A media quality scoring and evaluation system based on AI intelligent data analysis

Abstract

The present invention provides a media quality scoring and evaluation system based on AI intelligent data analysis, wherein the data acquisition module is used to acquire media traffic data, user interaction data and media historical conversion data, and perform data preprocessing, the virtual simulation module is used to perform simulation and obtain simulation results, the media screening module is used to screen media platforms and obtain candidate media sets, the clustering optimization module is used to perform clustering optimization operations on the candidate media sets and obtain optimized candidate media sets, the media evaluation module is used to perform quality evaluation on the optimized candidate media sets and obtain media quality scores, the intelligent auxiliary module is used to generate a media selection auxiliary report according to the media quality scores of the optimized candidate media sets, the optimization feedback module is used to obtain the actual media conversion rate of the user-selected media, generate an optimization feedback report, and perform optimization according to the optimization feedback report, thereby improving the accuracy of screening media platforms.

Description

Technical Field

[0001] The present invention relates to the technical field of media quality assessment, and in particular to a media quality scoring and assessment system based on AI intelligent data analysis. Background Art

[0002] In the digital advertising space, media platforms' screening methods are constantly improving, but there's still room for improvement. Current mainstream evaluation methods primarily rely on historical data to build scoring models. While these methods provide a basic basis for reference, they can encounter adaptability issues in practice. In particular, when market conditions change or new user behavior patterns emerge, existing evaluation rules may need to be adjusted to maintain effective predictions.

[0003] Existing technical solutions may have room for improvement in addressing the complex and ever-changing advertising landscape. For one thing, traditional evaluation methods often employ relatively fixed analytical frameworks, potentially failing to adequately account for random market factors. Furthermore, when screening results differ from expected outcomes, the system's ability to self-adjust may need improvement. While these issues may not necessarily affect overall evaluation effectiveness, they may reduce screening accuracy in certain circumstances. Summary of the Invention

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides a media quality scoring and evaluation system based on AI intelligent data analysis, comprising:

[0005] A data acquisition module, which is used to acquire media flow data, user interaction data, and media historical conversion data of each media platform, and perform data preprocessing on the media flow data, user interaction data, and media historical conversion data;

[0006] A virtual simulation module, which is used to perform simulation based on media traffic data, user interaction data, and historical media conversion data to obtain simulation results for each media platform, including steady-state performance indicators and chaos resilience indicators;

[0007] a media screening module, configured to screen media platforms according to the simulation results and preset media screening rules to obtain a candidate media set;

[0008] A clustering optimization module, configured to perform clustering optimization operations on the candidate media set to obtain an optimized candidate media set;

[0009] a media evaluation module, configured to perform a quality evaluation on the optimized candidate media set according to a media quality evaluation function to obtain a media quality score of the optimized candidate media set;

[0010] an intelligent assistance module, the intelligent assistance module being configured to generate a media selection assistance report based on the media quality scores of the optimized candidate media set, wherein the media selection assistance report can be used to assist a user in selecting media;

[0011] The optimization feedback module is used to obtain the actual media conversion rate of the media selected by the user, generate an optimization feedback report, and perform optimization according to the optimization feedback report.

[0012] As a further solution of the present invention, the simulation is performed based on the media traffic data, user interaction data, and media historical conversion data to obtain simulation results for each media platform, including:

[0013] The virtual simulation module integrates media traffic data with user interaction data to dynamically generate simulated user groups and their social behaviors, and predicts the response path and conversion rate of advertisements in the virtual environment based on historical media conversion data.

[0014] When performing simulation, the virtual simulation module injects a chaotic interference source into the virtual environment, detects the dynamic response state of each media platform in the virtual environment under strong disturbance, and generates simulation results of each media platform under strong interference environment based on the dynamic response state;

[0015] The simulation results include steady-state performance indicators and chaos resilience indicators, where the steady-state performance indicators are expressed as comprehensive indicators consisting of the expected media traffic, expected click-through rate and expected conversion rate of each media platform under preset benchmark conditions; the chaos resilience indicators are used to quantify the stability, fluctuation pattern and recovery ability of the key indicators of each media platform under strong disturbance conditions, and the key indicators are expressed as the media traffic, click-through rate and conversion rate of each media platform.

[0016] As a further solution of the present invention, the step of screening media platforms according to the simulation results and preset media screening rules to obtain a candidate media set includes:

[0017] The media screening module performs preliminary screening of media platforms based on the steady-state performance indicators and chaos resilience indicator thresholds corresponding to the preset simulation results in the preset media screening rules, and sorts the media platforms that pass the preliminary screening according to the priority rules contained in the preset media screening rules to generate a candidate media set.

