Film content scoring method and system based on multi-dimensional features

Abstract

The invention discloses a film content scoring method and system based on multi-dimensional features, and the method comprises the steps: obtaining the popularity data of actors in a performance library, grading the actors, and building a performance popularity library; the method comprises the following steps: collecting behavior data of current network users, media asset content metadata corresponding to the behavior data, staff information data and questionnaire survey data, performing standardization processing, defining multi-dimensional features, obtaining standardization numerical values corresponding to different dimension features, and calculating standardization scores Zi of the different dimension features; a weight Wi is allocated to the different dimension features, a weighted comprehensive scoring model is constructed, a popularity score Hi of a single actor and an average popularity score Havg of all actors in the same film are obtained, scoring interval calibration is carried out, and a film content score finalScore is obtained; the weight Wi allocation is dynamically adjusted. The considered scoring dimension is comprehensive, the adopted data processing method is scientific and effective, the scoring result is more accurate and reliable, and the efficiency and quality of multimedia content management can be improved.

Description

Technical Field

[0001] This invention relates to the field of multimedia data processing, and specifically to a method and system for scoring film content based on multi-dimensional features. Background Technology

[0002] In the field of multimedia content management, accurate rating of film content is crucial. Existing film rating methods often suffer from the following shortcomings: First, the feature dimensions on which the rating is based are relatively singular, usually only considering some user behavior data or simple content attributes, which makes it difficult to fully reflect the overall quality of the film. Secondly, when processing different types of data, there is a lack of effective data standardization and feature importance normalization methods, resulting in low accuracy and reliability of the scoring results. Third, the scoring model has limited application scenarios and cannot adequately meet the diverse needs of media asset introduction, intelligent recommendation, and operational cataloging.

[0003] For example, patent document CN120088022A discloses an automated film content rating method, system, terminal, and storage medium based on multi-modal data prediction. The method includes: receiving film-related information of a film to be evaluated input by a user; acquiring multi-dimensional data; integrating the multi-dimensional data with the film-related information to obtain a film dataset; analyzing the film data of the film to be evaluated based on the film dataset to obtain multiple film analysis results; performing a preliminary rating prediction on the film to be evaluated based on the film-related information to generate a preliminary rating prediction result; constructing a multi-factor weighted model; and performing a content rating on the film to be evaluated based on the film dataset, multiple film analysis results, and the preliminary rating prediction result to obtain a final rating result. Although this existing technology uses multi-dimensional data, the scope and types of multi-dimensional data are still not comprehensive enough to fully reflect the overall quality of the film. Furthermore, the prediction process is cumbersome and complex, and there is still a lack of effective feature importance normalization methods. The accuracy and reliability of the rating results need further improvement.

[0004] Therefore, there is an urgent need for a technical solution that can comprehensively integrate multi-dimensional features, scientifically process data, and be widely applied to film content rating. Summary of the Invention

[0005] To address the technical issues of existing scoring methods, such as their reliance on a single feature dimension, lack of effective data standardization and feature importance normalization, and limited application scenarios, a multi-dimensional feature-based film content scoring method is proposed, including the following: The system acquires popularity data of actors from the cast and crew database, classifies actors based on the popularity data, and establishes a cast and crew popularity database. The popularity data is multi-dimensional, including platform film and television work search index, award and honor data, work popularity ranking data, and social media influence data. Based on the aforementioned actor and performer popularity database, behavioral data of current internet users, media asset content metadata corresponding to the behavioral data, actor and performer information data, and questionnaire survey data are collected and then standardized to obtain standardized values ​​for different data. The behavioral data includes media asset content exposure, clicks, completion data, and current online user ratings data for the media asset content in the public domain; the media asset content metadata includes the release year and film type. Based on the behavioral data of existing users, the corresponding media content metadata and performer information data, and questionnaire data, multi-dimensional features are defined, and standardized values ​​corresponding to different dimensions of features are obtained. Calculate the standardized score Z of the different dimensional features. i Simultaneously, weights W are assigned to the different dimensional features. i Construct a weighted comprehensive scoring model to obtain the popularity score H of individual actors. i ; Based on the individual actor's popularity score H i Calculate the average popularity score H of all actors in the same film. avg And the average heat score H avg The scoring range is calibrated by mapping to the specified scoring range [minScore, maxScore] to obtain the finalScore of the film content score.

