Film and television regulatory analysis methods, systems, and storage media based on data mining

Abstract

This application relates to the field of data processing technology and discloses a film and television regulatory analysis method, system, and storage medium based on data mining. The method includes: comprehensively acquiring film and television data and performing cross-media transcoding on a multidimensional dataset; extracting video semantics, recognizing audio sentiment, and analyzing textual tendencies from the dataset to generate a time-series marker list; scoring key elements according to social impact, dissemination depth, and audience acceptance to form a dynamic scoring table; analyzing the dissemination path based on the scoring table to construct an impact map; ranking regulatory points according to the map to formulate an intelligent regulatory plan; and analyzing execution deviations through effect feedback to optimize regulatory parameters. In a massive multimedia content environment, this application achieves multi-dimensional content feature extraction, social impact assessment, and dissemination path analysis of film and television works, and accordingly forms a precise and differentiated intelligent regulatory plan, while also possessing self-optimization and adjustment capabilities.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a film and television regulatory analysis method, system and storage medium based on data mining. Background Technology

[0002] With the rapid development and widespread adoption of internet video platforms, the number of film and television works has exploded, and the content types have become increasingly diversified. Traditional film and television content supervision mainly relies on manual review, where professional reviewers watch, screen, and categorize works one by one, supplemented by simple technical means such as keyword filtering and blacklist management. This regulatory model was sufficient to meet basic needs in the early days of the internet when the scale of content was relatively limited. Various countries and regions have subsequently established rating systems and content review standards to guide the distribution and dissemination of film and television content. Some regions have also developed automated content recognition technologies, such as rule-based sensitive content blocking systems and simple image recognition and voice analysis tools, to improve review efficiency.

[0003] However, with the rise of short videos, live streaming, and user-generated content (UGC), traditional regulatory methods face numerous shortcomings and challenges. Manual review is inefficient and unable to handle the real-time review demands of massive amounts of content, resulting in a large amount of harmful content having a widespread impact before it is even discovered. Secondly, simple keyword filtering and rule-matching mechanisms are easily circumvented by content creators, making it difficult to identify harmful content with subtle expressions and cross-media integration. Thirdly, traditional regulatory methods lack the ability to analyze content dissemination paths and social impact, making it impossible to accurately assess the actual degree of harm and prioritize handling. Furthermore, existing technologies suffer from rigid regulatory standards, broad-based grading, and a lack of personalized measures, making it difficult to adapt to the differentiated needs of different audience groups. Most importantly, traditional regulatory methods lack self-learning and optimization mechanisms, making it difficult to cope with constantly changing content formats and dissemination models, and balancing regulatory effectiveness with implementation costs remains a challenge. Summary of the Invention

[0004] This application provides a data mining-based film and television regulatory analysis method, system, and storage medium, which is used to extract multi-dimensional content features, assess social impact, and analyze dissemination paths of film and television works in a massive multimedia content environment, and form a precise and differentiated intelligent regulatory solution based on this, while also having the ability to self-optimize and adjust.

[0005] Firstly, this application provides a film and television regulatory analysis method based on data mining. This method includes: collecting panoramic data on film and television works from video platforms; performing layered and standardized processing of video, audio, and text information through cross-media transcoding to obtain a multi-dimensional film and television content dataset; based on the multi-dimensional film and television content dataset, extracting scene semantics from video frames, performing multi-level emotion recognition on audio, and performing cultural value orientation analysis on text through content feature identification to obtain a time-series related content tagging list; based on the time-series related content tagging list, cross-scoring key elements according to social impact intensity, dissemination penetration depth, and audience psychological acceptance through hierarchical scoring to obtain a dynamic weighted content scoring table; based on the dynamic weighted content scoring table, analyzing the network dissemination path of film and television works through correlation tracking to obtain a three-dimensional content influence map; based on the three-dimensional content influence map, prioritizing regulatory points through decision rules to obtain an intelligent regulatory scheme including precise modification suggestions, gradient restriction measures, and personalized grading requirements; and based on the intelligent regulatory scheme, performing deviation correction analysis on the implementation effect of regulatory measures through effect feedback to obtain a self-optimizing regulatory parameter set.

[0006] Secondly, this application provides a film and television regulatory analysis system based on data mining, the film and television regulatory analysis system based on data mining includes:

[0007] The acquisition module is used to collect panoramic data from film and television works on video platforms. Through cross-media transcoding, video, audio, and text information are processed in a layered and standardized manner to obtain a multi-dimensional film and television content dataset.

