Advertisement putting effect intelligent monitoring method and system based on cloud video data analysis

Abstract

The application discloses an advertisement putting effect intelligent monitoring method and system based on cloud video data analysis, relates to the field of intelligent monitoring, generates advertisement characteristic information through cloud video platform server data, generates user behavior information according to thread data fluctuation, obtains behavior difference values through the difference between the behavior information before and after the user browses the advertisement, generates advertisement acceptance in combination with the advertisement characteristic information, obtains advertisement putting historical data, generates advertisement induced search probability, generates advertisement effect evaluation values and advertisement audience portraits in combination with the advertisement acceptance, the behavior difference values and the advertisement induced search probability, analyzes the advertisement induced search probability and the advertisement audience portrait according to the advertisement putting historical data, sets a user putting effect threshold, analyzes the advertisement effect evaluation values based on the user putting effect threshold, and judges whether the advertisement effect evaluation values are greater than the user putting effect threshold.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent monitoring, in particular to an advertisement delivery effect intelligent monitoring method and system based on cloud video data analysis. BACKGROUND

[0002] Cloud video data analysis refers to the process of collecting, storing, processing and analyzing video content, user behavior, and playback effect data through a cloud computing platform. With the powerful computing power and massive data storage capacity of cloud computing, the quality of video content and user experience can be improved. Advertisement delivery effect refers to the comprehensive impact of an advertisement after its delivery on communication, brand building, and sales promotion. It is a key indicator of whether an advertisement is successful.

[0003] Existing advertisement delivery effect intelligent monitoring methods and systems usually collect data from a single platform, resulting in incomplete data. Simple statistical analysis of the collected data cannot provide detailed information about user characteristics. The delivery effect can only be monitored by determining whether the advertisement has been viewed by the user. In-depth data analysis of user behavior motivation is difficult, and the intention of complex user behavior is ignored, resulting in a large difference between the monitoring results and the actual advertisement delivery effect, which cannot meet the needs of the advertisement delivery party. SUMMARY

[0004] To solve the above technical problems, an advertisement delivery effect intelligent monitoring method and system based on cloud video data analysis are provided. This technical solution solves the problem of being unable to understand user characteristics in detail, making it difficult to analyze user behavior motivation in depth, ignoring the intention of complex user behavior, and resulting in a large difference between the monitoring results and the actual advertisement delivery effect.

[0005] To achieve the above purposes, the technical solution adopted by the present application is as follows:

[0006] An advertisement delivery effect intelligent monitoring method based on cloud video data analysis includes:

[0007] Obtain and analyze cloud video platform server data to generate advertisement feature information and user behavior information;

[0008] Based on the user behavior information, obtain a behavior difference value, and combine the advertisement feature information to generate an advertisement acceptance degree;

[0009] Based on the cloud video platform, obtain advertisement delivery history data to generate an advertisement-induced search probability;

[0010] According to the advertisement acceptance degree, the behavior difference value, and the advertisement-induced search probability, generate an advertisement effect evaluation value and advertisement audience information;

[0011] Classify the advertising audience information to generate an advertising audience portrait;

[0012] According to the advertising delivery history data, analyze the advertising-induced search probability and the advertising audience portrait, and set a user delivery effect threshold;

[0013] Based on the user delivery effect threshold, analyze the advertising effect evaluation value, and determine whether the advertising effect evaluation value is greater than the user delivery effect threshold. If yes, optimize the advertising content according to the advertising audience portrait to maintain the freshness and attention of the audience to the advertising. If no, adjust the advertising delivery according to the advertising audience portrait.

[0014] Preferably, the generation of advertising feature information and user behavior information specifically includes:

[0015] Process the cloud video platform server data to obtain server thread occupation and advertising information, wherein the advertising information includes advertising product content and advertising duration;

[0016] Generate advertising feature information according to the advertising product content;

[0017] Obtain the user's login ID on the cloud video platform, bind the user's login ID with the server thread based on the cloud video server data, and obtain the user's associated thread;

[0018] According to the user's associated thread, analyze the server thread occupation to obtain thread fluctuation performance information, wherein the thread fluctuation performance information includes execution state fluctuation performance information, thread number fluctuation performance information, and resource occupation fluctuation performance information;

[0019] Generate user behavior information according to the thread fluctuation performance information;

[0020] Determine whether the user has paused the video playback through the execution state fluctuation performance information of the user's associated thread, determine whether the user has interacted with the video through the thread number fluctuation performance information of the user's associated thread, and determine whether the user has performed a download operation on the video through the resource occupation fluctuation performance information of the user's associated thread.