[0006] Furthermore, the film content scoring method also includes analyzing the impact of single-dimensional features on the average popularity score H. avg Contribution Dynamically adjust weight W i The allocation of features is optimized using multi-dimensional features and a weighted comprehensive scoring model.

[0007] Furthermore, the formula for calculating the film content score finalScore is as follows: ; Wherein, Score is the raw score, i.e., the average popularity score H. avg .

[0008] Furthermore, the average popularity score H of all actors in the same film. avg The calculation formula is expressed as: ; Where m is the total number of actors in the same film, and i = 1, 2, 3...m.

[0009] Furthermore, the popularity score H of the individual actor i The calculation formula is expressed as: ; Where n is the number of dimensions of the feature.

[0010] Furthermore, the standardized score Z i The calculation formula is expressed as: ; in, Let i be the standardized value of feature i; The mean of the standardized values ​​of feature i; Let be the standard deviation of the standardized values ​​of feature i.

[0011] Furthermore, the weight W i The allocation method is as follows: based on the defined multi-dimensional features, the importance of the multi-dimensional features is determined, the importance of the multi-dimensional features is normalized, and then weight values ​​of the importance of the multi-dimensional features are assigned, with the sum of the weight values ​​being 1.

[0012] Furthermore, the multi-dimensional features are defined as including features such as exposure, clicks, completion rate, conversion rate, image quality data, overall online popularity data, information completeness, film type, content tags, cast and crew, and director; the importance of the multi-dimensional features is determined by ranking the defined multi-dimensional features from high to low.

[0013] Furthermore, the single-dimensional feature affects the average heat score H. avg Contribution The calculation formula is expressed as follows: .

[0014] The present invention also provides a film content scoring system based on multi-dimensional features, which is implemented by the aforementioned film content scoring method based on multi-dimensional features. The film content scoring system includes a data acquisition module, a data preprocessing module, a cast and crew popularity processing module, a feature processing module, a scoring calculation module, and an application interaction module. The data acquisition module is used to receive the film content to be rated transmitted by the application interaction module, and to collect the behavior data of the current network users, the media asset content metadata and cast and crew information data, and the questionnaire survey data for the film content to be rated. The data preprocessing module is used to standardize the data collected by the data acquisition module to obtain standardized values ​​for different data. ; The actor and staff popularity processing module is used to establish the actor and staff popularity database and match popularity data for the actors and staff in the film to be rated. The feature processing module is used to define multi-dimensional features based on the data collected by the data acquisition module, normalize the importance of the multi-dimensional features, and assign weights W. i And obtain the standardized values ​​corresponding to features of different dimensions. ; The scoring calculation module is used to assign weights W according to the feature processing module. i and the obtained standardized values Calculate the popularity score H of each actor in the film to be rated. i And the average popularity score H of all actors avg And to perform scoring interval calibration and single-dimensional feature contribution. The system analyzes and obtains the finalScore of the video content score, and then transmits the finalScore to the application interaction module. The application interaction module is used to acquire the content of the film to be rated, input it into the data acquisition module, and output the film content rating finalScore acquired by the rating calculation module.