[0018] As a further solution of the present invention, the preset media screening rules include:

[0019] The preset media screening rules include anti-disturbance stability screening rules, behavior predictability screening rules, and risk exposure screening rules;

[0020] The anti-disturbance stability screening rule is to give priority to platforms with small deviations and fast recovery speeds in systemic indicators under strong interference. The systemic indicators are media platform media traffic, click-through rate, and conversion rate.

[0021] The behavior predictability screening rule is represented by a platform that favors user interaction patterns that do not exhibit abnormal variations during stress testing, wherein the user interaction patterns are represented by interaction patterns between virtual users generated based on user interaction data during simulation.

[0022] The risk exposure screening rule is expressed as eliminating platforms that trigger a chain collapse of systemic indicators under slight disturbances.

[0023] As a further solution of the present invention, performing cluster optimization on the candidate media set to obtain the optimized candidate media set includes:

[0024] The clustering optimization module extracts features from the candidate media set to obtain a media service attribute feature set, clusters the candidate media set based on the chaos resilience index and the media service attribute feature set corresponding to the candidate media set, and divides the candidate media set into multiple media clusters with similar internal features;

[0025] Calculating the chaos resilience comprehensive scores of the plurality of media clusters according to the chaos resilience comprehensive scoring function, eliminating the secondary clusters within the media cluster based on the chaos resilience comprehensive scores, and retaining the core cluster with the highest chaos resilience comprehensive score within the media cluster;

[0026] When the number of core clusters is lower than the preset core cluster number threshold, high-quality platforms with characteristics orthogonal to the core cluster are selected from the suboptimal clusters to fill the core clusters until the number of core clusters is no lower than the preset core cluster number threshold, and the core clusters are output as the optimized candidate media set.

[0027] As a further solution of the present invention, performing quality evaluation on the optimized candidate media set according to a media quality evaluation function to obtain a media quality score of the optimized candidate media set includes:

[0028] The media evaluation module obtains the comprehensive chaos resilience score and media business attributes of each candidate media platform included in the optimized candidate media set, and calculates the media quality score of each candidate media platform according to the media quality evaluation function.

[0029] As a further solution of the present invention, generating a media selection assistance report based on the media quality scores of the optimized candidate media set includes:

[0030] Media selection assistance reports include anti-chaos rating and input effectiveness ratio;

[0031] The anti-chaos rating is expressed as the level of the media platform's ability to resist market disturbances;

[0032] The investment-effectiveness ratio is expressed as the ratio of the quantified investment cost to the target user coverage efficiency.

[0033] As a further solution of the present invention, obtaining the actual media conversion rate of the user-selected media, generating an optimization feedback report, and performing optimization according to the optimization feedback report includes:

[0034] The actual media conversion rate of the media selected by the user is obtained, the actual media conversion rate is compared with the expected value in the corresponding steady-state performance indicator, and the corresponding preset rules are optimized according to the comparison result.

[0035] As a further solution of the present invention, obtaining media flow data, user interaction data, and media historical conversion data of each media platform, and performing data preprocessing on the media flow data, user interaction data, and media historical conversion data, includes:

[0036] The media traffic data includes at least the number of unique visitors, page views and average visit duration;

[0037] The user interaction data includes at least the number of user likes, comments and shares;

[0038] The media historical conversion data includes at least advertisement click-through rate and advertisement conversion rate;

[0039] Abnormal data in the media flow data, user interaction data, and media historical conversion data are eliminated and corrected, and the media flow data, the user interaction data, and the media historical conversion data are unified into the same dimension.

[0040] Based on the above aspects, the embodiment of the present application realizes the acquisition of media traffic data, user interaction data and media historical conversion data of each media platform based on the data acquisition module, and performs data preprocessing on the media traffic data, user interaction data and media historical conversion data. The virtual simulation module performs simulation based on the media traffic data, user interaction data and media historical conversion data to obtain the simulation results of each media platform. The simulation results include steady-state performance indicators and chaos resilience indicators. The media screening module screens the media platforms according to the simulation results and preset media screening rules to obtain the candidate media set, and obtains the expected performance of each media platform through simulation. At the same time, by injecting a strong interference source with chaotic characteristics to interfere with the simulation environment, the stability, fluctuation pattern and recovery ability of each media platform when encountering interference are judged, and the media platforms are screened according to the preset screening rules to eliminate unstable media platforms.

[0041] Based on the clustering optimization module, clustering optimization operations are performed on the candidate media set to obtain an optimized candidate media set. The media evaluation module performs quality evaluation on the optimized candidate media set according to the media quality evaluation function to obtain the media quality score of the optimized candidate media set. The media platforms are further screened through clustering operations, and media platforms with high similarity in the candidate media set are eliminated, so that the range of candidate media platforms is further narrowed. The media quality score is used as a mathematical indicator to quantify the value of each media platform.