[0021] Preferably, the generation of the behavior difference value in combination with the advertising feature information generates an advertising acceptance, specifically including:

[0022] According to the cloud video platform server data and the thread fluctuation performance information, obtain the thread fluctuation time;

[0023] According to the advertising duration, analyze the resource occupation fluctuation time to determine the advertising time period;

[0024] According to the advertisement time period, the thread fluctuation time is divided, and the thread occupation information before the advertisement and the thread occupation information after the advertisement are generated in combination with the server thread occupation situation;

[0025] With three minutes as a time period, the execution state fluctuation performance information, the thread quantity fluctuation performance information and the resource occupation fluctuation performance information are analyzed to obtain the execution state fluctuation frequency information, the thread quantity fluctuation frequency information and the resource occupation fluctuation frequency information;

[0026] From the execution state, the thread quantity and the resource occupation fluctuation, the thread occupation information before the advertisement and the thread occupation information after the advertisement are analyzed to generate the execution state fluctuation frequency difference, the thread quantity fluctuation frequency difference and the resource occupation fluctuation frequency difference respectively;

[0027] Based on the Pearson correlation coefficient calculation formula, the thread quantity fluctuation frequency difference and the resource occupation fluctuation frequency difference are analyzed with the execution state fluctuation frequency difference as a reference variable to obtain the thread quantity fluctuation correlation coefficient and the resource occupation fluctuation correlation coefficient respectively;

[0028] According to the ratio of the thread quantity fluctuation correlation coefficient and the resource occupation fluctuation correlation coefficient, the behavior difference value is obtained;

[0029] According to the advertisement delivery historical data, an advertisement perception table is generated;

[0030] The advertisement feature information is substituted into the advertisement perception table to obtain a standard difference value, and the advertisement acceptance degree is generated in combination with the behavior difference value, and the advertisement acceptance degree state transition probability of different behavior difference values is calculated to establish an advertisement acceptance degree matrix.

[0031] Preferably, the generation of the advertisement-induced search probability specifically comprises:

[0032] The advertisement delivery historical data is sorted and divided according to the advertisement acceptance degree to generate user influence state information;

[0033] According to the user influence state information, the advertisement delivery historical data is analyzed to calculate the user state probability of different acceptance degrees to generate a user influence state transition probability matrix;

[0034] The advertisement acceptance degree matrix and the user influence state transition probability matrix are analyzed to generate the advertisement-induced search probability;

[0035] The specific calculation formula of the advertisement-induced search probability is:

[0036] P[s k+1 |s k ]=P[s1|s2,s3,s4]P[v|v1,v2,v3];

[0037] P[s k+1 |s k ] is the probability of advertisement-induced search, s k+1 is the (k+1)th user influence state, s k is the kth user influence state, s1, s2, s3, s4 are user influence states, v is the current advertisement receptivity state, v1, v2, v3 are advertisement receptivity states, P[s1|s2, s3, s4] is the user influence state transition probability, and P[v|v1, v2, v3] is the advertisement receptivity state transition probability.

[0038] Preferably, the generating the advertisement effect evaluation value and the advertisement audience information specifically comprises:

[0039] Obtaining the registration information of the user login ID, classifying the user information in combination with the advertisement receptivity, obtaining the average advertisement receptivity of different user groups, and generating the advertisement audience information;

[0040] Obtaining the total number of advertisement browsing persons, determining the distribution of different user groups in the total number of advertisement browsing persons in combination with the advertisement audience information;

[0041] Weighted summing the advertisement-induced search probability corresponding to the average advertisement receptivity of each user group and the distribution of each user group in the total number of advertisement browsing persons, to generate the advertisement effect evaluation value.

[0042] Preferably, the setting the user delivery effect threshold value specifically comprises:

[0043] According to the advertisement audience portrait, classifying the same type of advertisement delivery historical data according to age information and regional information respectively, to generate age delivery effect data and regional delivery effect data;

[0044] Respectively, the age delivery effect data and the regional delivery effect data are analyzed and arranged to obtain the age delivery effect peak value data, the age delivery effect valley value data, the regional delivery effect peak value data and the regional delivery effect valley value data;

[0045] The age delivery effect peak value data and the age delivery effect valley value data are averaged and summed to generate the age delivery effect threshold value, and the regional delivery effect peak value data and the regional delivery effect valley value data are averaged and summed to generate the regional delivery effect threshold value;

[0046] According to the registration information of the advertisement browsing user, the corresponding age delivery effect threshold value and the regional delivery effect threshold value are obtained, and the average value is calculated to generate the user delivery effect threshold value.

[0047] Further, an advertisement delivery effect intelligent monitoring system based on cloud video data analysis is proposed, which is used to realize the above-mentioned monitoring method, and is characterized in that it specifically comprises:

[0048] A data collection module is configured to be connected with the cloud video platform server, acquire data of the cloud video platform server, acquire advertisement information from the advertisement party, and transmit the collected information data to the feature extraction module and the data analysis module.

