of advertising placement and achieve real-time monitoring and optimization of advertising effects.

[0042] Multimodal biometric data acquisition (S100): When a user browses a relevant website, multimodal biometric data acquisition is triggered. Biometric data is collected in real time through an EEG sensor and a pupil motion tracking camera. The EEG sensor collects the EEG signals of the user when watching advertisements, focusing on the neurophysiological signals in the alpha (8-12Hz) and beta (13-30Hz) bands. These signals represent the activity of the brain in different states. Alpha waves are related to the brain's relaxation and attention state, while beta waves are closely linked to states such as cognition and excitement. The sensor continuously acquires signals to ensure the accuracy and continuity of the data. At the same time, the pupil motion tracking camera records the user's pupil movement trajectory. Changes in the pupil can intuitively reflect changes in the user's emotions and attention. For example, when a user is interested in the advertisement content, the pupil may dilate. The collected EEG signals and pupil movement trajectory data are assigned a timestamp t, so that data from different sources can be accurately aligned in time and space, ensuring precise synchronization of the two types of data in time.

[0043] Construction of a Three-Layer Cognitive Map and Indicator Calculation (S200): The collected data is transmitted to the neurobehavioral decoding engine, which constructs a three-layer cognitive map, analyzes it from the physiological, behavioral, and cognitive levels, and calculates corresponding indicators. These indicators include: Physiological layer Attention Index (ACI), Behavioral layer Consistency Score (BCS), and Cognitive layer Memory Encoding Strength (MES). The physiological layer primarily focuses on the user's attention level to the advertisement at a physiological level. Based on the collected alpha wave energy, its ratio to the baseline value is calculated. Alpha wave energy is collected when the user is in a relatively calm state without specific stimuli. This serves as a reference standard to measure changes in current alpha wave energy. Simultaneously, an exponential decay function is constructed by combining this with the advertisement playback duration T. , This represents the alpha wave energy collected in real time at time t. It reflects the current intensity of alpha wave activity in the brain. As the advertising playback time T increases, the user's attention gradually decreases. The attention decay time constant is... To adjust the rate of this decay, alpha wave energy is continuously collected during the advertisement playback. And combined with the current ad playback duration T and baseline value The algorithm calculates ACl, and by continuously calculating ACl, it can understand in real time the changes in user attention to the advertisement at the physiological level. The behavioral layer measures user interest and engagement with the advertisement by analyzing the consistency between user physiological responses and behavioral actions. Dynamic time warping quantifies the temporal correlation between pupil diameter changes and interactive behavior. The window is set to... In the formula In the middle, cov Indicates the rate of change of pupil diameter With mouse movement speed In time shift The covariance under time shift τ reflects the degree of linear correlation between the two factors. If the changes in pupil diameter and mouse movement speed show a high covariance over time shift τ, it indicates that the user's physiological response and behavior are highly consistent at that point in time, meaning that the user may have become interested in the advertising content and made corresponding interactive actions. and These are the standard deviations of the pupil diameter change rate and mouse movement speed, respectively, used to normalize the covariance, eliminate the influence of different data scales, and make the calculation results more comparable. The maximum correlation coefficient is found within a time-shift window of -300ms to +300ms to capture the optimal temporal correlation between the two. By collecting pupil diameter change rate (pupil(t)) and mouse movement speed (cursor(t)) data in real time, at each sampling time t, within the time range of -300ms to +300ms, the covariance cov(pupil(t), cursor(t+τ)) under different time shifts τ is calculated and divided by the corresponding standard deviation. and The maximum value among these calculations is the current BCS value, used to understand the consistency changes between user behavior and physiological responses, and thus to determine the user's interest in and engagement with the advertisement. The cognitive layer uses wavelet coherence analysis to detect sudden increases in beta-band phase synchronization, thereby assessing the degree to which the user's memory encodes the advertisement content—that is, the depth of the impression the advertisement leaves in the user's brain. ,in, These are the wavelet transform coefficients for the β band. The square of the modulus of the wavelet transform coefficients in the β band reflects the energy distribution of the signal at different times and frequencies. Integrating it over the frequency range of 30Hz to 40Hz and dividing by the frequency f allows for a comprehensive consideration of the signal energy within this frequency range, yielding the cognitive layer memory encoding strength (MES). Within this frequency range, β wave activity is related to brain cognition and memory. When a phase synchronization surge occurs, it means that the brain has made a strong memory encoding of the advertising content at that moment. By continuously calculating the MES, the changes in the user's memory encoding strength of the advertising content can be monitored in real time.