[0015] The beneficial effects of this invention are as follows: This invention establishes a cast and crew popularity database and classifies the popularity of actors in the database to determine key dimensional data that can be used to evaluate actor popularity. Then, it classifies multi-dimensional features related to the content of the film to be rated, comprehensively integrating these features, including user behavior data, content metadata, and cast and crew information, to fully reflect the overall quality of the film. It employs Z-score standardization, analytic hierarchy process (AHP), and principal component analysis (PCA) for data preprocessing and feature processing, achieving data standardization and feature importance normalization to ensure the accuracy and reliability of the scoring results. By analyzing the contribution of single-dimensional features to the film content score, it dynamically adjusts the weight allocation, optimizes the multi-dimensional features and weighted comprehensive scoring model, and expands the application scope. It can be widely used for media asset introduction reference, intelligent recommendation recall and ranking, and operational cataloging assistance, improving the efficiency and quality of multimedia content management. Simultaneously, it updates the scoring and analysis of content in specific scenarios in real time for different time periods and seasons. Attached Figure Description

[0016] Figure 1 This is a flowchart of the film content scoring method based on multi-dimensional features provided by the present invention; Figure 2 This is a logic diagram of the film content rating system based on multi-dimensional features provided by the present invention. Detailed Implementation

[0017] The technical solution of the present invention is further described below, but the scope of protection is not limited to what is described.

[0018] This invention provides a method for rating film content based on multi-dimensional features, including the following: Step S100: Obtain the popularity data of actors in the cast and crew database, classify the actors according to the popularity data, and establish a cast and crew popularity database; the popularity data is multi-dimensional data, including platform film and television work search index, award and honor data, work popularity ranking data, and social media influence data; it covers the number of views, ratings, discussion, and hot list rankings of film and television works in which the actors and crew participate on major platforms, and selects relevant data of the most recent works in which the actors and crew participate; the major platforms include Douban, Tencent, iQiyi, Maoyan, and Lighthouse.

[0019] The classification is specifically as follows: based on factors such as the number of works, box office performance, and social media attention of actors within a certain period of time, actors in the cast and crew database are divided into different popularity levels, such as first-tier, second-tier, and third-tier, to form a cast and crew popularity database.

[0020] Step S200: Based on the actor and performer popularity database, collect behavioral data of current network users, media asset content metadata corresponding to the behavioral data, actor and performer information data, and questionnaire survey data, and then perform standardization processing to obtain standardized values ​​for different data. The behavioral data includes media asset content exposure, clicks, completion data, and current online user ratings data for the media asset content in the public domain; the media asset content metadata includes the release year and film type; the questionnaire survey data is user evaluation data of the film content obtained through questionnaire surveys.

[0021] For example, we can obtain behavioral data such as media asset exposure, clicks, and completion rates from user data on our proprietary IPTV platform, and obtain user ratings for film content from public platforms through the VGS system. We can also extract media asset content and its corresponding cast and crew information, including actors and directors, through our proprietary media asset management system. Simultaneously, we can obtain metadata such as the film's release year and genre, including comedy, action, and science fiction.

[0022] The standardization process includes converting exposure and click data into a uniform numerical range through normalization or standardization methods; and performing one-hot encoding or tag encoding on film type data.

[0023] Step S300: Based on the behavioral data of the existing network users, the media asset content metadata and performer information data corresponding to the behavioral data, and the questionnaire survey data, define multi-dimensional features and obtain standardized values ​​corresponding to different dimensional features. Calculate the standardized score Z of the different dimensional features. i Simultaneously, weights W are assigned to the different dimensional features. i Construct a weighted comprehensive scoring model to obtain the popularity score H of individual actors. i ; The standardized score Z i The calculation formula is expressed as: (1); in, Let i be the standardized value of feature i; The mean of the standardized values ​​of feature i; Let be the standard deviation of the standardized values ​​of feature i.

[0024] The weight W i The allocation method is as follows: based on the defined multi-dimensional features, the importance of the multi-dimensional features is determined, the importance of the multi-dimensional features is normalized, and then weight values ​​of the importance of the multi-dimensional features are assigned, with the sum of the weight values ​​being 1.

[0025] The multi-dimensional features are defined as including features such as exposure, clicks, completion rate, conversion rate, image quality data, overall online popularity data, information completeness, film type, content tags, cast and crew, and director; it also includes the ratio of completion rate to clicks.

[0026] The importance of the multi-dimensional features is defined by ranking the multi-dimensional features from high to low.