[0049] A feature extraction module is configured to extract features from the received data according to the required features, generate advertisement feature information according to the advertisement product content, extract the age information and the regional information of the user according to the registration information of the user login ID, and transmit the extracted feature information to the data analysis module.

[0050] A data analysis module is configured to analyze the received data, generate thread fluctuation performance information, behavior difference value, advertisement acceptance, advertisement induced search probability, advertisement effect evaluation value, advertisement audience information, advertisement audience portrait and user delivery effect threshold by arranging and analyzing the data of the data collection module and the feature extraction module, and transmit the data to the intelligent monitoring module and the effect display module.

[0051] An intelligent monitoring module is configured to compare the advertisement effect evaluation value transmitted from the data analysis module with the user delivery effect threshold, monitor whether the advertisement effect evaluation value meets the standard, and transmit the monitoring result to the effect display module.

[0052] An effect display module is configured to arrange and analyze the received data, and display the advertisement monitoring result.

[0053] Preferably, the data collection module specifically comprises:

[0054] A first collection unit is configured to be connected with the cloud video platform server, and acquire thread information from the cloud video platform server.

[0055] A second collection unit is configured to be connected with the advertisement party, and acquire advertisement information from the advertisement party.

[0056] Preferably, the feature extraction module specifically comprises:

[0057] A first extraction unit is configured to extract features from the received advertisement information, generate advertisement product content, advertisement duration, and advertisement feature information according to the advertisement product content.

[0058] The second extraction unit is used for data feature extraction of the received user ID registration information according to age and region respectively, and generates an advertisement audience portrait.

[0059] Preferably, the data analysis module specifically comprises:

[0060] The first analysis unit is used for analyzing the received thread fluctuation performance information, and generating execution state fluctuation performance information, thread quantity fluctuation performance information and resource occupation fluctuation performance information.

[0061] The second analysis unit is used for analyzing resource occupation fluctuation time according to the advertisement duration, and determining an advertisement time period.

[0062] The third analysis unit is used for taking the execution state fluctuation frequency difference as a reference variable, analyzing the thread quantity fluctuation frequency difference and the resource occupation fluctuation frequency difference, and respectively acquiring a thread quantity fluctuation correlation coefficient and a resource occupation fluctuation correlation coefficient.

[0063] The fourth analysis unit is used for analyzing the advertisement receptivity matrix and the user influence state transition probability matrix through a specific advertisement-induced search probability calculation formula, and generating an advertisement-induced search probability.

[0064] Compared with the prior art, the present application has the following beneficial effects:

[0065] The present application proposes an advertisement delivery effect intelligent monitoring scheme based on cloud video data analysis, generates advertisement feature information and user behavior information through cloud video platform server data, acquires a behavior difference value based on the user behavior information, generates advertisement receptivity in combination with the advertisement feature information, generates an advertisement-induced search probability through advertisement delivery historical data, generates an advertisement effect evaluation value and an advertisement audience portrait in combination with the advertisement receptivity, the behavior difference value and the advertisement-induced search probability, analyzes the advertisement-induced search probability and the advertisement audience portrait according to the advertisement delivery historical data, sets a user delivery effect threshold value, analyzes the advertisement effect evaluation value based on the user delivery effect threshold value, judges whether the advertisement effect evaluation value is greater than the user delivery effect threshold value, and through this way, the behavior information of the user can be effectively acquired according to the thread data fluctuation condition in the cloud video platform server, the advertisement receptivity is generated through the difference between the behavior information before and after the user browses the advertisement, the behavior motive of the user is analyzed in combination with the advertisement delivery historical data, the evaluation result of the advertisement effect is more accurate, the difference between the monitoring result and the actual advertisement delivery effect is reduced, and it is ensured that the advertisement party can more clearly understand the advertisement delivery effect. BRIEF DESCRIPTION OF DRAWINGS

[0066] Figure 1The flow chart of the advertisement putting effect intelligent monitoring method based on cloud video data analysis provided by the present application is shown in the figure.

[0067] Figure 2 The flow chart of the step of generating advertisement feature information and user behavior information in the present application is shown in the figure.

[0068] Figure 3 The flow chart of the step of obtaining behavior difference value and generating advertisement acceptance in the present application is shown in the figure.

[0069] Figure 4 The flow chart of the step of generating advertisement induced search probability in the present application is shown in the figure.

[0070] Figure 5 The flow chart of the step of generating advertisement effect evaluation value and advertisement audience information in the present application is shown in the figure.

[0071] Figure 6 The flow chart of the step of setting user putting effect threshold in the present application is shown in the figure.

[0072] Figure 7 The structural diagram of the advertisement putting effect intelligent monitoring system based on cloud video data analysis provided by the present application is shown in the figure. DETAILED DESCRIPTION

[0073] The following description is used to disclose the present application so that those skilled in the art can implement the present application. The preferred embodiments in the following description are only used as examples, and other obvious modifications can be thought by those skilled in the art.