[0044] Real-time conversion probability prediction (S300): The neurobehavioral decoding engine integrates three indicators: physiological level attention index (ACI), behavioral level consistency score (BCS), and cognitive level memory encoding strength (MES), and generates a real-time conversion probability value formula through logistic regression prediction. Based on the respective impacts of the physiological attention index (ACI), behavioral consistency score (BCS), and cognitive encoding strength (BCS) on conversion, different weights of 0.7, 0.25, and 0.05 were assigned. These weights reflect the relative importance of each indicator in predicting user conversion behavior. The physiological attention index (ACI) reflects the user's basic level of attention to the advertisement and has a significant impact on conversion, hence its high weight. The behavioral consistency score (BCS) reflects the user's engagement and interest, and its weight is secondary. Through this formula, the three indicators are combined to obtain a probability value that reflects whether a user will convert (such as purchasing a product or clicking a link) while watching an advertisement. As the ACI, BCS, and MES values ​​change during ad playback, the real-time conversion probability value... It will also be updated in real time to provide a basis for subsequent advertising adjustments.

[0045] Dynamically adjust the ad element gene unit (S400): based on the calculated real-time conversion probability value. The base combination of advertising element gene units is dynamically adjusted. These advertising element gene units include visual genes, auditory genes, and narrative genes. The trigger condition for adjusting the gene combination is as follows:

[0046] Condition 1: When When the ACI is <0.65 and ACI <0.6, the combination of visual gene weight of 70% and narrative gene weight of 30% is executed for 3 seconds. This is because when the conversion probability is low and the physiological attention index is also low, it means that the advertisement has not attracted users in terms of visual presentation and content narrative, so it is necessary to focus on adjusting these two aspects of the gene.

[0047] Condition 2: When a single test When the value is <0.65 and BCS <0.4, the combination of auditory gene weight of 60% and visual gene weight of 40% is executed. This indicates that the consistency between the user's physiological response and behavior is poor, indicating that the coordination between the auditory and visual elements of the advertisement is poor. Therefore, these two genes should be adjusted.

[0048] Condition 3: When When MES < 0.3, and the duration is 5 seconds, the combination of narrative gene weight of 80% and auditory gene weight of 20% is executed. At this time, the real-time conversion probability is at a medium level, but the cognitive layer memory encoding strength is low, indicating that the advertisement needs to be improved in terms of content narrative and auditory assistance in order to deepen the user's memory of the advertisement.

[0049] Condition 4: When When the value is >0.8, a combination of 50% weight for the visual gene and 50% weight for the narrative gene is triggered. This means that the advertisement is currently performing well. Maintaining the balance between the visual and narrative genes is important to keeping users engaged.

[0050] Dynamic Ad Version Switching (S500): Based on the historical contribution decay weighted value of ad gene units, genes with low contribution are eliminated, and the ad gene library rule base evolves autonomously to achieve dynamic switching of ad versions, through the historical contribution decay weighted value. Calculate the contribution score for each advertising gene unit g, where w is the time decay factor and w=0.98, indicating that the gene's contribution gradually decreases over time. This is because the advertising market and user needs are constantly changing, and past contributions have less reference value in the present. k is the gene activation number and k≥1. The conversion rate increase at the kth activation and This represents the real-time conversion probability before gene activation. The real-time conversion probability after gene activation is represented by , where n is the total number of activations. The conversion improvement rate of the gene at different activation counts and the decay effect of time on its contribution are comprehensively considered. The average contribution score of the gene unit is obtained by weighted summing of the conversion improvement rates after each gene activation and dividing by the total number of activations. A lower score indicates a smaller contribution of the gene unit to the advertising effect. During ad delivery, whenever a gene is activated (i.e., the advertising gene unit is adjusted), the real-time conversion probability before activation is recorded. and real-time conversion probability after activation And calculate the conversion rate. As the number of activations increases, the contribution score of the gene unit is continuously updated. Similarly, as advertising continues, the contribution score of each gene unit is constantly updated based on new activations. The contribution scores of all advertising gene units are periodically ranked, and the gene unit with the lowest contribution score is removed from the advertising gene pool. This is because low-contribution gene units cannot effectively improve advertising conversion rates. Simultaneously, during the removal of low-contribution genes, the reasons for their removal and their combinations with other genes are analyzed. Based on the analysis results, the advertising gene pool rule base evolves. This dynamic switching mechanism ensures that advertisements are always presented to users in the most attractive and convertible form, improving advertising effectiveness. In this way, it is possible to more accurately understand user needs and interests, adjust advertising strategies in a timely manner, and improve advertising attractiveness and conversion rates. Furthermore, the self-evolutionary capability allows the advertising gene pool and rule base to continuously adapt to market changes and dynamic adjustments to user needs, providing an efficient and intelligent solution for advertising effectiveness analysis and optimization.