[0008] The extraction module is used to extract scene semantics from video frames, perform multi-level emotion recognition on audio, and perform cultural value orientation analysis on text based on the multi-dimensional film and television content dataset, so as to obtain a list of time-related content tags.

[0009] The scoring module is used to cross-score key elements based on the time-related content tag list, according to the intensity of social impact, the depth of dissemination penetration, and the psychological acceptance of the audience, to obtain a dynamic weighted content scoring table.

[0010] The analysis module is used to analyze the network propagation path of film and television works by correlation tracking based on the dynamic weighted content scoring table, and obtain a three-dimensional content influence map.

[0011] The sorting module is used to prioritize the regulatory points according to the three-dimensional content influence map and through decision rules to obtain an intelligent regulatory solution that includes precise modification suggestions, gradient restriction measures and personalized classification requirements.

[0012] The analysis module is used to perform deviation correction analysis on the implementation effect of the regulatory measures based on the intelligent regulatory scheme and through effect feedback, so as to obtain a self-optimizing regulatory parameter set.

[0013] A third aspect of the present invention provides a computer device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor invokes the instructions in the memory to cause the computer device to execute the above-described data mining-based film and television regulatory analysis method.

[0014] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the above-described data mining-based film and television regulatory analysis method.

[0015] The technical solution provided in this application achieves layered and standardized organization of video, audio, and text information through panoramic data acquisition and cross-media transcoding, overcoming the problems of fragmented data acquisition and inconsistent formats in traditional film and television supervision. It establishes a multi-dimensional film and television content dataset, providing a high-quality data foundation for subsequent analysis. Through content feature recognition technology, it extracts scene semantics from video frames, performs multi-level emotion recognition on audio, and analyzes cultural value orientations in text, breaking through the limitations of single-media analysis. This achieves semantic association of cross-media content, forming a temporally related content tagging list and enhancing the ability to identify obscure and harmful content. Based on a graded scoring mechanism, key elements are categorized according to their social impact intensity, dissemination penetration depth, and target audience. The system innovatively introduces a multi-dimensional evaluation system by cross-scoring public psychological acceptance, generating a dynamic weighted content scoring table to make regulatory assessments more comprehensive and objective. Through correlation tracking technology, it analyzes the network-like dissemination paths of film and television works, revealing the spatiotemporal dynamic characteristics of content dissemination and constructing a three-dimensional content impact map, providing a basis for precise regulation. Based on decision-making rules, it prioritizes regulatory points, forming an intelligent regulatory solution that includes precise modification suggestions, tiered restrictions, and personalized grading requirements, achieving optimized allocation and differentiated processing of regulatory resources. Through an effect feedback mechanism, it analyzes the deviation correction of the implementation effect of regulatory measures, generating a self-optimizing regulatory parameter set, enabling the system to have self-learning and continuous optimization capabilities. It is particularly worth emphasizing that this solution fully leverages the technological contributions of artificial intelligence algorithms and models in specific functions and application areas. For example, the deep learning visual recognition model used in scene semantic extraction can accurately identify sensitive content in complex scenes; the sentiment computing algorithm used in multi-level sentiment recognition can capture the emotional changes implied in sound; the natural language processing technology introduced in cultural value orientation analysis can understand the deep semantics of text; the network propagation algorithm used in propagation path analysis can predict the scope of influence and propagation trend of content; and the multi-objective optimization algorithm applied in regulatory decision-making can balance regulatory effectiveness and resource consumption. The organic combination of these algorithmic features makes this solution demonstrate significant technical advantages in the field of massive multimedia content supervision, improving both the accuracy and efficiency of supervision while reducing labor costs. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of one embodiment of the film and television regulatory analysis method based on data mining in this application.

[0018] Figure 2 This is a schematic diagram of one embodiment of the film and television supervision and analysis system based on data mining in this application.

[0019] Figure 3 This is a schematic block diagram of the structure of the computer device in an embodiment of the present invention. Detailed Implementation

[0020] This application provides a data mining-based method, system, and storage medium for film and television regulatory analysis. The terms "first," "second," "third," "fourth," etc. (if present)," in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0021] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the film and television regulatory analysis method based on data mining in this application includes:

[0022] Step S101: Collect panoramic data of film and television works on the video platform, and perform layered and standardized organization of video, audio and text information through cross-media transcoding to obtain a multi-dimensional film and television content dataset.