[0074] REFERENCE Figure 1 As shown in the figure, an advertisement putting effect intelligent monitoring method based on cloud video data analysis comprises:

[0075] Obtaining and analyzing cloud video platform server data to generate advertisement feature information and user behavior information;

[0076] Based on the user behavior information, obtaining behavior difference value, combining with the advertisement feature information, generating advertisement acceptance;

[0077] Based on the cloud video platform, obtaining advertisement putting history data to generate advertisement induced search probability;

[0078] According to the advertisement acceptance, the behavior difference value and the advertisement induced search probability, generating the advertisement effect evaluation value and the advertisement audience information;

[0079] Classifying the advertisement audience information to generate advertisement audience portrait;

[0080] According to the advertisement putting history data, analyzing the advertisement induced search probability and the advertisement audience portrait to set the user putting effect threshold;

[0081] Based on the user delivery effect threshold, the advertisement effect evaluation value is analyzed to determine whether the advertisement effect evaluation value is greater than the user delivery effect threshold, if yes, the advertisement content is optimized according to the advertisement audience portrait, the freshness and attention of the audience to the advertisement are maintained, if not, the advertisement delivery is adjusted according to the advertisement audience portrait.

[0082] The scheme generates advertisement feature information and user behavior information through cloud video platform server data, obtains a behavior difference value based on the user behavior information, generates an advertisement acceptance degree in combination with the advertisement feature information, generates an advertisement-induced search probability through advertisement delivery historical data, generates an advertisement effect evaluation value and an advertisement audience portrait in combination with the advertisement acceptance degree, the behavior difference value and the advertisement-induced search probability, analyzes the advertisement-induced search probability and the advertisement audience portrait according to the advertisement delivery historical data, sets a user delivery effect threshold, analyzes the advertisement effect evaluation value based on the user delivery effect threshold, and determines whether the advertisement effect evaluation value is greater than the user delivery effect threshold. Through this way, the behavior information of the user can be effectively obtained according to the thread data fluctuation of the cloud video platform server, the advertisement acceptance degree is generated through the difference between the behavior information before and after the user browses the advertisement, the behavior motivation of the user is analyzed in combination with the advertisement delivery historical data, and the evaluation result of the advertisement effect is more accurate.

[0083] Referring to Figure 2 The advertisement feature information and the user behavior information are generated, and specifically include:

[0084] The cloud video platform server data is processed to obtain server thread occupation and advertisement information, the advertisement information including advertisement product content and advertisement duration;

[0085] The advertisement feature information is generated according to the advertisement product content;

[0086] The login ID of the user on the cloud video platform is obtained, and the login ID of the user is bound with the server thread based on the cloud video server data to obtain the user associated thread;

[0087] The server thread occupation is analyzed according to the user associated thread to obtain thread fluctuation performance information, the thread fluctuation performance information including execution state fluctuation performance information, thread number fluctuation performance information and resource occupation fluctuation performance information;

[0088] The user behavior information is generated according to the thread fluctuation performance information;

[0089] The execution state fluctuation performance information of the user associated thread is used to determine whether the user pauses the video playback, the thread quantity fluctuation performance information of the user associated thread is used to determine whether the user interacts with the video, and the resource occupation fluctuation performance information of the user associated thread is used to determine whether the user performs a video download operation.

[0090] It can be understood that the cloud video platform server data includes cloud video server data and advertising information implanted by an advertising party in the cloud video platform server, the advertising content is extracted from the product type, product spokesperson, product function and product advantage, the advertising feature information is generated, the user sends a login request to the cloud video platform server when logging in the cloud video platform, the cloud video platform server generates an independent data space according to the network communication address of the user, which is used to record the thread data of the user in the cloud video platform, the thread data corresponds to the operation behavior of the user, the login ID of the user in the cloud video platform is bound with the network communication address, so that the user associated thread can be obtained, the thread fluctuation performance information and the user behavior maintain a one-to-one corresponding relationship, the execution state of the thread includes start, pause, resume and stop, the start indicates that the user opens a new video, the pause indicates that the user performs a pause operation on the video being watched, the resume indicates that the user continues to perform a play operation on the paused video, and the stop indicates that the user closes the video, the increase of the thread quantity indicates that the user interacts with the video content under the condition that the current video is normally played, and the increase of the thread resource occupation indicates that the user performs a video download operation.