[0042] The intelligent assistance module generates a media selection assistance report based on the media quality scores of the optimized candidate media set. The media selection assistance report can be used to assist users in selecting media, and assist users in selecting a media platform that meets their needs in the form of an assistance report.

[0043] The optimization feedback module obtains the actual media conversion rate of the media selected by the user, generates an optimization feedback report, performs optimization based on the optimization feedback report, and continuously optimizes the preset rules through feedback optimization to adapt to the complex and changing market environment, thereby improving the accuracy of media screening. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a schematic diagram of the execution flow of control steps in a media quality scoring and evaluation system based on AI intelligent data analysis provided by an embodiment of the present invention;

[0045] Figure 2 This is a schematic diagram of a media quality scoring and evaluation system based on AI intelligent data analysis provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0046] The present invention will be described in detail below with reference to the accompanying drawings. Figure 2A schematic diagram of a media quality scoring and evaluation system based on AI intelligent data analysis, which can realize the concept of the present application, is shown in some embodiments of the present application.

[0047] Specifically, a media quality scoring and evaluation system based on AI intelligent data analysis includes:

[0048] A data acquisition module, which is used to acquire media flow data, user interaction data, and media historical conversion data of each media platform, and perform data preprocessing on the media flow data, user interaction data, and media historical conversion data;

[0049] A virtual simulation module, which is used to perform simulation based on media traffic data, user interaction data, and historical media conversion data to obtain simulation results for each media platform, including steady-state performance indicators and chaos resilience indicators;

[0050] a media screening module, configured to screen media platforms according to the simulation results and preset media screening rules to obtain a candidate media set;

[0051] A clustering optimization module, configured to perform clustering optimization operations on the candidate media set to obtain an optimized candidate media set;

[0052] a media evaluation module, configured to perform a quality evaluation on the optimized candidate media set according to a media quality evaluation function to obtain a media quality score of the optimized candidate media set;

[0053] an intelligent assistance module, the intelligent assistance module being configured to generate a media selection assistance report based on the media quality scores of the optimized candidate media set, wherein the media selection assistance report can be used to assist a user in selecting media;

[0054] The optimization feedback module is used to obtain the actual media conversion rate of the media selected by the user, generate an optimization feedback report, and perform optimization according to the optimization feedback report.

[0055] Figure 1 This is a schematic diagram of the execution flow of control steps in a media quality scoring and evaluation system based on AI intelligent data analysis provided by an embodiment of the present invention. The control steps in the media quality scoring and evaluation system based on AI intelligent data analysis are introduced in detail below.

[0056] Step S10: acquiring media flow data, user interaction data, and media historical conversion data of each media platform based on the data acquisition module, and performing data preprocessing on the media flow data, user interaction data, and media historical conversion data.

[0057] Specifically, the data acquisition module collects multi-source data from each media platform to obtain media flow data, user interaction data, and media historical conversion data of each media platform.

[0058] It can be understood that the media traffic data includes at least the number of independent visitors, page views and average visit duration; the user interaction data includes at least the number of user likes, comments and shares; and the media historical conversion data includes at least the advertising click-through rate and advertising conversion rate.

[0059] Furthermore, abnormal data of the media traffic data, the user interaction data and the media historical conversion data are eliminated and corrected. For example, if the number of independent visitors of a certain media platform suddenly increases by 200% in one day, it is diagnosed as CDN hijacked traffic after tracing the source, and the data is truncated according to the IQR method, and the media traffic data, the user interaction data and the media historical conversion data are unified into the same dimension.

[0060] It is understandable that the Lyapunov exponent in chaos theory is introduced to quantify the sensitivity of media platforms to traffic fluctuations. Through chaotic time series analysis, such as Takens embedding theorem, it is possible to detect whether there are abnormal fluctuation patterns in platform traffic, such as pseudo-random characteristics caused by brushing.

[0061] As you can understand, the Lyapunov index calculation algorithm is selected according to different data types. For long-term monitoring data, such as media traffic data over 90 days, the Wolf algorithm is used to capture system sensitivity and obtain the corresponding Lyapunov index. For short-term behavioral data, such as 7-day active page click streams, the Rosenstein algorithm is used for rapid evaluation to obtain the corresponding Lyapunov index.

[0062] At the same time, the obtained Lyapunov index is compared with the benchmark value of the corresponding media industry, and abnormal media are marked based on the comparison results. For example, if the Lyapunov index corresponding to the historical media conversion data of an e-commerce platform is 0.25, which exceeds the industry benchmark of 0.12, the e-commerce platform is marked as having abnormal brushing or lacking anti-interference ability. During the simulation, the marked platform will be verified in detail.

[0063] In step S20, the virtual simulation module performs simulation based on media traffic data, user interaction data, and media historical conversion data to obtain simulation results of each media platform. The simulation results include steady-state performance indicators and chaos resilience indicators. The media screening module screens media platforms based on the simulation results and preset media screening rules to obtain a candidate media set.