[0027] The normalization methods for the importance of multidimensional features include the analytic hierarchy process (AHP) and principal component analysis (PCA).

[0028] Step S400, based on the popularity score H of the individual actor i Calculate the average popularity score H of all actors in the same film. avg And the average heat score H avg The scoring range is calibrated by mapping to the specified scoring range [minScore, maxScore] to obtain the finalScore of the film content score.

[0029] The popularity score H of the individual actor i The calculation formula is expressed as: (2); Where n is the number of dimensions of the feature.

[0030] The average popularity score H of all actors in the same film avg The calculation formula is expressed as: (3); Where m is the total number of actors in the same film, and i = 1, 2, 3...m.

[0031] The scoring interval calibration involves mapping the calculated raw scores to a specified scoring range [minScore, maxScore], such as [0, 100], through methods such as linear transformation, to facilitate intuitive understanding and application, thereby obtaining the film content score finalScore. The calculation formula for the film content score finalScore is as follows: (4) Wherein, Score is the raw score, i.e., the average popularity score H. avg .

[0032] Step S500, the film content scoring method further includes analyzing the average popularity score H based on single-dimensional features. avg Contribution Dynamically adjust weight W i The allocation of features is optimized using multi-dimensional features and a weighted comprehensive scoring model.

[0033] The single-dimensional feature affects the average heat score H. avg Contribution The calculation formula is expressed as follows: (5).

[0034] This invention also provides a film content scoring system based on multi-dimensional features, implemented using the aforementioned film content scoring method based on multi-dimensional features. The film content scoring system includes a data acquisition module, a data preprocessing module, a cast and crew popularity processing module, a feature processing module, a scoring calculation module, and an application interaction module. The data acquisition module is used to receive the film content to be rated transmitted by the application interaction module, and to collect the behavior data of the current network users, the media asset content metadata and cast and crew information data, and the questionnaire survey data for the film content to be rated. The data preprocessing module is used to standardize the data collected by the data acquisition module to obtain standardized values ​​for different data. ; The actor and staff popularity processing module is used to establish the actor and staff popularity database and match popularity data for the actors and staff in the film to be rated. The feature processing module is used to define multi-dimensional features based on the data collected by the data acquisition module, normalize the importance of the multi-dimensional features, and assign weights W. iAnd obtain the standardized values ​​corresponding to features of different dimensions. ; The scoring calculation module is used to assign weights W according to the feature processing module. i and the obtained standardized values Calculate the popularity score H of each actor in the film to be rated. i And the average popularity score H of all actors avg And to perform scoring interval calibration and single-dimensional feature contribution. The system analyzes and obtains the finalScore of the video content score, and transmits the finalScore to the application interaction module. The application interaction module is used to acquire the content of the film to be rated, input it into the data acquisition module, and output the film content rating finalScore acquired by the rating calculation module.

[0035] The above-disclosed embodiments are merely specific examples of the present invention. However, the present invention is not limited thereto, and any variations that can be conceived by those skilled in the art should fall within the protection scope of the present invention.

Claims

1. A film content scoring method based on multi-dimensional features, characterized in that, Includes the following: The system acquires popularity data of actors from the cast and crew database, classifies actors based on the popularity data, and establishes a cast and crew popularity database. The popularity data is multi-dimensional, including platform film and television work search index, award and honor data, work popularity ranking data, and social media influence data. Based on the aforementioned actor and performer popularity database, behavioral data of current internet users, media asset content metadata corresponding to the behavioral data, actor and performer information data, and questionnaire survey data are collected and then standardized to obtain standardized values ​​for different data. The behavioral data includes media asset content exposure, clicks, completion data, and current online user ratings data for the media asset content in the public domain; the media asset content metadata includes the release year and film type. Based on the behavioral data of existing users, the corresponding media content metadata and performer information data, and questionnaire data, multi-dimensional features are defined, and standardized values ​​corresponding to different dimensions of features are obtained. Calculate the standardized score Z of the different dimensional features. i Simultaneously, weights W are assigned to the different dimensional features. i Construct a weighted comprehensive scoring model to obtain the popularity score H of individual actors. i ; Based on the individual actor's popularity score H i Calculate the average popularity score H of all actors in the same film. avg And the average heat score H avg The scoring range is calibrated by mapping to the specified scoring range [minScore, maxScore] to obtain the finalScore of the film content score.