[0091] Referring to Figure 3 The behavior difference value is obtained, the advertising acceptance is generated in combination with the advertising feature information, and specifically includes:

[0092] The thread fluctuation time is obtained according to the cloud video platform server data and the thread fluctuation performance information;

[0093] The resource occupation fluctuation time is analyzed according to the advertising duration, and the advertising time period is determined;

[0094] The thread fluctuation time is divided according to the advertising time period, and the thread occupation information before the advertising and the thread occupation information after the advertising are generated in combination with the server thread occupation;

[0095] The execution state fluctuation frequency information, the thread quantity fluctuation frequency information and the resource occupation fluctuation frequency information are obtained by analyzing the execution state fluctuation performance information, the thread quantity fluctuation performance information and the resource occupation fluctuation performance information in a three-minute time period;

[0096] The thread occupation information before the advertisement period and the thread occupation information after the advertisement period are analyzed from the aspects of execution state, thread quantity and resource occupation fluctuation, respectively, to generate execution state fluctuation frequency difference, thread quantity fluctuation frequency difference and resource occupation fluctuation frequency difference;

[0097] Based on the Pearson correlation coefficient calculation formula, the thread quantity fluctuation frequency difference and the resource occupation fluctuation frequency difference are analyzed with the execution state fluctuation frequency difference as a reference variable to obtain thread quantity fluctuation correlation coefficient and resource occupation fluctuation correlation coefficient respectively.

[0098] The behavior difference value is obtained according to the ratio of the thread quantity fluctuation correlation coefficient and the resource occupation fluctuation correlation coefficient.

[0099] The advertisement perception table is generated according to the advertisement delivery historical data.

[0100] The standard difference value is obtained by substituting the advertisement feature information into the advertisement perception table, and the advertisement acceptance is generated in combination with the behavior difference value, and the advertisement acceptance state transition probability of different behavior difference values is calculated to establish the advertisement acceptance matrix.

[0101] It can be understood that when the thread information fluctuates, the cloud video platform server will record the corresponding time, so by arranging the cloud video platform server data through the thread fluctuation performance information, the thread fluctuation time can be obtained. When the advertisement is played in the video, the thread execution state becomes paused, the thread quantity increases by one thread, and the resource occupation remains in an increased steady state for the same length of time as the advertisement duration. By searching the cloud video platform server data through these feature information, the advertisement period can be determined. The thread quantity fluctuation correlation coefficient is the correlation coefficient of the execution state fluctuation frequency difference and the thread quantity fluctuation frequency difference, and the resource occupation fluctuation correlation coefficient is the correlation coefficient of the execution state fluctuation frequency difference and the resource occupation fluctuation frequency difference. According to the advertisement delivery historical data, the historical behavior difference values corresponding to different product types, product endorsers, product functions and product advantages can be obtained. By the method of controlling variables, the product types, product endorsers, product functions and product advantages are compared in turn to obtain the historical average behavior difference values corresponding to the single feature information of the product type, product endorser, product function and product advantage information. The historical average behavior difference values of the single feature information are statistically integrated to generate the advertisement perception table. The advertisement acceptance is divided into high acceptance, medium acceptance and low acceptance. There is a corresponding relationship between the behavior difference value and the advertisement acceptance, but there are individual differences among users, and the behavior difference value of the user may shift the advertisement acceptance. By calculating the proportion of the user whose behavior difference value corresponds to the advertisement acceptance in the total number of historical data, the advertisement acceptance state transition probability of different behavior difference values is determined, and the advertisement acceptance matrix is established.

[0102] Referring to Figure 4 As shown in the figure, the advertisement-induced search probability is generated, specifically including:

[0103] The advertisement delivery history data is sorted and divided according to the advertisement acceptance, and the user influence state information is generated;

[0104] According to the user influence state information, the advertisement delivery history data is analyzed, the user state probability of different acceptance is calculated, and the user influence state transition probability matrix is generated;

[0105] The advertisement acceptance matrix and the user influence state transition probability matrix are analyzed, and the advertisement-induced search probability is generated;

[0106] The specific calculation formula of the advertisement-induced search probability is:

[0107] P[s k+1 |s k ]=P[s1|s2,s3,s4]P[v|v1,v2,v3];

[0108] In the formula, P[s k+1 |s k ] is the advertisement-induced search probability, s k+1 is the k+1 user influence state, s k is the k user influence state, s1, s2, s3, s4 is the user influence state, v is the current advertisement acceptance state, v1, v2, v3 is the advertisement acceptance state, P[s1|s2, s3, s4] is the user influence state transition probability, and P[v|v1, v2, v3] is the advertisement acceptance state transition probability.

[0109] It can be understood that the user influence state is divided into the user not being affected by the advertisement, the user having certain interest in the advertisement but not triggering the search, the user clicking the advertisement and starting the search, and the user leaving the search page without completing the search. The advertisement delivery history data is sorted according to the advertisement acceptance, the user influence state corresponding to different advertisement acceptance can be obtained, the number of user influence state transition corresponding to the advertisement acceptance in the historical data is calculated by sorting and analyzing the advertisement delivery history data, the user state probability of different acceptance is determined, and the user influence state transition probability matrix is established.

