Audience prediction system and method of child video PK operation mechanism

Abstract

The invention relates to the technical field of video audience prediction, and particularly discloses an audience prediction system and method of a child video PK operation mechanism. Comprising a child video number acquisition module, a child video feature information extraction module, a child video feature information analysis module, a child video quality analysis and evaluation module, a child video audience prediction module, a child video prediction optimization module and a child video prediction result output module. The method comprises the following steps: acquiring and numbering each to-be-predicted child video, extracting a video feature data value and an audience behavior data value, analyzing a feature correlation coefficient and evaluating video quality, establishing an audience prediction model, performing audience prediction scoring on each child video, and establishing a prediction optimization index model. The audience prediction score of the child video is optimized; by introducing a big data analysis technology, accurate prediction and optimization of child video audiences are realized, and powerful support is provided for content recommendation and PK strategies of a platform.

Description

Technical Field

[0001] The present invention relates to the technical field of video audience prediction, and in particular to an audience prediction system and method for the operation mechanism of children's video PK. Background Art

[0002] With the rapid development of Internet technology, video content has become an important way for children to obtain information and entertainment. In recent years, video content designed specifically for children has emerged in an endless stream in the market and is deeply loved by parents and children. In order to improve the user experience and the diversity of content, many platforms have introduced a children's video PK mechanism, that is, multiple children's videos compete within the same time period, and the winner is determined according to the interaction of the audience. This mechanism not only stimulates the enthusiasm of creators but also brings more choices and a sense of participation to the audience.

[0003] However, the audience prediction system of the traditional children's video PK operation mechanism has many deficiencies. First of all, the traditional audience prediction system often relies on simple statistical analysis and artificial experience, and it is difficult to accurately capture the complexity and diversity of audience behavior, resulting in inaccurate prediction results. Secondly, the traditional audience prediction system does not conduct in-depth research on the correlation between video feature data and audience behavior data, lacks comprehensive analysis of these factors, and is difficult to accurately predict the selection tendency of the audience in the PK scenario. In addition, the traditional audience prediction system mostly adopts prediction models based on statistics or simple rules, lacks flexibility and adaptability, is difficult to capture the complexity of audience behavior and the diversity of video content, and lacks a continuous optimization and update mechanism, and cannot optimize the audience prediction model, making it difficult to adapt to market changes and new trends in audience behavior, reducing the accuracy of prediction. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide an audience prediction system and method for the operation mechanism of children's video PK to solve the problems raised in the above background art.

[0005] To achieve the above object, the present invention provides the following technical solution: An audience prediction system for the operation mechanism of children's video PK, including: a children's video number acquisition module, a children's video feature information extraction module, a children's video feature information analysis module, a children's video quality analysis and evaluation module, a children's video audience prediction module, a children's video prediction optimization module, and a children's video prediction result output module.

[0006] The children's video number acquisition module: used to acquire the works of children's video PK, mark them as each to-be-predicted children's video, and perform numbering processing on each to-be-predicted children's video, and mark them as 1, 2,..., i,..., n in sequence, where i is the number of each to-be-predicted children's video; Children's Video Feature Information Extraction Module: Extract respective video feature data values and respective viewer behavior data values corresponding to each to-be-predicted children's video from each to-be-predicted children's video; Children's Video Feature Information Analysis Module: Based on the respective video feature data values and respective viewer behavior data values corresponding to each to-be-predicted children's video, analyze to obtain the feature correlation coefficients of each to-be-predicted children's video, and conduct a correlation assessment on the children's video features, outputting the correlation assessment results; Children's Video Quality Analysis and Evaluation Module: Based on the color saturation, resolution, and volume among the respective video feature data values corresponding to each to-be-predicted children's video, analyze to obtain the quality evaluation coefficients of each to-be-predicted children's video; Children's Video Viewer Prediction Module: Based on the feature correlation coefficients, quality evaluation coefficients, and click-through rates of each to-be-predicted children's video, establish a viewer prediction model for the children's video PK operation mechanism, and analyze to obtain the viewer prediction scores of each to-be-predicted children's video; Children's Video Prediction Optimization Module: Based on the viewer prediction scores of each to-be-predicted children's video, establish a prediction optimization index model, and analyze to obtain the prediction optimization indices of each to-be-predicted children's video; Children's Video Prediction Result Output Module: Optimize the viewer prediction scores of the children's videos based on the prediction optimization indices of each to-be-predicted children's video, and output the viewer prediction results of each to-be-predicted children's video.