[0051] In summary, this embodiment demonstrates in detail the complete operation process of the AI-based advertising effectiveness analysis method in a real-world advertising scenario. Starting with the collection of multimodal biometric data, it calculates various indicators by constructing a three-layer cognitive mapping model to predict real-time conversion probabilities, then dynamically adjusts the advertising element gene units based on probabilities, and finally achieves dynamic switching of advertising versions based on gene contribution, thus realizing real-time monitoring and optimization of advertising effectiveness.

[0052] Example 2

[0053] like Figure 1 As shown in the figure, this embodiment provides a specific process for advertising promotion and optimization using an artificial intelligence-based advertising effectiveness analysis method:

[0054] (1) Data collection:

[0055] Using brainwave sensor devices, the alpha and beta brainwave signals of users when they watch advertisements are collected.

[0056] Using an eye-tracking camera, the user's pupil movement trajectory is recorded at a high frame rate, including changes in pupil size and direction of movement;

[0057] Accurate timestamps are added to the collected EEG signals and pupil movement trajectory data to ensure precise alignment of the data in time and space;

[0058] (2) User interaction behavior data recording:

[0059] Record user interaction behavior during ad viewing, such as mouse movement speed, clicks, and page scrolling;

[0060] (3) Data Analysis

[0061] Based on the collected alpha wave energy and the duration of the advertisement playback, the physiological attention index (ACI) is calculated to measure the user's level of attention to the advertisement at the physiological level.

[0062] By quantifying the temporal correlation between pupil diameter changes and interactive behaviors through dynamic time warping, a Behavioral Consistency Score (BCS) is calculated to assess the consistency between users' physiological responses and behavioral actions.

[0063] Wavelet coherence analysis was used to detect sudden phase synchronization increases in the β-band, and the cognitive layer memory encoding strength (MES) was calculated to determine the degree of user memory encoding of advertising content.

[0064] (4) Predicting real-time conversion probability:

[0065] The three indicators—physiological attention index (ACI), behavioral consistency score (BCS), and cognitive memory encoding strength (MES)—are combined and real-time conversion probability values ​​are generated through logistic regression prediction.

[0066] Based on real-time conversion probability values ​​and ACI, BCS, and MES indicators, the gene combination of advertising element gene units is dynamically adjusted according to triggering conditions. Advertising element gene units include visual genes, auditory genes, and narrative genes.

[0067] (5) Dynamic switching of ad versions:

[0068] Calculate the historical contribution decay weighted value for each advertising gene unit. This value comprehensively considers the conversion improvement rate of the gene under different activation times and the decay effect of time on the contribution.

[0069] Regularly rank all advertising gene units by contribution score and eliminate gene units with low contribution scores;

[0070] Based on the updates to the gene pool and the evolution of the rule base, the current advertising effect and user feedback are rendered in real time, and the advertising version is dynamically switched, with priority given to displaying the advertising version containing high-contribution gene combinations.

[0071] (6) Effectiveness evaluation and continuous optimization:

[0072] Continuously monitor real-time conversion rates, ACI, BCS, MES, and other metrics after ad placement to evaluate the effectiveness of ad adjustments;

[0073] Analyze user feedback data to understand users' views and opinions on the advertisements;

[0074] Based on the results of the effectiveness evaluation, the gene combination of advertising element gene units and the advertising version will be further adjusted to continuously optimize the advertising effect;

[0075] Based on dynamic adjustments to user needs, we continuously update the advertising gene library and rule library to improve the accuracy and effectiveness of ad placement.

[0076] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An artificial intelligence-based method for analyzing advertising effectiveness, characterized in that, The specific steps of this method are as follows: S100: Collect the user's EEG alpha and beta neurophysiological signals and pupil movement trajectory, and perform spatiotemporal alignment using timestamp t; S200. A three-layer cognitive mapping is constructed through a neurobehavioral decoding engine, including a physiological layer, a behavioral layer, and a cognitive layer. The physiological layer attention index (ACI), the behavioral layer consistency score (BCS), and the cognitive layer memory encoding strength (MES) are calculated. The state dynamics of the three-layer cognitive mapping are as follows: The Physiological Attention Index (ACI) is calculated by constructing an exponential decay function based on the ratio of alpha wave energy to baseline value and the duration of ad playback, which measures the user's level of attention to the ad at the physiological level. Behavioral consistency score (BCS) analysis: quantifying the temporal correlation between pupil diameter changes and interactive behavior through dynamic time warping; MES for assessing cognitive layer memory encoding strength: Detecting beta-band wavelet coherence surge events, calculating phase synchronization cumulative energy of the frequency band, and then assessing the degree of user memory encoding of advertising content; S300 performs logistic regression prediction on ACI, BCS, and MES to generate real-time conversion probability values; S400. Dynamically adjust the gene combination of the advertising element gene unit according to the real-time conversion probability value. The advertising element gene unit includes visual genes, auditory genes, and narrative genes. S500 uses a weighted average of the historical contribution of advertising element gene units to eliminate the visual, auditory, and narrative gene units with the lowest contribution and replace them with new genes, thus dynamically switching the advertising version.