[0110] Referring to Figure 5 As shown in the figure, the advertisement effect evaluation value and the advertisement audience information are generated, specifically including:

[0111] The registration information of the user login ID is obtained, the user information is classified in combination with the advertisement acceptance, the average advertisement acceptance of different user groups is obtained, and the advertisement audience information is generated;

[0112] Obtaining the total number of advertisement browsing, combining the advertisement audience information, determining the distribution of different user groups in the total number of advertisement browsing;

[0113] Weighted sum of the average advertisement acceptance of each user group and the distribution of each user group in the total number of advertisement browsing, generating an advertisement effect evaluation value.

[0114] It can be understood that the registration information of the user login ID includes user age information and user region information, according to the registration information of the user login ID, the users are classified according to age and region, different age group users and different regional group users are generated, different age groups are divided into intervals of ten years, and different regional group users are divided into intervals of cities, according to the advertisement acceptance of different users in the advertisement delivery history data, the average advertisement acceptance of each user group is calculated, the average advertisement acceptance of each user group is compared and arranged in order from high to low, and the advertisement audience information is generated.

[0115] Referring to Figure 6 , a user delivery effect threshold is set, specifically including:

[0116] According to the advertisement audience portrait, the same type of advertisement delivery history data is classified according to age information and regional information, and age delivery effect data and regional delivery effect data are generated;

[0117] The age delivery effect data and the regional delivery effect data are analyzed respectively, and the age delivery effect peak data, the age delivery effect valley data, the regional delivery effect peak data and the regional delivery effect valley data are obtained;

[0118] The age delivery effect peak data and the age delivery effect valley data are averaged, and the age delivery effect threshold is generated, and the regional delivery effect peak data and the regional delivery effect valley data are averaged, and the regional delivery effect threshold is generated;

[0119] According to the registration information of the advertisement browsing user, the corresponding age delivery effect threshold and regional delivery effect threshold are obtained, and the average value is calculated, and the user delivery effect threshold is generated.

[0120] Further, referring to Figure 7 , an advertisement delivery effect intelligent monitoring system based on cloud video data analysis is proposed, which is used to realize the above-mentioned monitoring method, and the specific steps are as follows:

[0121] The data collection module is configured to be connected with the cloud video platform server, acquire data of the cloud video platform server, acquire advertising information from an advertising party, and transmit the collected information data to the feature extraction module and the data analysis module.

[0122] The feature extraction module is configured to extract features from the received data according to requirements, generate advertising feature information according to advertising product content, extract age information and regional information of a user according to registration information of a user login ID, and transmit the extracted feature information to the data analysis module.

[0123] The data analysis module is configured to analyze the received data, generate thread fluctuation performance information, behavior difference value, advertising acceptance, advertising induced search probability, advertising effect evaluation value, advertising audience information, advertising audience portrait and user delivery effect threshold by collating and analyzing the data of the data collection module and the feature extraction module, and transmit the data to the intelligent monitoring module and the effect display module.

[0124] The intelligent monitoring module is configured to compare the advertising effect evaluation value transmitted by the data analysis module with the user delivery effect threshold, monitor whether the advertising effect evaluation value meets the standard, and transmit the monitoring result to the effect display module.

[0125] The effect display module is configured to analyze and collate the received data, and display the advertising monitoring result.

[0126] Specifically, the data collection module comprises:

[0127] The first collection unit is configured to be connected with the cloud video platform server, acquire thread information from the cloud video platform server.

[0128] The second collection unit is configured to be connected with the advertising party, and acquire advertising information from the advertising party.

[0129] Specifically, the feature extraction module comprises:

[0130] The first extraction unit is configured to extract features from the received advertising information, generate advertising product content, advertising duration, and advertising feature information according to the advertising product content.

[0131] The second extraction unit is configured to extract data features from the received user ID registration information according to age and region respectively, and generate an advertising audience portrait.

[0132] Specifically, the data analysis module comprises:

[0133] The first analysis unit is used for analyzing the received thread fluctuation performance information to generate execution state fluctuation performance information, thread quantity fluctuation performance information and resource occupation fluctuation performance information.

[0134] The second analysis unit is used for analyzing resource occupation fluctuation time according to advertisement time length to determine an advertisement time period.

[0135] The third analysis unit is used for taking the execution state fluctuation frequency difference as a reference variable, analyzing the thread quantity fluctuation frequency difference and the resource occupation fluctuation frequency difference, and respectively obtaining a thread quantity fluctuation correlation coefficient and a resource occupation fluctuation correlation coefficient.

[0136] The fourth analysis unit is used for analyzing an advertisement receptivity matrix and a user influence state transition probability matrix through an advertisement-induced search probability specific calculation formula to generate an advertisement-induced search probability.