[0007] Preferably, the specific implementation manner of the Children's Video Feature Information Extraction Module is as follows: Extract respective video feature data values and respective viewer behavior data values corresponding to each to-be-predicted children's video from each to-be-predicted children's video; Obtain the respective video feature data values corresponding to each to-be-predicted children's video , v representing the numbers of the respective video feature data values, v = 1, 2,... c where i represents the number of each to-be-predicted children's video, i = 1, 2,... n ; among them, the respective video feature data values include but are not limited to the color saturation, resolution, clarity, frame rate, average brightness, lens switching frequency, picture clarity, and volume of the picture; the video feature data value vector corresponding to the i-th to-be-predicted children's video is: ; Obtain the respective viewer behavior data values corresponding to each to-be-predicted children's video , p representing the numbers of the respective viewer behavior data values, p = 1, 2,... k; Among them, each of the audience behavior data values includes, but is not limited to, the view count, the number of likes, the number of comments, the number of shares, the viewing duration, and the number of repeated viewings; the vector of audience behavior data values corresponding to the i-th child video to be predicted is: .

[0008] Preferably, the specific execution manner of the child video feature information analysis module is as follows: Calculate the feature correlation coefficients of each child video to be predicted. The specific calculation formula is as follows: , where represents the feature correlation coefficient of the i -th child video to be predicted, i represents the number of each child video to be predicted, represents the i -th video feature data value corresponding to the v -th child video to be predicted, represents the i -th preset value of video feature data corresponding to the v -th child video to be predicted, v represents the number of each video feature data value, c represents the total number of video feature data values, represents the i -th audience behavior data value corresponding to the p -th child video to be predicted, represents the i -th preset value of audience behavior data corresponding to the p -th child video to be predicted, p represents the number of each audience behavior data value, k represents the total number of audience behavior data values, represents a very small positive number.

[0009] Preferably, the specific execution manner of the child video quality analysis and evaluation module is as follows: Extract the color saturation, resolution, and volume from each video feature data value corresponding to each child video to be predicted, and mark them as , , ; At the same time, extract the standard value of color saturation corresponding to the child video from the management database. From the formula , analyze and obtain the color saturation deviation SD i corresponding to the i-th child video to be predicted; Calculate the quality evaluation coefficient of each child video to be predicted. The specific calculation formula is as follows: , whereQEC i Represents the quality evaluation coefficient of the i-th children's video to be predicted. SD max Represents the maximum preset color saturation deviation. Represents the volume of the i-th children's video to be predicted. Represents the volume standard value corresponding to the children's video. Represents the resolution of the i-th children's video to be predicted. e Represents the natural constant.

[0010] Preferably, the specific implementation content of the children's video audience prediction module is as follows: Obtain the click count i of the i-th children's video to be predicted Cc i and the display count Df i , and substitute them into the formula respectively to obtain the click-through rate i of the i-th children's video to be predicted. Ctr i ; Establish an audience prediction model for the children's video PK operation mechanism. The specific prediction model is as follows: , where PS i represents the audience prediction score of the i-th children's video to be predicted, represents the weight value v of the j-th video feature data, represents the feature correlation coefficient i of the i-th children's video to be predicted, QEC i represents the quality evaluation coefficient of the i-th children's video to be predicted.