2. The film content scoring method based on multi-dimensional features as described in claim 1, characterized in that, The film content scoring method also includes analyzing single-dimensional features on the average popularity score H. avg Contribution Dynamically adjust weight W i The allocation of features is optimized using multi-dimensional features and a weighted comprehensive scoring model.

3. The film content scoring method based on multi-dimensional features as described in claim 2, characterized in that, The formula for calculating the finalScore of the film content score is as follows: ； Wherein, Score is the raw score, i.e., the average popularity score H. avg .

4. The film content scoring method based on multi-dimensional features as described in claim 2, characterized in that, The average popularity score H of all actors in the same film avg The calculation formula is expressed as: ； Where m is the total number of actors in the same film, and i = 1, 2, 3...m.

5. The film content scoring method based on multi-dimensional features as described in claim 2, characterized in that, The popularity score H of the individual actor i The calculation formula is expressed as: ； Where n is the number of dimensions of the feature.

6. The film content scoring method based on multi-dimensional features as described in claim 2, characterized in that, The standardized score Z i The calculation formula is expressed as: ； in, Let i be the standardized value of feature i; The mean of the standardized values ​​of feature i; Let be the standard deviation of the standardized values ​​of feature i.

7. The film content scoring method based on multi-dimensional features as described in claim 2, characterized in that, The weight W i The allocation method is as follows: based on the defined multi-dimensional features, the importance of the multi-dimensional features is determined, the importance of the multi-dimensional features is normalized, and then weight values ​​of the importance of the multi-dimensional features are assigned, with the sum of the weight values ​​being 1.

8. The film content scoring method based on multi-dimensional features as described in claim 7, characterized in that, The multi-dimensional features are defined as including features such as exposure, clicks, completion rate, conversion rate, image quality data, overall online popularity data, information completeness, film type, content tags, cast and crew, and director; the importance of the multi-dimensional features is arranged from high to low according to the order of the multi-dimensional features.

9. The film content scoring method based on multi-dimensional features as described in claim 2, characterized in that, The single-dimensional feature affects the average heat score H. avg Contribution The calculation formula is expressed as follows: 。 10. A film content rating system based on multi-dimensional features, characterized in that, The film content scoring method based on multi-dimensional features as described in any one of claims 1-9 is applied. The film content scoring system includes a data acquisition module, a data preprocessing module, a cast and crew popularity processing module, a feature processing module, a scoring calculation module, and an application interaction module. The data acquisition module is used to receive the film content to be rated transmitted by the application interaction module, and to collect the behavior data of the current network users, the media asset content metadata and cast and crew information data, and the questionnaire survey data for the film content to be rated. The data preprocessing module is used to standardize the data collected by the data acquisition module to obtain standardized values ​​for different data. ; The actor and staff popularity processing module is used to establish the actor and staff popularity database and match popularity data for the actors and staff in the film to be rated. The feature processing module is used to define multi-dimensional features based on the data collected by the data acquisition module, normalize the importance of the multi-dimensional features, and assign weights W. i And obtain the standardized values ​​corresponding to features of different dimensions. ; The scoring calculation module is used to assign weights W according to the feature processing module. i and the obtained standardized values Calculate the popularity score H of each actor in the film to be rated. i And the average popularity score H of all actors avg And to perform scoring interval calibration and single-dimensional feature contribution. The system analyzes and obtains the finalScore of the video content score, and then transmits the finalScore to the application interaction module. The application interaction module is used to acquire the content of the film to be rated, input it into the data acquisition module, and output the film content rating finalScore acquired by the rating calculation module.

Images