[0137] In summary, the advantages of the present application are as follows:

[0138] The user's behavior information can be effectively obtained according to the thread data fluctuation in the cloud video platform server, the evaluation result of the advertisement effect is more accurate by analyzing the user's behavior motivation, the difference between the monitoring result and the actual advertisement effect is reduced, and the advertisement party can more clearly understand the advertisement effect.

[0139] The basic principles, main features and advantages of the present application are shown and described above. It should be understood by those skilled in the art that the present application is not limited by the above examples, and the above examples and descriptions in the specification are only the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection required by the present application is defined by the appended claims and their equivalents.

Claims

1. An intelligent monitoring method for advertising delivery effects based on cloud video data analysis, characterized in that: include: Obtain and analyze cloud video platform server data to generate advertising feature information and user behavior information; Based on user behavior information, obtain behavior difference values, combine with advertising feature information, and generate advertising acceptance; Based on the cloud video platform, historical advertising data is obtained to generate the probability of advertising-induced searches; Generate advertising effect evaluation value and advertising audience information based on advertising acceptance, behavioral difference value and advertising-induced search probability; Classify advertising audience information and generate advertising audience portraits; Based on historical advertising data, analyze the probability of ad-induced searches and the profile of the advertising audience, and set user delivery effect thresholds; Based on the user delivery effect threshold, the advertising effect evaluation value is analyzed to determine whether the advertising effect evaluation value is greater than the user delivery effect threshold. If so, the advertising content is optimized according to the advertising audience profile to maintain the audience's freshness and attention to the advertisement. If not, the advertising delivery is adjusted according to the advertising audience profile; The step of obtaining the behavior difference value and combining it with the advertisement feature information to generate the advertisement acceptance specifically includes: Obtain thread fluctuation time based on cloud video platform server data and thread fluctuation performance information; Analyze resource usage fluctuations based on ad duration and determine ad time slots; Divide the thread fluctuation time according to the advertising period, and generate the thread occupancy information of the pre-ad period and the post-ad period based on the server thread occupancy status; Analyze the execution status fluctuation performance information, thread number fluctuation performance information, and resource usage fluctuation performance information in a three-minute period to obtain execution status fluctuation frequency information, thread number fluctuation frequency information, and resource usage fluctuation frequency information; Analyze the thread occupancy information before and after the commercials from the perspectives of execution status, thread number, and resource occupancy fluctuations, generating execution status fluctuation frequency difference, thread number fluctuation frequency difference, and resource occupancy fluctuation frequency difference, respectively. Based on the Pearson correlation coefficient calculation formula, the execution state fluctuation frequency difference is used as the benchmark variable. The thread number fluctuation frequency difference and resource usage fluctuation frequency difference are analyzed to obtain the thread number fluctuation correlation coefficient and resource usage fluctuation correlation coefficient respectively. Obtain the behavior difference value based on the ratio of the thread number fluctuation correlation coefficient and the resource usage fluctuation correlation coefficient; Generate an advertising impression table based on historical advertising data; Substitute the advertising feature information into the advertising perception table to obtain the standard difference value, combine it with the behavioral difference value, generate advertising acceptance, and calculate the advertising acceptance state transition probability of different behavioral difference values ​​to establish an advertising acceptance matrix.

2. The method for intelligently monitoring advertising delivery effects based on cloud video data analysis according to claim 1, characterized in that: The generating of advertisement feature information and user behavior information specifically includes: Processing cloud video platform server data to obtain server thread occupancy and advertising information, including advertising product content and advertising duration; Generate advertising feature information according to the content of the advertising product; Obtain the user's login ID on the cloud video platform, bind the user's login ID to the server thread based on the cloud video server data, and obtain the user-associated thread; Analyze the server thread occupancy according to the user-associated threads to obtain thread fluctuation performance information, including execution state fluctuation performance information, thread number fluctuation performance information, and resource occupancy fluctuation performance information; Generate user behavior information based on thread fluctuation performance information; Through the execution status fluctuation performance information of the user-associated thread, it is determined whether the user has paused the video playback; through the thread number fluctuation performance information of the user-associated thread, it is determined whether the user has interacted with the video; through the resource usage fluctuation performance information of the user-associated thread, it is determined whether the user has downloaded the video.

3. The method for intelligently monitoring advertising delivery effects based on cloud video data analysis according to claim 2, characterized in that: Generating the advertisement-induced search probability specifically includes: Organize historical advertising data, divide it by advertising acceptance, and generate user influence status information; Based on user influence status information, historical advertising data is analyzed to calculate the user status probabilities of different acceptance levels and generate a user influence status transition probability matrix; Analyze the advertising acceptance matrix and the user influence state transition probability matrix to generate the advertising-induced search probability; The specific calculation formula for the probability of advertisement-induced search is: ； Where, The probability of induced search for advertisements, For the User influence status, For the User influence status, Affects the state for the user, is the current advertising acceptance status, is the advertising acceptance status, The user influences the state transition probability, is the state transition probability of advertising acceptance.