[0011] Preferably, the method for obtaining the weight value v of the j-th video feature data is specifically as follows: Step 1: Extract the values of each video feature data corresponding to each children's video to be predicted , and perform standardization processing on each video feature data value through linear standardization; Step 2: Substitute the values of each video feature data corresponding to each children's video to be predicted into the formula to obtain the proportion of each video feature data value in each children's video to be predicted, where i is the number of each children's video to be predicted, i = 1, 2, 3,... n, v represents the number of each video feature data value, v= 1, 2, ... c ; Step 3, calculate the entropy value of each video feature data: ; Step 4: Calculate the weight value of each video feature data : .

[0012] Preferably, the specific execution method of the children's video prediction optimization module is as follows: Based on the audience prediction scores of each children's video to be predicted, obtain the true values of the audience prediction scores corresponding to each historical prediction for each children's video to be predicted and the predicted values ; Define the loss function , where RMSE i The loss function of the i-th children's video to be predicted, h represents the total number of historical predictions, u represents the number of the historical prediction times, u = 1, 2, ... h , 、 respectively represent the true value and the predicted value of the i-th children's video to be predicted corresponding to the j-th historical prediction; i The i-th u times of historical prediction; Establish a prediction optimization index model, and the specific model is: , where POI i The prediction optimization index of the i-th children's video to be predicted.

[0013] Preferably, the specific execution method of the children's video prediction result output module is as follows: Obtain the prediction optimization index of the i-th children's video to be predicted POI i , optimize the audience prediction scores of the children's videos, and obtain the optimized output values of the audience prediction scores of each children's video to be predicted , and output the results to the mobile devices of the management personnel.

[0014] To achieve the above object, the present invention provides the following technical solutions: A method for predicting the audience of a children's video PK operation mechanism, implementing the above-mentioned system for predicting the audience of a children's video PK operation mechanism, including the following steps: S1: Obtain the children's video numbers: Obtain the works of the children's video PK, mark them as each children's video to be predicted, and number each children's video to be predicted, and mark them as 1, 2, … i …, n in sequence, where i is the number of each children's video to be predicted; S2: Extraction of children's video feature information: Extract the respective video feature data values and respective viewer behavior data values corresponding to each to-be-predicted children's video from each to-be-predicted children's video; S3: Analysis of children's video feature information: Based on the respective video feature data values and respective viewer behavior data values corresponding to each to-be-predicted children's video, analyze to obtain the feature correlation coefficients of each to-be-predicted children's video, and conduct a correlation assessment on the children's video features, and output the correlation assessment results; S4: Analysis and evaluation of children's video quality: Based on the color saturation, resolution, and volume in the respective video feature data values corresponding to each to-be-predicted children's video, analyze to obtain the quality evaluation coefficients of each to-be-predicted children's video; S5: Prediction of children's video viewers: Based on the feature correlation coefficients, quality evaluation coefficients, and click-through rates of each to-be-predicted children's video, establish a viewer prediction model for the children's video PK operation mechanism, and analyze to obtain the viewer prediction scores of each to-be-predicted children's video; S6: Prediction optimization of children's videos: Based on the viewer prediction scores of each to-be-predicted children's video, establish a prediction optimization index model, and analyze to obtain the prediction optimization indices of each to-be-predicted children's video; S7: Output of children's video prediction results: Optimize the viewer prediction scores of children's videos based on the prediction optimization indices of each to-be-predicted children's video, and output the viewer prediction results of each to-be-predicted children's video.