[0023] Step S102: Based on the multidimensional film and television content dataset, extract scene semantics from video frames, perform multi-level emotion recognition on audio, and conduct cultural value orientation analysis on text through content feature recognition to obtain a list of time-related content tags.

[0024] Step S103: Based on the time-series related content tagging list, the key elements are cross-scored according to the intensity of social impact, the depth of dissemination penetration, and the psychological acceptance of the audience through hierarchical scoring to obtain a dynamic weighted content scoring table;

[0025] Step S104: Based on the dynamic weighted content scoring table, analyze the network dissemination path of film and television works through correlation tracking to obtain a three-dimensional content influence map;

[0026] Step S105: Based on the three-dimensional content impact map, prioritize the regulatory points using decision rules to obtain an intelligent regulatory solution that includes precise modification suggestions, gradient restriction measures, and personalized grading requirements.

[0027] Step S106: Based on the intelligent supervision scheme, conduct deviation correction analysis on the implementation effect of supervision measures through effect feedback to obtain a self-optimizing supervision parameter set.

[0028] It is understood that the executing entity of this application can be a film and television supervision and analysis system based on data mining, or it can be a terminal or a server; the specific implementation is not limited here. This application's embodiment uses a server as an example for illustration.

[0029] Specifically, panoramic data acquisition technology is used to comprehensively acquire data from film and television works on video platforms. For video, audio, and text data of different formats, cross-media transcoding technology is employed to convert heterogeneous data into a unified standard format. Specifically, for video data, keyframes are extracted and encoded for standardization; for audio data, sampling rate and bit depth are unified; and for text data, encoding format conversion and structural standardization are performed. This processed data is then hierarchically structured to construct a multi-dimensional film and television content dataset, which includes time-aligned media content and metadata information.

[0030] Based on a multidimensional film and television content dataset, content feature identification is performed. For video frames, scene semantic extraction technology is used to identify key elements, scene types, and visual semantics within the frame. For audio content, multi-level sentiment recognition technology is used to analyze features such as tone, rhythm, and volume changes to extract emotional tendencies. For text content, cultural value orientation analysis technology is used to uncover the underlying value orientations and cultural stances. These three types of feature data are linked and integrated according to timestamps to form a time-series associated content tag list. This list contains multimodal content features and tags for each time point in the film and television works.

[0031] Based on a time-series-related content tagging list, a tiered scoring technique is used to evaluate key elements from multiple dimensions. The social impact intensity score is based on the content's sensitivity and public attention; the dissemination penetration depth score is based on the content's dissemination speed and coverage; and the audience psychological acceptance score considers the psychological feelings and acceptance levels of different groups. Through a cross-scoring method, these three dimensions are weighted and calculated to generate a content scoring table with dynamic weight coefficients, achieving precise quantitative rating of film and television content. Based on the dynamic weighted content scoring table, correlation tracking technology is used to perform network analysis of the dissemination path of film and television works, tracking the dissemination trajectory and impact range of content across different platforms, regions, and populations. By establishing a content correlation network, key nodes and dissemination path characteristics are identified. Combining time and geographical data, a three-dimensional content impact map is constructed, intuitively displaying the distribution and development trend of the social impact of film and television content.

[0032] Based on a three-dimensional content impact map, decision rule technology is applied to systematically analyze and prioritize regulatory points. High-risk regulatory points undergo content compliance assessments, generating targeted modification suggestions. Tiered restriction measures are formulated by combining dissemination scope and influence data; personalized tiered requirements are generated based on audience characteristics. These integrated elements form an intelligent regulatory solution, providing precise regulatory decision support for different types of film and television content. Based on the implementation of the intelligent regulatory solution, execution data of regulatory measures are collected through an effect feedback mechanism. The differences between expected and actual effects are compared to conduct root cause analysis of deviations. Parameter adjustment strategies are developed for different deviation factors, and after verification, a self-optimizing regulatory parameter set is formed, achieving continuous optimization and self-adjustment of the regulatory system.