[0064] In this embodiment, step S20 includes:

[0065] In step S21 , the virtual simulation module performs simulation based on the media traffic data, user interaction data, and media historical conversion data to obtain simulation results for each media platform.

[0066] Specifically, the virtual simulation module integrates media traffic data with user interaction data, dynamically generates simulated user groups and the social behaviors of the simulated user groups, and predicts the response path and conversion rate of advertisements in the virtual environment based on historical media conversion data to generate steady-state performance indicators.

[0067] It can be understood that when performing simulation, the virtual simulation module injects chaotic interference sources into the virtual environment, detects the dynamic response state of each media platform in the virtual environment under strong disturbance, and generates simulation results of each media platform in a strong interference environment based on the dynamic response state.

[0068] Specifically, non-periodic dynamic noise is injected into the virtual environment, and the noise intensity is dynamically adjusted according to the Lyapunov exponent corresponding to each media platform. After the noise is injected, the response behavior of the key indicators of each media platform is tracked in real time. The key indicators are represented by the media traffic, click-through rate and conversion rate of each media platform.

[0069] It can be understood that by real-time tracking the changes in the distance between adjacent points of traffic data after the injection of non-periodic dynamic noise, and taking the divergence speed per minute as the chaos response index, the stability of the media platform is quantified based on the chaos response index; the peak values ​​of the three indicators of media traffic, click-through rate and conversion rate deviating from the baseline are taken as the maximum relative deviation, and the volatility risk of the media platform is quantified based on the maximum relative deviation; the time required for the indicators to return to the steady state after the noise is removed is obtained, and the recovery capacity level of the media platform is divided based on the time required for the indicators to return to the steady state after the noise is removed.

[0070] Furthermore, a chaos resilience index is generated based on the chaos response index, the maximum relative deviation and the recovery capability level, and anomaly detection is performed based on the chaos resilience index. If the deviation between the real-time chaos resilience index and the corresponding Lyapunov exponent exceeds 0.05, the anomaly detection mechanism is triggered to verify the mechanical oscillation characteristics of the media platform. For example, the typical spectrum peak spacing of the volume-boosting platform is less than 0.5Hz. Finally, the steady-state performance index and the chaos resilience index are output as simulation results.

[0071] It can be understood that the simulation results include steady-state performance indicators and chaos resilience indicators, where the steady-state performance indicators are expressed as comprehensive indicators consisting of the expected media traffic, expected click-through rate and expected conversion rate of each media platform under preset benchmark conditions, and the chaos resilience indicators are used to quantify the stability, fluctuation pattern and recovery ability of the key indicators of each media platform under strong disturbance conditions.

[0072] In step S22 , the media screening module screens media platforms according to the simulation results and preset media screening rules to obtain a candidate media set.

[0073] Specifically, the media screening module performs an initial screening of the media platform based on the steady-state performance indicators and chaos resilience indicator thresholds corresponding to the preset simulation results in the preset media screening rules. For example, it requires that the average daily expected media traffic in the steady-state performance indicators be no less than 500,000 and the fluctuation range cannot exceed 15% of the expected media traffic, and the expected conversion rate cannot be lower than 80% of the industry average and the deviation cannot exceed 2%. It requires that the chaos response index of the chaos resilience indicator be no less than 0.6, the maximum relative deviation be no more than 20%, and the recovery time corresponding to the recovery capability level be no more than 10 minutes.

[0074] Furthermore, the media platforms that have passed the initial media platform screening are sorted according to the priority rules included in the preset media screening rules to generate a candidate media set.

[0075] It can be understood that the preset media screening rules include anti-disturbance stability screening rules, behavior predictability screening rules and risk exposure screening rules.

[0076] The anti-disturbance stability screening rule is to give priority to platforms with small deviations and fast recovery speeds in systemic indicators under strong interference. The systemic indicators are media platform media traffic, click-through rate, and conversion rate.

[0077] The behavior predictability screening rule is represented by a platform that favors user interaction patterns that do not exhibit abnormal variations during stress testing, wherein the user interaction patterns are represented by interaction patterns between virtual users generated based on user interaction data during simulation.

[0078] The risk exposure screening rule is expressed as eliminating platforms that trigger a chain collapse of systemic indicators under slight disturbances.

[0079] Step S30: performing a cluster optimization operation on the candidate media set based on the cluster optimization module to obtain an optimized candidate media set; and performing a quality evaluation on the optimized candidate media set by the media evaluation module according to a media quality evaluation function to obtain a media quality score of the optimized candidate media set.

[0080] In this embodiment, step S30 includes:

[0081] Step S31 : performing cluster optimization operations on the candidate media set based on the cluster optimization module to obtain an optimized candidate media set.