[0015] Technical effects and advantages of the present invention: 1. By numbering each to-be-predicted children's video, the present invention extracts the respective video feature data values and respective viewer behavior data values corresponding to each to-be-predicted children's video, analyzes to obtain the feature correlation coefficients of each to-be-predicted children's video, conducts a correlation assessment on the children's video features, and outputs the correlation assessment results. By analyzing the color saturation, resolution, and volume in the respective video feature data values corresponding to each to-be-predicted children's video, the present invention analyzes to obtain the quality evaluation coefficients of each to-be-predicted children's video. Based on the feature correlation coefficients, quality evaluation coefficients, and click-through rates of each to-be-predicted children's video, the present invention analyzes to obtain the viewer prediction scores of each to-be-predicted children's video. By establishing a prediction optimization index model, the present invention optimizes the viewer prediction scores of children's videos and outputs the viewer prediction results of each to-be-predicted children's video. By introducing big data analysis technology, the present invention realizes accurate prediction and optimization of children's video viewers, providing strong support for the content recommendation and PK strategies of the platform; 2. The present invention significantly improves the prediction accuracy of audience preferences in children's video PK by deeply mining video feature data values and audience behavior data values and applying advanced data analysis techniques, providing more reliable prediction results for the platform. At the same time, based on the accurate prediction results, the platform can optimize content strategies, adjust the children's video PK mechanism, and thus improve user retention rate and platform activity. In addition, the optimized prediction model significantly reduces manual intervention and remarkably improves operation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The present invention will be further described with reference to the accompanying drawings. However, the embodiments in the drawings do not constitute any limitation to the present invention. For those of ordinary skill in the art, other drawings can be obtained based on the following drawings without creative efforts.

[0017] Figure 1 It is a schematic structural diagram of an audience prediction system for a children's video PK operation mechanism of the present invention.

[0018] Figure 2 It is a schematic flowchart of an audience prediction method for a children's video PK operation mechanism of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0020] Embodiment 1 Please refer to Figure 1 As shown, the present invention provides an audience prediction system for a children's video PK operation mechanism, including: a children's video number acquisition module, a children's video feature information extraction module, a children's video feature information analysis module, a children's video quality analysis and evaluation module, a children's video audience prediction module, a children's video prediction optimization module, and a children's video prediction result output module; The children's video number acquisition module is connected to the children's video feature information extraction module, the children's video feature information extraction module is respectively connected to the children's video feature information analysis module and the children's video quality analysis and evaluation module, the children's video feature information extraction module is connected to the children's video audience prediction module, the children's video quality analysis and evaluation module is connected to the children's video audience prediction module, the children's video audience prediction module is connected to the children's video prediction optimization module, and the children's video prediction optimization module is connected to the children's video prediction result output module.

[0021] Children's Video Number Acquisition Module: It is used to obtain the works of children's video PK, mark them as each to-be-predicted children's video, and number each to-be-predicted children's video, successively marked as 1, 2, … i …, n, where i is the number of each to-be-predicted children's video; Children's Video Feature Information Extraction Module: Extract respective video feature data values and respective audience behavior data values corresponding to each to-be-predicted children's video from each to-be-predicted children's video; In this embodiment, it should be specifically noted that the specific execution manner of the Children's Video Feature Information Extraction Module is as follows: Extract respective video feature data values and respective audience behavior data values corresponding to each to-be-predicted children's video from each to-be-predicted children's video; Obtain respective video feature data values corresponding to each to-be-predicted children's video , v representing the numbers of respective video feature data values, v = 1, 2,... c , where i represents the number of each to-be-predicted children's video, i = 1, 2,... n ; among them, the respective video feature data values include but are not limited to the color saturation, resolution, clarity, frame rate, average brightness, lens switching frequency, picture clarity, and volume of the picture; the video feature data value vector corresponding to the i-th to-be-predicted children's video is: ; Obtain respective audience behavior data values corresponding to each to-be-predicted children's video , p representing the numbers of respective audience behavior data values, p = 1, 2,... k ; among them, the respective audience behavior data values include but are not limited to the view count, like count, comment count, share count, viewing duration, and repeated viewing times; the audience behavior data value vector corresponding to the i-th to-be-predicted children's video is: .