4. The method for intelligently monitoring advertising delivery effects based on cloud video data analysis according to claim 3, characterized in that: The generating of the advertising effect evaluation value and the advertising audience information specifically includes: Obtain the registration information of the user login ID, classify the user information based on the advertising acceptance, obtain the average advertising acceptance of different user groups, and generate advertising audience information; Obtain the total number of ad viewers and, combined with ad audience information, determine the distribution of different user groups within the total number of ad viewers; The advertising effect evaluation value is generated by taking the weighted sum of the advertising-induced search probability corresponding to the average advertising acceptance of each user group and the distribution of each user group in the total number of ad viewers.

5. The method for intelligently monitoring advertising delivery effects based on cloud video data analysis according to claim 4, characterized in that: The setting of user delivery effect threshold specifically includes: Based on the advertising audience portrait, historical data of the same type of advertising is classified by age information and region information to generate age advertising effect data and region advertising effect data; Organize and analyze the age-based delivery effect data and the regional delivery effect data respectively to obtain the age-based delivery effect peak data, age-based delivery effect valley data, regional delivery effect peak data, and regional delivery effect valley data; The age-based delivery effect peak data and age-based delivery effect valley data are averaged and summed to generate the age-based delivery effect threshold. The region-based delivery effect peak data and region-based delivery effect valley data are averaged and summed to generate the region-based delivery effect threshold. According to the registration information of the ad browsing user, the corresponding age delivery effect threshold and regional delivery effect threshold are obtained, and the average value is calculated to generate the user delivery effect threshold.

6. An intelligent monitoring system for advertising delivery effects based on cloud video data analysis, applicable to the monitoring method according to any one of claims 1 to 5, characterized in that: Specifically include: A data collection module is used to establish a data connection with the cloud video platform server, obtain data from the cloud video platform server, obtain advertising information from advertisers, and transmit the collected information data to the feature extraction module and the data analysis module; A feature extraction module is used to extract features from the received data according to the required requirements, generate advertising feature information based on the content of the advertising product, extract the user's age information and regional information based on the registration information of the user login ID, and transmit the extracted feature information to the data analysis module; A data analysis module, which is used to analyze and process the received data. By collating and analyzing the data from the data collection module and the feature extraction module, the module generates thread fluctuation performance information, behavior difference values, advertising acceptance, advertising-induced search probability, advertising effect evaluation values, advertising audience information, advertising audience portraits, and user delivery effect thresholds, and transmits the data to the intelligent monitoring module and the effect display module; An intelligent monitoring module, which is used to compare the advertising effect evaluation value transmitted by the data analysis module with the user delivery effect threshold, monitor whether the advertising effect evaluation value meets the standard, and transmit the monitoring result to the effect display module; The effect display module is used to analyze and organize the received data and display the advertising monitoring results.

7. The intelligent monitoring system for advertising delivery effects based on cloud video data analysis according to claim 6 is characterized in that: The data collection module specifically includes: A first collecting unit, configured to establish a data connection with a cloud video platform server and obtain thread information from the cloud video platform server; The second collecting unit is used to establish a communication connection with the advertiser and obtain advertisement information from the advertiser.

8. The intelligent monitoring system for advertising delivery effects based on cloud video data analysis according to claim 6 is characterized in that: The feature extraction module specifically includes: a first extraction unit configured to extract features from the received advertising information, generate advertising product content and advertising duration, and generate advertising feature information based on the advertising product content; The second extraction unit is used to extract data features of the received user ID registration information according to age and region to generate an advertising audience portrait.

9. The intelligent monitoring system for advertising delivery effects based on cloud video data analysis according to claim 6 is characterized in that: The data analysis module specifically includes: a first analyzing unit configured to analyze received thread fluctuation performance information to generate execution state fluctuation performance information, thread quantity fluctuation performance information, and resource occupancy fluctuation performance information; a second analyzing unit, configured to analyze resource occupancy fluctuation time according to advertisement duration to determine an advertisement period; a third analyzing unit configured to analyze the thread quantity fluctuation frequency difference and the resource occupancy fluctuation frequency difference using the execution state fluctuation frequency difference as a benchmark variable, and obtain a thread quantity fluctuation correlation coefficient and a resource occupancy fluctuation correlation coefficient respectively; The fourth analysis unit, the third analysis unit is used to analyze the advertisement acceptance matrix and the user influence state transition probability matrix through the specific calculation formula of the advertisement-induced search probability, and generate the advertisement-induced search probability.

Images