2. The advertising effectiveness analysis method based on artificial intelligence according to claim 1, characterized in that, The physiological attention index (ACI) comprehensively considers the impact of real-time collected alpha wave energy and the duration of advertising playback on attention. ,in, Indicates at time The real-time acquisition of alpha wave energy reflects the current intensity of alpha wave activity in the brain. It is the baseline energy of the alpha wave. For the duration of the advertisement, This represents the attention decay time constant.

3. The advertising effectiveness analysis method based on artificial intelligence according to claim 1, characterized in that, The Behavioral Consistency Score (BCS) analysis assesses user interest and engagement with the advertisement. Changes in pupil diameter reflect changes in user emotion and attention, while mouse movement speed reflects user interaction behavior. ,in For time-shifting windows, cov Indicates the rate of change of pupil diameter With mouse movement speed In time shift The covariance under these conditions reflects the degree of linear correlation between the two. and These are the rate of change of pupil diameter and the standard deviation of mouse movement speed, respectively.

4. The advertising effectiveness analysis method based on artificial intelligence according to claim 1, characterized in that, The cognitive layer memory encoding strength MES uses wavelet coherence analysis to detect sudden increases in β-band phase synchronization, assessing the degree to which users encode advertising content, i.e., the depth of the impression left by the advertisement in the user's brain. The calculation formula is as follows: ,in, These are the wavelet transform coefficients for the β band. This represents the square of the modulus of the β-band wavelet transform coefficients.

5. The advertising effectiveness analysis method based on artificial intelligence according to claim 1, characterized in that, The indicators obtained from the S300's integrated three-layer cognitive mapping are, through... To predict the real-time conversion probability of users, the influence of physiological attention index, behavioral consistency score, and cognitive memory encoding strength on conversion is assigned weights of 0.7, 0.25, and 0.05, respectively, to calculate the probability value of a user converting while watching an advertisement. .

6. The advertising effectiveness analysis method based on artificial intelligence according to claim 1, characterized in that, The S400 gene combination initiates the gene recombination process when any of the triggering conditions are met: Condition 1: When When the ACI is <0.65 and ACI <0.6, the combination of visual gene weighting of 70% and narrative gene weighting of 30% is executed for 3 seconds. Condition 2: When a single test When the value is <0.65 and BCS <0.4, the combination of auditory gene weighting of 60% and visual gene weighting of 40% is executed. Condition 3: When When MES < 0.3 and the value is in the range [0.65, 0.8), the combination of narrative gene weight of 80% and auditory gene weight of 20% is executed for 5 seconds. Condition 4: When When the value is >0.8, a combination of 50% weight for the visual gene and 50% weight for the narrative gene is triggered.

7. The advertising effectiveness analysis method based on artificial intelligence according to claim 1, characterized in that, The S500 historical contribution decay weighted value is used to dynamically evaluate the historical contribution of each advertising gene unit. ,pass Gene units with the lowest contribution scores are moved to the adjustment pool, where... The contribution score of gene unit g is given, where w is the time decay factor and w=0.98, and k is the gene activation number and k≥1. The conversion rate increase at the kth activation and , This represents the real-time conversion probability before gene activation. denoted as , where is the real-time conversion probability after gene activation, and n is the total number of activations.

8. An AI-based advertising effectiveness analysis system, applicable to the AI-based advertising effectiveness analysis method described in any one of claims 1-7, characterized in that, The system comprises: a multimodal biometric acquisition module, a neural behavior decoding module, a dynamic advertising gene bank module, a real-time rendering module, and a gene bank optimization module. The multimodal biometric acquisition module is used to collect EEG signals and pupil movement data of users when they watch advertisements in real time, and to preprocess the collected data and transmit the processed data to the neurobehavioral decoding module. The neurobehavioral decoding module receives data preprocessed by the multimodal biometric acquisition module, constructs a three-layer cognitive mapping model, and calculates the physiological layer attention index, the behavioral layer consistency score, and the cognitive layer memory encoding strength. The dynamic advertising gene library module predicts the real-time conversion probability based on the physiological layer attention index, the behavioral layer consistency score, and the cognitive layer memory encoding strength, and sends the real-time conversion probability and indicator data to the real-time rendering module. The real-time rendering module adjusts the advertising content in a targeted manner based on the real-time conversion probability and indicators output by the neurobehavioral decoding module, combined with the advertising gene unit. The gene pool optimization module eliminates genes with low contribution based on the historical contribution decay weighting value of advertising gene units, and autonomously evolves the advertising gene pool rule base.

Images