[0022] Children's Video Feature Information Analysis Module: Based on the respective video feature data values and respective audience behavior data values corresponding to each to-be-predicted children's video, analyze to obtain the feature correlation coefficients of each to-be-predicted children's video, and conduct a correlation assessment on the children's video features, and output the correlation assessment result; In this embodiment, it should be specifically noted that the specific execution manner of the Children's Video Feature Information Analysis Module is as follows: Calculate the feature correlation coefficients of each to-be-predicted children's video, and the specific calculation formula is as follows: , where, represents the iThe feature correlation coefficient of the children's video to be predicted, where i represents the number of each children's video to be predicted. represents the i th video feature data value corresponding to the v th children's video to be predicted. represents the i th video feature data preset value corresponding to the v th children's video to be predicted. v represents the number of each video feature data value. c represents the total number of video feature data values. represents the i th viewer behavior data value corresponding to the p th children's video to be predicted. represents the i th viewer behavior data preset value corresponding to the p th children's video to be predicted. p represents the number of each viewer behavior data value. k represents the total number of viewer behavior data values. represents an extremely small positive number, aiming to avoid the situation where the denominator is zero and ensure that the formula is meaningful in any case. In this embodiment, it should be specifically noted that if the i th video feature data value corresponding to the v th children's video to be predicted is color saturation, and the i th viewer behavior data value corresponding to the p th children's video to be predicted is the view count, and the feature correlation coefficient value of the i th children's video to be predicted analyzed is relatively large and close to 1, then it is determined that the correlation between the video feature data value of color saturation and the viewer behavior data value of view count is relatively high, indicating that the video feature data value of color saturation has a greater impact on viewer attraction.

[0023] Perform a correlation assessment on the children's video features, specifically: Extract the feature correlation coefficient i of the th children's video to be predicted. The value range is between -1 and 1. If the feature correlation coefficient i of the th children's video to be predicted is close to 1, then it is determined that the video feature data of the i th children's video to be predicted has a positive impact on the viewer behavior data. If the feature correlation coefficient i of the th children's video to be predicted is close to -1, then it is determined that the iThe video feature data of a child video to be predicted has a negative impact on the audience behavior data. If the i feature correlation coefficient of the th child video to be predicted i is close to 0, it is determined that the video feature data of the

[0024] th child video to be predicted has no impact on the audience behavior data. In this embodiment, it should be specifically noted that the specific execution method of the child video quality analysis and evaluation module is as follows: Extract the color saturation, resolution, and volume in the video feature data values corresponding to each child video to be predicted, and mark them as and ; At the same time, extract the standard value of the color saturation corresponding to the child video from the management database. From the formula , analyze and obtain the color saturation deviation SD i corresponding to the th child video to be predicted; Calculate the quality evaluation coefficient of each child video to be predicted. The specific calculation formula is as follows: QEC i represents the quality evaluation coefficient of the SD max th child video to be predicted, represents the maximum value of the preset color saturation deviation, represents the volume of the th child video to be predicted, e represents the standard value of the volume corresponding to the child video,

[0025] represents the resolution of the In this embodiment, it should be specifically noted that the specific execution content of the child video audience prediction module is as follows: i Obtain the click count Cc i and the display count Df i, substitute them into the formula respectively, and obtain the i click-through rate of the i-th children's video to be predicted Ctr i ; Establish an audience prediction model for the children's video PK operation mechanism. The specific prediction model is as follows: , where PS i represents the audience prediction score of the i-th children's video to be predicted, represents the weight value of the v -th video feature data, represents the feature correlation coefficient of the i -th children's video to be predicted, QEC i represents the quality evaluation coefficient of the i-th children's video to be predicted.

[0026] In this embodiment, it should be specifically noted that the method for obtaining the weight value v of the -th video feature data is specifically as follows: Step 1: Extract the values of each video feature data corresponding to each children's video to be predicted , and perform standardization processing on the values of each video feature data through linear standardization; Step 2: Substitute the values of each video feature data corresponding to each children's video to be predicted into the formula to obtain the proportion of the values of each video feature data in each children's video to be predicted , where i is the number of each children's video to be predicted, i = 1, 2, 3,... n, v represents the number of each video feature data value, v = 1, 2,... c ; Step 3: Calculate the entropy value of each video feature data: ; Step 4: Calculate the weight value of each video feature data: .