[0082] Specifically, the clustering optimization module extracts features from the candidate media set to obtain a media business attribute feature set, which includes platform type, such as short video or information, core user groups, such as age distribution and regional concentration, content update frequency, and advertising adaptation industries, such as e-commerce, education, or entertainment. The candidate media set is clustered based on the chaos resilience index corresponding to the candidate media set and the media business attribute feature set, and the candidate media set is divided into multiple groups of media clusters with similar internal features.

[0083] For example, the multiple media platforms included in the candidate media set are divided into three clusters: high-resilience precise adaptation group, medium-resilience general traffic group, and scenario efficiency group. The characteristics of the high-resilience precise adaptation group are that it is mainly based on live broadcast platforms, users have strong consumption power, the beauty industry advertising matching degree exceeds 90%, and the anti-disturbance ability is outstanding, that is, the average maximum relative deviation of traffic of the media platforms in the cluster in the chaos test is only 9%, and the time required for the indicators to return to steady state after the noise is removed is on average within 5 minutes; the characteristics of the medium-resilience general traffic group are that it is dominated by short video platforms, with a wide user coverage but lower user consumption power than the high-resilience precise adaptation group, and the advertising industry adaptability is about It is 65%, and the anti-interference ability performance is qualified, that is, the average maximum relative deviation of the traffic of the media platform in the cluster in the chaos test is only 18%, and the time required for the indicators to return to the steady state after the noise is removed is an average of less than 10 minutes; the characteristics of the scenario efficiency group are mainly vertical scenario platforms for IoT devices such as in-vehicle voice, and users have high scenario conversion potential. For example, the purchase intention during commuting hours increases by 3.5 times, the precise reach rate of cars or smart homes is 81%, and the decision-making link is compressed to 3.7 minutes. The anti-interference ability performance is excellent, that is, the average maximum relative deviation of the traffic of the media platform in the cluster in the chaos test is only 5%, and the time required for the indicators to return to the steady state after the noise is removed is an average of less than 3 minutes.

[0084] Furthermore, the chaos resilience comprehensive scores of the multiple media clusters are calculated according to the chaos resilience comprehensive scoring function. The secondary clusters within the media cluster are eliminated based on the chaos resilience comprehensive scores, and the core cluster with the highest chaos resilience comprehensive score within the media cluster is retained. The chaos resilience comprehensive scoring function can be expressed as:

[0085] ;

[0086] in, Expressed as a comprehensive score of chaos resilience, Expressed as the chaos response index weight, the default value is 0.4, Expressed as the chaos response index, Expressed as the maximum relative deviation weight, the default value is 0.3, Expressed as the maximum relative deviation, It represents the weight of the time required for the indicator to return to steady state after noise removal. The default value is 0.3. It is expressed as the time required for the indicator to return to steady state after the noise is removed.

[0087] It can be understood that all cluster scores are arranged in descending order, the quantile of the bottom 25% is calculated, all clusters below this quantile are determined as secondary clusters, and the secondary clusters are eliminated. At the same time, the cluster with the highest score is selected from the retained clusters as the core cluster, such as selecting the top 25% quantile and determining all clusters above this quantile as the core cluster.

[0088] It can be understood that when the number of core clusters is lower than the preset core cluster number threshold, high-quality platforms with characteristics orthogonal to the core cluster are selected from the suboptimal clusters to fill the core clusters until the number of core clusters is not lower than the preset core cluster number threshold, and the core clusters are output as the optimized candidate media set.

[0089] For example, if the number of core clusters is less than the preset threshold, the orthogonal supplement mechanism is activated to expand the cluster. In the first round, it is determined that only the live broadcast cluster meets the core standards. It is necessary to screen the suboptimal platform from the suboptimal cluster with the largest difference in user characteristics from the live broadcast cluster and qualified anti-interference. Among them, the live broadcast cluster is mainly composed of women aged 25 to 40, so platforms with a user overlap rate of less than 20% are given priority. The advertising conversion rate of the in-vehicle voice cluster during the commuting peak is 18%, which is higher than the 4.2% of the live broadcast cluster. The maximum relative deviation of the anti-interference test is 7%, which is relatively stable. The user characteristics of the in-vehicle voice cluster overlap with the live broadcast cluster by 12%, and there is no overlap between the commuting scene and the prime time of live broadcast. The fitness APP cluster is excluded because the overlap rate of young female users is 41%. The smart home cluster has a user overlap rate of 23% and a scene overlap of 35%. As the second choice, the in-vehicle voice platform with the highest orthogonality index is given priority, and the smart home cluster is expanded to the core cluster.

[0090] In step S32 , the media evaluation module performs quality evaluation on the optimized candidate media set according to a media quality evaluation function to obtain a media quality score of the optimized candidate media set.