[0033] Taking a popular TV drama as an example, video, audio, and text data from the drama were collected, unified into a standard format through cross-media transcoding, and a multi-dimensional film and television content dataset was constructed. Semantic extraction was then performed on violent scenes appearing in the video frames, multi-layered emotion recognition was conducted on anger expressed in character dialogues, and cultural orientation analysis was performed on the value orientation of the dialogue text, generating a list of time-related content tags. Based on this list, the social impact intensity of the violent content was assessed as high-risk, the penetration depth among teenagers was 76%, and the audience's psychological acceptance was moderate to low, thus generating a dynamic weighted content scoring table. Through correlation tracking, it was found that the violent content spread in a network on social media, with its influence continuously expanding, forming a three-dimensional content impact map. Decision rules identified this content as a high-priority regulatory point, generating precise modification suggestions, age-group-based tiered restrictions, and personalized grading requirements, forming an intelligent regulatory solution. After implementation, feedback showed that the contact rate among teenagers decreased to an acceptable range, but some restrictions were circumvented. After deviation correction analysis, the regulatory parameters were adjusted to form a self-optimizing regulatory parameter set, achieving effective regulation of the film and television content.

[0034] In this embodiment, panoramic data acquisition and cross-media transcoding are used to achieve hierarchical and standardized organization of video, audio, and text information, overcoming the problems of fragmented data acquisition and inconsistent formats in traditional film and television supervision. A multi-dimensional film and television content dataset is established, providing a high-quality data foundation for subsequent analysis. Content feature recognition technology is used to extract scene semantics from video frames, perform multi-level emotion recognition on audio, and analyze cultural value orientations in text, breaking through the limitations of single-media analysis and achieving semantic association of cross-media content. This forms a temporally related content tagging list, enhancing the ability to identify obscure and harmful content. Based on a graded scoring mechanism, key elements are categorized according to their social impact intensity, dissemination penetration depth, and audience perception. The system innovatively introduces a multi-dimensional evaluation system by cross-scoring the acceptance of content, generating a dynamic weighted content scoring table to make regulatory assessments more comprehensive and objective. Through correlation tracking technology, it analyzes the network-like dissemination paths of film and television works, revealing the spatiotemporal dynamic characteristics of content dissemination and constructing a three-dimensional content impact map, providing a basis for precise regulation. Based on decision-making rules, it prioritizes regulatory points, forming an intelligent regulatory solution that includes precise modification suggestions, tiered restrictions, and personalized grading requirements, achieving optimized allocation and differentiated processing of regulatory resources. Through an effect feedback mechanism, it analyzes the deviation correction of the implementation effect of regulatory measures, generating a self-optimizing regulatory parameter set, enabling the system to have self-learning and continuous optimization capabilities. It is particularly worth emphasizing that this solution fully leverages the technological contributions of artificial intelligence algorithms and models in specific functions and application areas. For example, the deep learning visual recognition model used in scene semantic extraction can accurately identify sensitive content in complex scenes; the sentiment computing algorithm used in multi-level sentiment recognition can capture the emotional changes implied in sound; the natural language processing technology introduced in cultural value orientation analysis can understand the deep semantics of text; the network propagation algorithm used in propagation path analysis can predict the scope of influence and propagation trend of content; and the multi-objective optimization algorithm applied in regulatory decision-making can balance regulatory effectiveness and resource consumption. The organic combination of these algorithmic features makes this solution demonstrate significant technical advantages in the field of massive multimedia content supervision, improving both the accuracy and efficiency of supervision while reducing labor costs.

[0035] In one specific embodiment, the process of performing step S101 may specifically include the following steps:

[0036] (1) Deeply penetrate and collect film and television content from multiple video platforms through an adaptive capture network to obtain full-spectrum film and television raw data;

[0037] (2) Perform digital fingerprint recognition on the full spectrum of film and television raw data, and perform intelligent detection on video frame rate, audio sampling and text encoding to obtain media feature map;

[0038] (3) Based on the media feature map, the heterogeneous media formats are spatiotemporally synchronized through cross-encoding conversion to obtain a unified standard media stream;

[0039] (4) Perform semantic segmentation on the unified standard media stream, and use keyframe localization technology to accurately slice video, audio and text to obtain a semantically related slice set;

[0040] (5) The semantic association slice set is hierarchically classified by multidimensional cross mapping to construct a content association network and obtain a structured content matrix;

[0041] (6) Write the structured content matrix into the data warehouse through distributed storage, establish a multi-dimensional retrieval channel, and obtain a multi-dimensional film and television content dataset.