[0027] Children's video prediction optimization module: Based on the audience prediction scores of each children's video to be predicted, establish a prediction optimization index model, and analyze to obtain the prediction optimization index of each children's video to be predicted; In this embodiment, it should be specifically noted that the specific execution method of the children's video prediction optimization module is as follows: Based on the audience prediction scores of each children's video to be predicted, obtain the true values of the audience prediction scores of each historical prediction corresponding to each children's video to be predicted and predicted value ; Define the loss function , where RMSE i The loss function of the i-th child video to be predicted. h represents the total number of historical predictions, u represents the number of historical prediction times, u = 1, 2,... h , 、 respectively represent the true value and the predicted value of the i -th historical prediction corresponding to the i-th child video to be predicted; u Establish a prediction optimization index model. The specific model is: , where POI i The prediction optimization index of the i-th child video to be predicted.

[0028] Child video prediction result output module: Optimize the predicted scores of the audiences of the child videos based on the prediction optimization indexes of each child video to be predicted, and output the predicted results of the audiences of each child video to be predicted.

[0029] In this embodiment, it should be specifically noted that the specific execution method of the child video prediction result output module is as follows: Obtain the prediction optimization index of the i-th child video to be predicted POI i , optimize the predicted scores of the audiences of the child videos, obtain the output values of the predicted scores of the audiences of each child video to be predicted after optimization , and output the results to the mobile device of the administrator.

[0030] Embodiment 2 Please refer to Figure 2 shown. The present invention provides a method for predicting audiences of a child video PK operation mechanism, including the following steps: S1: Obtain child video numbers: Obtain the works of the child video PK, mark them as each child video to be predicted, and number each child video to be predicted, and mark them as 1, 2,... i..., n in sequence. i is the number of each child video to be predicted; S2: Extract child video feature information: Extract the respective video feature data values and respective audience behavior data values corresponding to each child video to be predicted from each child video to be predicted; S3: Analyze child video feature information: Based on the respective video feature data values and respective audience behavior data values corresponding to each child video to be predicted, analyze to obtain the feature correlation coefficients of each child video to be predicted, and evaluate the correlation of the child video features, and output the correlation evaluation results;​ S4: Analysis and Evaluation of Children's Video Quality: Based on the color saturation, resolution, and volume in the respective video feature data values corresponding to each to-be-predicted children's video, analyze and obtain the quality evaluation coefficients of each to-be-predicted children's video; S5: Prediction of Children's Video Audience: Based on the feature correlation coefficients, quality evaluation coefficients, and click-through rates of each to-be-predicted children's video, establish an audience prediction model for the children's video PK operation mechanism, and analyze and obtain the audience prediction scores of each to-be-predicted children's video; S6: Prediction Optimization of Children's Video: Based on the audience prediction scores of each to-be-predicted children's video, establish a prediction optimization index model, and analyze and obtain the prediction optimization indices of each to-be-predicted children's video; S7: Output of Children's Video Prediction Results: Optimize the audience prediction scores of children's videos based on the prediction optimization indices of each to-be-predicted children's video, and output the audience prediction results of each to-be-predicted children's video.

[0031] Finally: The above are only the preferred embodiments of the present invention and are not used to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

[0032] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed in the present application can easily think of changes or replacements, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claimed rights.