[0091] Specifically, the media evaluation module obtains the comprehensive chaos resilience score and media service attributes of each candidate media platform included in the optimized candidate media set, and calculates the media quality score of each candidate media platform according to the media quality evaluation function. The media quality evaluation function can be expressed as:

[0092] ;

[0093] in, Expressed as steady-state advertising value weight, Expressed as steady-state advertising value, Expressed as the comprehensive score weight of chaos resilience, Expressed as a comprehensive score of chaos resilience, Expressed as industry adaptation coefficient, Expressed as time period efficiency coefficient.

[0094] It can be understood that the steady-state advertising value is composed of the traffic value coefficient, the click value coefficient and the conversion value coefficient. The traffic value coefficient is expressed as the expected media traffic of the media platform divided by the median media traffic of the industry's top platforms. For example, if the expected media traffic of a certain media platform is 1.8 million and the median media traffic of the industry's top platforms is 2 million, then the corresponding traffic value coefficient is 0.9; the click value coefficient is expressed as the expected click-through rate of the media platform divided by the industry platform's click-through rate benchmark. For example, if the expected click-through rate of a certain media platform is 2.4% and the industry platform's click-through rate benchmark is 2.0%, then the corresponding click value coefficient is 1.2; the conversion value coefficient is expressed as the expected conversion rate of the media platform divided by the industry platform's conversion rate benchmark. For example, if the expected conversion rate of a certain media platform is 2% and the industry platform's conversion rate benchmark is 2%, then the corresponding conversion value coefficient is 1.0.

[0095] As you can understand, the industry adaptation coefficient is set based on the platform industry label and the matching degree of the advertiser. For example, the vertical field precise matching is set to [1.1, 1.2], the general industry matching is set to 1.0, and the low correlation is set to [0.8, 0.9].

[0096] It can be understood that the setting of the time period efficiency coefficient is expressed as dynamic adjustment according to the advertising plan period, where the prime time is represented by 19:00 to 22:00, the regular time is 7:00 to 18:59, and the off-peak time is 22:01 to 6:59. The time period efficiency coefficients of 1.2, 1.0 and 0.9 are set for the prime time, regular time and off-peak time respectively.

[0097] Step S40: The intelligent assistance module generates a media selection assistance report according to the media quality scores of the optimized candidate media set. The media selection assistance report can be used to assist the user in selecting media.

[0098] It can be understood that the media selection auxiliary report includes anti-chaos rating and investment efficiency ratio. The anti-chaos rating represents the level of the media platform's ability to resist market disturbances. The anti-chaos rating is divided into three levels: S, A and B. Among them, the S level indicates that the chaos resilience score is greater than 0.85, the chaos response index is greater than 0.9, and the conversion fluctuation is less than 10% when experiencing price wars with competitors. Media platforms of this level are recommended to adapt to high-budget brand advertising; A level indicates that the chaos resilience score is in the range of [0.75, 0.85), the chaos response index is in the range of [0.7, 0.8], and the maximum relative deviation is less than 20%. Media platforms of this level are recommended to adapt to regular promotional delivery; B level indicates that the chaos resilience score is in the range of [0.65, 0.75), and the time required for the indicators to return to steady state after noise removal is more than 15 minutes. Media platforms of this level are recommended to be observed.

[0099] The investment effectiveness ratio is expressed as the product of the target user reach and the scene conversion gain divided by the cost per exposure. The target user reach is expressed as the number of effective exposures based on user portrait matching, such as the proportion of female users aged 25-40 reached by beauty ads; the scene conversion gain is expressed as the conversion lift rate of orthogonal scene delivery, such as the difference between the 3.1% conversion rate of the in-vehicle platform during commuting hours and the average of 0.9%; the cost per exposure is expressed as the industry benchmark cost adjustment value.

[0100] Step S50: The optimization feedback module obtains the actual media conversion rate of the media selected by the user, generates an optimization feedback report, and performs optimization according to the optimization feedback report.

[0101] Specifically, the actual media conversion rate of the media selected by the user is obtained, the actual media conversion rate is compared with the expected value in the corresponding steady-state performance indicator, and the corresponding preset rule is optimized according to the comparison result.

[0102] For example, a beauty brand placed an advertisement on a luxury live streaming platform with an S-level Chaos rating. The expected conversion rate was 2.5%, but the actual conversion rate for three consecutive periods was only 1.8% to 2.0%. Based on the recent actual conversion data collected from the platform, it was found that the decision-making cycle of high-end customers had prolonged. Therefore, the original rule of "requiring the conversion rate benchmark to be greater than 2.0%" was optimized to generate a new rule of "lowering the conversion rate benchmark for similar platforms to 1.9%". Simulation was then carried out to verify the effect of the optimized rule; a new energy vehicle placed an advertisement on an in-vehicle voice platform. The expected conversion rate during commuting hours was 3.0%, but the actual conversion rate reached 4.1%, exceeding the expected value. Therefore, the original industry adaptation coefficient of 1.0 was optimized to 1.25 during peak hours.