[0042] Specifically, an adaptive capture network is used to deeply penetrate and collect film and television content from multiple video platforms. The adaptive capture network is a distributed crawler cluster technology that dynamically adjusts its crawling strategy based on the data structure characteristics of different video platforms, overcoming platform anti-crawling mechanisms and achieving effective collection of encrypted content, streaming media, and hidden pages. This network consists of multiple layers of proxy nodes, which maintain high-efficiency collection while avoiding platform blocking through IP rotation, simulated user behavior, and parallel task allocation. During the collection process, the network automatically adjusts the request frequency based on the platform's response speed and employs differentiated crawling depths for different types of content. The acquired full-spectrum film and television raw data includes a complete set of content such as video files, audio streams, subtitle text, comment data, and metadata information. The acquired full-spectrum film and television raw data undergoes digital fingerprinting processing, and unique data features are extracted from the video, audio, and text content using feature extraction algorithms. Video frame rate detection calculates the frame rate of the original video by analyzing the inter-frame time interval; audio sampling recognition determines the audio sampling rate, bit depth, and channel configuration through spectral analysis; and text encoding detection identifies the text encoding format through character set features. These parameters form the feature identifiers of each media file, constructing a media feature map.

[0043] Based on media feature maps, further spatiotemporal synchronization processing is performed on heterogeneous media formats through cross-encoding conversion. Cross-encoding conversion converts media files from different sources into a unified standard format. Video transcoding adopts the H.264 / AVC encoding standard, based on a frame rate of 25fps and a resolution of 1080p; audio transcoding is unified to AAC format with a sampling rate of 44.1kHz and a depth of 16 bits; and text content is converted to the standard text format encoded in UTF-8. Spatiotemporal synchronization processing ensures precise alignment of audio, video, and text on the timeline, generating a unified standard media stream. Semantic segmentation processing is performed on the unified standard media stream, and keyframe localization technology is applied to slice video, audio, and text content. Video keyframe localization is based on shot transition detection algorithms, identifying scene change points by analyzing visual differences between adjacent frames; audio segmentation is based on acoustic feature changes and silence detection, identifying natural separation points of sound content; and text segmentation is based on semantic units and syntactic structure, dividing continuous text into meaningful sentence segments. Each media slice is associated and matched according to its timestamp to form a semantically associated slice set. Each slice contains video, audio and text content within the same time period.

[0044] This study employs a multi-dimensional cross-mapping technique to hierarchically categorize semantically related video clips, establishing a model of relationships between data across different dimensions. Clips are tagged with features based on content theme, sentiment, and characters / objects. Then, similarity calculations are used to cluster clips with similar features, establishing connections between clips and forming a multi-layered content association network. This content association network reflects the logical structure and semantic relationships of film and television content, expressed in matrix form to form a structured content matrix. The structured content matrix is ​​written to a data warehouse using a distributed storage mechanism. Sharding technology is used to distribute large-scale data across multiple nodes, establishing multi-dimensional retrieval channels to support efficient queries. During storage, an index structure based on multiple dimensions such as content features, timestamps, and sentiment is established, supporting multi-angle and multi-level data retrieval and analysis, forming a complete multi-dimensional film and television content dataset, providing a data foundation for subsequent regulatory analysis.

[0045] Taking a popular online drama series as an example, the adaptive capture network, based on the distribution of the series across three major video platforms, set a targeted crawling strategy to collect full-spectrum raw film and television data, including 100 episodes of the main content, 50 behind-the-scenes videos, and 3,000 user comments. Subsequently, digital fingerprinting was performed, detecting that the video formats included MP4 and FLV, with frame rates of 24fps and 30fps respectively; the audio sampling rates included 44.1kHz and 48kHz; and the text encoding included UTF-8 and GBK, generating a media feature map. Based on the feature map, cross-encoding conversion was performed, uniformly converting all videos to H.264 encoding, 25fps MP4 format; audio to 44.1kHz, 16-bit AAC format; and text to UTF-8 encoding, achieving spatiotemporal synchronization to obtain a unified standard media stream. Next, semantic segmentation was performed. Keyframe algorithms identified 1200 scene transition points in the video, audio analysis identified 850 sound change points, and text analysis extracted 3500 semantic units. These were then matched by timestamps to form a semantically related slice set. Multidimensional cross-mapping was used to categorize the slices, constructing a content association network based on three dimensions: "plot development - character interaction - emotional change," generating a structured content matrix. The matrix data was then fragmented and stored across multiple nodes in a data warehouse, establishing retrieval channels based on content type, temporal order, and semantic association. This generated a multidimensional film and television content dataset, laying the foundation for content monitoring and analysis of this popular drama series.