Claims

1. An audience prediction system for a children's video PK operation mechanism, characterized in that, Including: Children's video number acquisition module: used to obtain the works of children's video PK, mark them as each to-be-predicted children's video, and number each to-be-predicted children's video, sequentially marked as 1, 2, … i …, n, where i is the number of each to-be-predicted children's video; Children's video feature information extraction module: extract each video feature data value and each audience behavior data value corresponding to each to-be-predicted children's video from each to-be-predicted children's video respectively; Children's video feature information analysis module: based on each video feature data value and each audience behavior data value corresponding to each to-be-predicted children's video, analyze to obtain the feature correlation coefficient of each to-be-predicted children's video, and conduct a correlation assessment on the children's video features, and output the correlation assessment result; Children's video quality analysis and evaluation module: based on the color saturation, resolution, and volume in each video feature data value corresponding to each to-be-predicted children's video, analyze to obtain the quality evaluation coefficient of each to-be-predicted children's video; Children's video audience prediction module: based on the feature correlation coefficient, quality evaluation coefficient, and click-through rate of each to-be-predicted children's video, establish an audience prediction model for the children's video PK operation mechanism, and analyze to obtain the audience prediction score of each to-be-predicted children's video; Children's video prediction optimization module: based on the audience prediction score of each to-be-predicted children's video, establish a prediction optimization index model, and analyze to obtain the prediction optimization index of each to-be-predicted children's video; Children's video prediction result output module: optimize the audience prediction score of the children's video based on the prediction optimization index of each to-be-predicted children's video, and output the audience prediction results of each to-be-predicted children's video.

2. The audience prediction system of a children's video PK operation mechanism according to claim 1, characterized in that: The specific execution method of the children's video feature information extraction module is as follows: Extract each video feature data value and each audience behavior data value corresponding to each to-be-predicted children's video from each to-be-predicted children's video respectively; Obtain the video feature data values corresponding to each child video to be predicted , v represent the numbers of the video feature data values, v = 1, 2,... c , where i represents the number of each child video to be predicted, i = 1, 2,... n ; among them, the video feature data value vector corresponding to the i-th child video to be predicted is: ; Obtain the values of each audience behavior data corresponding to each child video to be predicted , p denote the numbers of the values of each audience behavior data, p = 1, 2,... k ; among them, the vector of the values of the audience behavior data corresponding to the i-th child video to be predicted is: .

3. The audience prediction system for the children's video PK operation mechanism according to claim 1, characterized in that: The specific execution method of the children's video feature information analysis module is as follows: Calculate the feature correlation coefficient of each to-be-predicted children's video, and the specific calculation formula is as follows: , where represents the feature correlation coefficient of the i i-th child video to be predicted, and i represents the number of each child video to be predicted. represents the i j-th video feature data value corresponding to the v i-th child video to be predicted. represents the i k-th preset value of the video feature data corresponding to the v i-th child video to be predicted. v represents the number of each video feature data value. c represents the total number of video feature data values. represents the i m-th viewer behavior data value corresponding to the p i-th child video to be predicted. represents the i n-th preset value of the viewer behavior data corresponding to the p i-th child video to be predicted. p represents the number of each viewer behavior data value. k represents the total number of viewer behavior data values. represents an extremely small positive number.

4. The audience prediction system for a children's video PK operation mechanism according to claim 1, wherein: The specific execution method of the children's video quality analysis and evaluation module is as follows: Extract the color saturation, resolution, and volume in the respective video feature data values corresponding to each child video to be predicted, and mark them respectively as , , ; Extract the standard value of color saturation corresponding to the children's video from the management database simultaneously , from the formula , analyze and obtain the color saturation deviation corresponding to the i-th children's video to be predicted SD i ; Calculate the quality evaluation coefficient of each to-be-predicted children's video, and the specific calculation formula is as follows: , where QEC i represents the quality evaluation coefficient of the i-th child video to be predicted, SD max represents the maximum preset color saturation deviation, represents the volume of the i-th child video to be predicted, represents the volume standard value corresponding to the child video, represents the resolution of the i-th child video to be predicted, e represents the natural constant.