[0103] The specific usage and function of this embodiment are described below:

[0104] First, based on the data acquisition module, the media traffic data, user interaction data and media historical conversion data of each media platform are obtained, and the media traffic data, user interaction data and media historical conversion data are preprocessed. The virtual simulation module performs simulation based on the media traffic data, user interaction data and media historical conversion data to obtain the simulation results of each media platform. The simulation results include steady-state performance indicators and chaos resilience indicators. The media screening module screens the media platforms according to the simulation results and preset media screening rules to obtain the candidate media set. The expected performance of each media platform is obtained through simulation. At the same time, the simulation environment is interfered by injecting a strong interference source with chaotic characteristics to evaluate and quantify the stability, fluctuation pattern and recovery ability of each media platform when encountering interference. Subsequently, the media platforms are screened according to the preset screening rules to eliminate unstable media platforms, thereby narrowing the screening range of the media platform and reducing the calculation amount and processing time of the data.

[0105] Secondly, based on the clustering optimization module, cluster optimization operations are performed on the candidate media set to obtain an optimized candidate media set. The media evaluation module performs quality evaluation on the optimized candidate media set according to the media quality evaluation function to obtain the media quality score of the optimized candidate media set. The clustering operation is used to further screen the media platforms, eliminate redundant media platforms with high similarity in the candidate media set, further narrow the range of candidate media platforms, and use the media quality score as a mathematical indicator to quantify the value of the remaining media platforms.

[0106] Then, the intelligent assistance module generates a media selection assistance report according to the media quality scores of the optimized candidate media set, and assists the user in selecting a media platform that meets his or her needs by generating the assistance report.

[0107] The final optimization feedback module obtains the actual conversion rate of the media selected by the user, generates an optimization feedback report, performs optimization based on the optimization feedback report, judges the accuracy of the screening based on the actual conversion rate, and generates a corresponding optimization strategy. The preset rules are optimized according to the optimization strategy to adapt to the complex and changing market environment, thereby improving the accuracy of media screening.

[0108] In addition, an embodiment of the present invention further provides an electronic device, including:

[0109] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method proposed in the first embodiment of the present invention.

[0110] The following is a detailed introduction to the various components of electronic equipment:

[0111] The term "processor" is the control center of an electronic device and can be a single processor or a collective term for multiple processing elements. For example, the processor can be one or more central processing units (CPUs), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the first embodiment of the present invention, such as one or more digital signal processors (DSPs) or one or more field programmable gate arrays (FPGAs).

[0112] The processor can execute various functions of the electronic device by running or executing software programs stored in the memory and calling data stored in the memory.

[0113] The memory is used to store the software program for executing the solution of the present invention, and the execution is controlled by the processor. The specific implementation method can refer to the above method embodiment and will not be repeated here.

[0114] The memory may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, a random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only (CD-ROM), or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto. The memory may be integrated with the processor or exist independently and be coupled to the processor via an interface circuit of the electronic device, and this is not specifically limited in the embodiments of the present invention.

[0115] The above embodiments can be implemented in whole or in part via software, hardware (e.g., circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in accordance with the embodiments of the present invention are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wireless communication (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer, or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0116] It should be understood that the term "and / or" as used herein simply describes an association between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent the existence of A alone, the existence of both A and B, or the existence of B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the related objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0117] It should be understood that in the embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0118] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A media quality scoring and evaluation system based on AI intelligent data analysis, characterized in that: include: A data acquisition module, which is used to acquire media flow data, user interaction data, and media historical conversion data of each media platform, and perform data preprocessing on the media flow data, user interaction data, and media historical conversion data; A virtual simulation module, which is used to perform simulation based on media traffic data, user interaction data, and historical media conversion data to obtain simulation results for each media platform, including steady-state performance indicators and chaos resilience indicators; The virtual simulation module integrates media traffic data with user interaction data to dynamically generate simulated user groups and their social behaviors, and predicts the response path and conversion rate of advertisements in the virtual environment based on historical media conversion data. When performing simulation, the virtual simulation module injects a chaotic interference source into the virtual environment, detects the dynamic response state of each media platform in the virtual environment under strong disturbance, and generates simulation results of each media platform under strong interference environment based on the dynamic response state; The simulation results include steady-state performance indicators and chaos resilience indicators. The steady-state performance indicators are expressed as comprehensive indicators consisting of the expected media traffic, expected click-through rate, and expected conversion rate of each media platform under preset benchmark conditions. The chaos resilience indicators are used to quantify the stability, fluctuation pattern, and recovery ability of key indicators of each media platform under strong disturbance conditions. The key indicators are expressed as media traffic, click-through rate, and conversion rate of each media platform. a media screening module, configured to screen media platforms according to the simulation results and preset media screening rules to obtain a candidate media set; A clustering optimization module, configured to perform clustering optimization operations on the candidate media set to obtain an optimized candidate media set; a media evaluation module, configured to perform a quality evaluation on the optimized candidate media set according to a media quality evaluation function to obtain a media quality score of the optimized candidate media set; an intelligent assistance module, the intelligent assistance module being configured to generate a media selection assistance report based on the media quality scores of the optimized candidate media set, wherein the media selection assistance report can be used to assist a user in selecting media; An optimization feedback module, which is used to obtain the actual media conversion rate of the media selected by the user, generate an optimization feedback report, and perform optimization according to the optimization feedback report; The actual media conversion rate of the media selected by the user is obtained, the actual media conversion rate is compared with the expected value in the corresponding steady-state performance indicator, and the corresponding preset rules are optimized according to the comparison result.