[0046] In one specific embodiment, the process of performing step S102 may specifically include the following steps:

[0047] (1) Extract video key frame sequences from the multidimensional film and television content dataset, perform edge detection and region division on the video screen through scene segmentation technology, and obtain video scene combination features;

[0048] (2) Use a semantic understanding engine to classify thematic elements of the video scene combination features, extract key visual elements and contextual relationships in the scene, and obtain a scene semantic mapping set;

[0049] (3) Separate the audio data stream from the multidimensional film and television content dataset, and perform waveform decomposition of the audio emotional features through spectrum analysis to obtain the emotional audio feature spectrum;

[0050] (4) Multi-level labeling of emotional audio feature spectrum, and emotional classification of pitch, speech rate and rhythm changes by emotional intensity gradient quantization to obtain multi-level emotional labeling data;

[0051] (5) Extract text content from multidimensional film and television content datasets, extract keywords and analyze value orientation of text through semantic network to obtain cultural value orientation indicators;

[0052] (6) Integrate and associate the scene semantic mapping set, multi-level sentiment tag data and cultural value orientation indicators according to the time axis, establish the temporal mapping relationship between content elements, and obtain a temporal associated content tag list.

[0053] Specifically, representative frames are selected from the standardized video data. The extraction process combines frame difference and edge energy thresholding to calculate the pixel difference between adjacent frames. When the difference exceeds a preset threshold, it is marked as a scene transition point, generating a set of keyframes. After acquiring the keyframes, scene segmentation techniques are used to perform edge detection and region division on the image. Specifically, this includes: using the Canny edge detection algorithm to identify the outlines of objects in the image; using region growing to merge pixel groups with similar textures and colors into semantic regions; and using hierarchical clustering to combine related regions according to spatial adjacency and feature similarity to form video scene combination features. These features include the spatial distribution, shape features, and visual attribute information of the main objects in the image. A semantic understanding engine is used to classify thematic elements based on the combined features of video scenes, establishing a visual element dictionary that includes basic element categories such as people, scenes, and objects. Then, feature matching is used to map the visual regions in the image to the element categories in the dictionary, identifying the specific content objects contained in the scene. Next, the spatial positional relationships and interaction states between objects are analyzed to extract the contextual relevance between elements. Scene semantic description tags are generated, and a scene semantic mapping set is constructed. This mapping set records the main visual elements, element categories, element combination relationships, and overall semantic expression contained in each scene, providing a semantic understanding foundation for visual content for subsequent regulatory judgment.

[0054] Audio data streams are separated from multi-dimensional film and television content datasets, and independent sound information is obtained by extracting audio channels. Spectral analysis is performed on the audio data, and a short-time Fourier transform (STFT) is used to convert the time-domain signal into a time-frequency domain representation. Then, MFCC (Melbourne Frequency Cepstral Coefficients) features are extracted, which simulate the human ear's perception of sound. Next, spectral analysis is performed to identify energy distribution, pitch variations, and rhythmic features in the audio. Waveform decomposition techniques are used to separate complex audio into components of different frequency bands, forming an emotional audio feature spectrum. This feature spectrum contains the frequency distribution, energy intensity, and modulation characteristics of the audio signal at different time segments, providing fundamental data features for emotion recognition. Multi-level labeling processing is performed on the emotional audio feature spectrum to establish an emotional acoustic parameter model. This model includes the mapping relationship between key acoustic features such as pitch, speech rate, volume, and rhythm changes and emotional states. Then, the parameters in the audio feature spectrum are quantified to extract the emotional expression features. Next, the emotional intensity is divided into multiple levels, such as low intensity, medium intensity, and high intensity, using emotional intensity gradient quantization technology. Based on the acoustic feature combination pattern, the audio segments are labeled with emotional categories, such as basic emotional types like anger, joy, sadness, and calmness, resulting in multi-level emotional labeling data. This data includes information on the emotional category, emotional intensity level, and emotional change trend of each time period of the audio.

[0055] Text content, including dialogue subtitles and narration, is extracted from a multidimensional film and television content dataset. Semantic network technology is used to analyze the text, performing Chinese word segmentation and part-of-speech tagging to identify basic language units. Then, the importance weights of words are calculated using the TF-IDF algorithm to extract key topic words. Next, a word co-occurrence network is established to analyze the semantic association strength between words. A value orientation dictionary is introduced to perform value orientation analysis on the text content, identifying positive, negative, or neutral values ​​to form a cultural value orientation index. This index quantifies the intensity and orientation of various value concepts expressed in the text content.