5. The audience prediction system of a children's video PK operation mechanism according to claim 1, wherein: The specific execution content of the children's video audience prediction module is as follows: Obtain the i click-through rate of the Cc i th child video to be predicted Df i and the display times, and substitute them into the formula respectively to obtain the i click-through rate of the Ctr i th child video to be predicted; Establish an audience prediction model for the children's video PK operation mechanism, and the specific prediction model is as follows: , where PS i represents the predicted audience rating of the i-th child video to be predicted, represents the v weight value of the j-th video feature data, represents the i feature correlation coefficient of the k-th child video to be predicted, QEC i represents the quality assessment coefficient of the i-th child video to be predicted.

6. The audience prediction system of a children's video PK operation mechanism according to claim 5, characterized in that: The weight value of the v th video feature data is obtained as follows: Step 1: Extract the video feature data values corresponding to each child video to be predicted , and standardize each video feature data value through linear standardization; Step 2: Substitute the video feature data values corresponding to each child video to be predicted into the formula to obtain the proportion of each video feature data value in each child video to be predicted , where i is the number of each child video to be predicted, i = 1, 2, 3,... n, v represents the number of each video feature data value, v = 1, 2,... c ; Step 3, calculate the entropy value of each video feature data: ; Step 4: Calculate the weight values of each video feature data : 。 7. The audience prediction system of a children's video PK operation mechanism according to claim 1, characterized in that: The specific execution method of the children's video prediction optimization module is as follows: Based on the predicted audience scores of each children's video to be predicted, obtain the true values of the predicted audience scores of each historical prediction corresponding to each children's video to be predicted and the predicted values ; Define the loss function ,in, RMSE i The loss function of the i-th child video to be predicted, h represents the total number of historical predictions, u The number indicating the number of historical predictions. u =1,2,... h , , Respectively represent i The child videos to be predicted correspond to u The actual value and predicted value of the historical forecast; Establish a prediction optimization index model, and the specific model is as follows: , where POI i The prediction optimization index of the i-th child video to be predicted.

8. The audience prediction system of a children's video PK operation mechanism according to claim 1, characterized in that: The specific execution method of the children's video prediction result output module is as follows: Obtain the prediction optimization index of the i-th child video to be predicted POI i , optimize the audience prediction score of the child video to obtain the optimized output values of the audience prediction scores of each child video to be predicted , and output the results to the mobile devices of the management personnel 9. A method for predicting the audience of a children's video PK operation mechanism, which is used for the audience prediction system of a children's video PK operation mechanism described in any one of the above claims 1-8, and is characterized in that, Including the following steps: S1: Children's video number acquisition: Obtain the works of children's video PK, mark them as each to-be-predicted children's video, and number each to-be-predicted children's video, sequentially marked as 1, 2, … i …, n, where i is the number of each to-be-predicted children's video; S2: Children's video feature information extraction: Extract each video feature data value and each audience behavior data value corresponding to each to-be-predicted children's video from each to-be-predicted children's video respectively; S3: Analysis of children's video feature information: Based on the respective video feature data values and respective viewer behavior data values corresponding to each children's video to be predicted, analyze to obtain the feature correlation coefficients of each children's video to be predicted, and conduct a correlation assessment on the children's video features, and output the correlation assessment results; S4: Analysis and evaluation of children's video quality: Based on the color saturation, resolution, and volume in the respective video feature data values corresponding to each children's video to be predicted, analyze to obtain the quality assessment coefficients of each children's video to be predicted; S5: Prediction of children's video viewers: Based on the feature correlation coefficients, quality assessment coefficients, and click-through rates of each children's video to be predicted, establish a viewer prediction model for the children's video PK operation mechanism, and analyze to obtain the viewer prediction scores of each children's video to be predicted; S6: Optimization of children's video prediction: Based on the viewer prediction scores of each children's video to be predicted, establish a prediction optimization index model, and analyze to obtain the prediction optimization indices of each children's video to be predicted; S7: Output of children's video prediction results: Optimize the viewer prediction scores of children's videos based on the prediction optimization indices of each children's video to be predicted, and output the viewer prediction results of each children's video to be predicted.

Images