2. The media quality scoring and evaluation system based on AI intelligent data analysis according to claim 1 is characterized in that: The step of screening media platforms according to the simulation results and preset media screening rules to obtain a candidate media set includes: The media screening module performs preliminary screening of media platforms based on the steady-state performance indicators and chaos resilience indicator thresholds corresponding to the preset simulation results in the preset media screening rules, and sorts the media platforms that pass the preliminary screening according to the priority rules contained in the preset media screening rules to generate a candidate media set.

3. The media quality scoring and evaluation system based on AI intelligent data analysis according to claim 2 is characterized in that: The preset media screening rules include: The preset media screening rules include anti-disturbance stability screening rules, behavior predictability screening rules, and risk exposure screening rules; The anti-disturbance stability screening rule is to give priority to platforms with small deviations and fast recovery speeds in systemic indicators under strong interference. The systemic indicators are media platform media traffic, click-through rate, and conversion rate. The behavior predictability screening rule is represented by a platform that favors user interaction patterns that do not exhibit abnormal variations during stress testing, wherein the user interaction patterns are represented by interaction patterns between virtual users generated based on user interaction data during simulation. The risk exposure screening rule is expressed as eliminating platforms that trigger a chain collapse of systemic indicators under slight disturbances.

4. The media quality scoring and evaluation system based on AI intelligent data analysis according to claim 1 is characterized in that: The clustering optimization operation is performed on the candidate media set to obtain the optimized candidate media set, including: The clustering optimization module extracts features from the candidate media set to obtain a media service attribute feature set, clusters the candidate media set based on the chaos resilience index and the media service attribute feature set corresponding to the candidate media set, and divides the candidate media set into multiple media clusters with similar internal features; Calculating the chaos resilience comprehensive scores of the plurality of media clusters according to the chaos resilience comprehensive scoring function, eliminating the secondary clusters within the media cluster based on the chaos resilience comprehensive scores, and retaining the core cluster with the highest chaos resilience comprehensive score within the media cluster; When the number of core clusters is lower than the preset core cluster number threshold, high-quality platforms with characteristics orthogonal to the core cluster are selected from the suboptimal clusters to fill the core clusters until the number of core clusters is no lower than the preset core cluster number threshold, and the core clusters are output as the optimized candidate media set.

5. The media quality scoring and evaluation system based on AI intelligent data analysis according to claim 1 is characterized in that: The performing quality evaluation on the optimized candidate media set according to the media quality evaluation function to obtain a media quality score of the optimized candidate media set includes: The media evaluation module obtains the comprehensive chaos resilience score and media business attributes of each candidate media platform included in the optimized candidate media set, and calculates the media quality score of each candidate media platform according to the media quality evaluation function.

6. The media quality scoring and evaluation system based on AI intelligent data analysis according to claim 1 is characterized in that: Generating a media selection assistance report according to the media quality scores of the optimized candidate media set includes: Media selection assistance reports include anti-chaos rating and input effectiveness ratio; The anti-chaos rating is expressed as the level of the media platform's ability to resist market disturbances; The investment-effectiveness ratio is expressed as the ratio of the quantified investment cost to the target user coverage efficiency.

7. The media quality scoring and evaluation system based on AI intelligent data analysis according to claim 1 is characterized in that: The acquiring of media flow data, user interaction data, and media historical conversion data of each media platform, and performing data preprocessing on the media flow data, user interaction data, and media historical conversion data, includes: The media traffic data includes at least the number of unique visitors, page views and average visit duration; The user interaction data includes at least the number of user likes, comments and shares; The media historical conversion data includes at least advertisement click-through rate and advertisement conversion rate; Abnormal data in the media flow data, user interaction data, and media historical conversion data are eliminated and corrected, and the media flow data, the user interaction data, and the media historical conversion data are unified into the same dimension.

Images