[0056] The scene semantic mapping set, multi-level sentiment marker data, and cultural value orientation indicators are integrated and associated along the time axis to establish a unified time benchmark and ensure the alignment of the three types of data in the time dimension. Then, the multimodal data within the same time window are associated and matched using the time window sliding method. Next, a temporal association matrix between content elements is constructed to record the mutual influence relationship between visual, auditory, and textual content. A temporal association content marker list is generated, which records the comprehensive content characteristics of each segment of the film and television work in chronological order, including visual scene description, emotional expression state, and cultural value orientation.

[0057] Taking a certain animated film for teenagers as an example, after extracting keyframe sequences from its multi-dimensional film and television content dataset, scene segmentation was used to identify character images, environmental scenes, and interactive actions in the scene, generating video scene combination features. Semantic understanding analysis was performed on these features to identify key visual elements such as "fighting scenes" and "weapon usage," forming a scene semantic mapping set. Simultaneously, spectral features were extracted from the audio, analyzing characteristics such as high pitch and fast tempo to generate an emotional audio feature spectrum. Through emotional intensity quantification, these audio features were labeled into emotional categories such as "moderate tension" and "mild fear," forming multi-level emotional labeling data. Keywords were extracted from the subtitle text, analyzing the value orientation expressions such as "courage" and "justice" contained within, generating cultural value orientation indicators. Integrating these three types of data along a timeline revealed that in a certain battle scene, the visuals presented intense fighting, the audio expressed tension, and the text emphasized the value of "fighting for justice." This comprehensive analysis of multimodal data formed a time-series related content labeling list.

[0058] In one specific embodiment, the process of executing step S103 may specifically include the following steps:

[0059] (1) Filter the elements of the time-related content tag list, and use key point extraction technology to identify and separate sensitive elements and core content in film and television content to obtain a set of key content elements;

[0060] (2) The key elements of the content are quantitatively scored according to the social impact intensity standard. The social sensitivity and guiding role of the content elements are evaluated by referring to public opinion, and the social impact intensity index is obtained.

[0061] (3) Construct a propagation link model based on the set of key content elements, and calculate the diffusion speed and coverage of the content in different groups of people through user behavior analysis to obtain the propagation penetration depth value;

[0062] (4) Psychological acceptance is measured based on the set of key elements of the content. The psychological acceptance of different age groups and cultural backgrounds is analyzed by matching audience profiles to obtain the psychological acceptance level.

[0063] (5) The social impact intensity index, the depth of communication penetration and the psychological acceptance level are cross-correlated in multiple dimensions. The three indicators are dynamically weighted by the weight allocation matrix to obtain the comprehensive score data.

[0064] (6) The comprehensive scoring data is mapped in a hierarchical manner, and the content scoring is segmented by adjusting the threshold to establish a variable weight scoring system and obtain a dynamic weight content scoring table.

[0065] Specifically, key point extraction technology is used to identify and separate sensitive elements and core content in film and television content. This technology employs a variant of the TF-IDF algorithm, weighting each content element in the time-series markers according to its frequency and distinguishability to select key content points with regulatory value. The process calculates the feature weight value of each content element, ranks them according to their importance in the overall content, sets a threshold, and extracts elements with weights exceeding the threshold to form a set of key content elements. This set includes video scenes, audio emotions, and textual expressions requiring focused monitoring; these elements are assigned unique identifiers for subsequent scoring and tracking. The set of key content elements is then quantitatively scored according to the social impact intensity standard, which refers to the degree to which content may guide, stimulate, or influence a social group. The quantification process uses social sentiment reference technology, matching content elements with cases in a historical public opinion database to analyze the intensity of social reactions triggered by similar content in the past. The evaluation dimensions include sensitivity, controversy, and guidance; each dimension is scored using a five-level scoring standard, and a weighted average is used to calculate the social impact intensity index. This index reflects the extent to which content elements may influence social values, public order, or group behavior, providing an important reference for subsequent content classification.

[0066] A propagation chain model is constructed based on a set of key content elements to analyze the propagation characteristics of content among different groups. The propagation chain model is based on information diffusion theory and uses user behavior analysis technology to mathematically model the content propagation process. The model calculation process can be expressed as follows:

[0067]

[0068] in, Indicates the propagation penetration depth value; Indicates the number of users; This represents the weight coefficient of the g-th group; Indicates the group influence coefficient; This represents the number of users in the g-th group; This represents the propagation capability index of the i-th user; This represents the time decay coefficient of the g-th population; Indicates the duration of transmission; Indicates the number of dissemination channels; This represents the influence coefficient of the j-th channel; This represents the coverage area of ​​the j-th channel; This represents the maximum channel influence coefficient; This represents the total target coverage area. The formula comprehensively considers the characteristics of different groups, user dissemination behavior, time decay factors, and dissemination channel effects to calculate the content's penetration depth, reflecting the speed and scope of content spread.

[0069] Psychological acceptability is measured based on a set of key content elements, and audience profiling matching technology is used to analyze the psychological acceptance levels of different age groups and cultural backgrounds. The psychological acceptability measurement constructs a multi-dimensional audience profiling database, including psychological characteristic models of groups with different age groups, cultural backgrounds, and education levels. Then, the matching degree between key content elements and each group's profile is calculated to assess the potential psychological impact of the content on different groups. Acceptance levels are categorized based on the matching results, typically into four levels: high acceptability, moderate acceptability, low acceptability, and inappropriateness. Psychological acceptability levels provide an audience-oriented reference for regulatory decisions, helping to determine the appropriate target audience for content. The social impact intensity index, dissemination penetration depth value, and psychological acceptability level are cross-correlated in multiple dimensions, and the three indicators are dynamically weighted using a weighting matrix. The weighting matrix dynamically adjusts the weights of each indicator based on content type, dissemination environment, and regulatory focus, making the scoring results more consistent with actual regulatory needs. The weighted calculation process uses a weighted geometric average method, comprehensively considering the interaction relationship of the three indicators to generate comprehensive scoring data. This data has multidimensional attributes, including not only the overall score but also the independent scores and weights of each dimension, providing comprehensive data support for subsequent classification processing.

[0070] The comprehensive scoring data is mapped into different levels, and content scores are segmented using threshold adjustment technology. The mapping defines different scoring ranges, including categories such as suitable for all ages, suitable for teenagers, suitable for adult supervision, and restricted. The threshold ranges for each level are then dynamically adjusted based on content characteristics to ensure the scoring results comply with current regulatory standards. A variable-weight scoring system is established, with differentiated scoring strategies for different types of content, resulting in a dynamic-weight content scoring table. This scoring table assigns detailed rating information to each content element, including suitable audience, restrictions, and adjustment suggestions, providing a direct basis for subsequent regulatory decisions.

[0071] Taking a reality talent show as an example, the list of time-related content tags was filtered to identify sensitive elements such as "intense competition among contestants," "severe comments from judges," and "emotional breakdowns of contestants," forming a set of key content elements. The social impact intensity of these elements was assessed. By comparing the social discussion intensity triggered by similar content in the past, the sensitivity of the "emotional breakdown" scenario was rated at level 4 (out of 5), the controversy at level 3, and the guiding nature at level 4, resulting in a social impact intensity index of 3.8. A dissemination chain model was constructed to analyze the dissemination characteristics of this content among adolescents. Considering the high activity level of this group (γ=0.85) and the high dissemination coefficient of short video platforms (φ=0.92), the dissemination penetration depth was calculated to be 0.73 using the above formula. Psychological acceptance was measured, revealing a low relevance of the content to the 12-16 year old adolescent group, resulting in a psychological acceptance rating of "low acceptance." The three indicators were input into a weighting matrix. Considering the program's primary target audience is teenagers, the psychological acceptance indicator was assigned a high weight of 0.5, while the social impact and dissemination depth indicators were assigned weights of 0.3 and 0.2 respectively, resulting in a comprehensive score of 3.62. Through threshold mapping, this score falls into the "Watch with caution, teenagers" range (3.5-4.0), and is marked as "Parental guidance required" in the dynamic weighted content scoring table. Specific modification suggestions were generated, including reducing the duration of emotional breakdown scenes and increasing positive guidance content.

[0072] In one specific embodiment, the process of executing step S104 may specifically include the following steps:

[0073] (1) Extract high-weight content tags from the dynamic weighted content scoring table, and perform matching analysis on related content in different film and television works through similarity calculation to obtain a content association link library;

[0074] (2) Construct a timeline propagation sequence of film and television works based on the content association link library, and track the evolution trend of the content in different periods through time series data association to obtain a time series propagation trajectory map;

[0075] (3) Spatial dimension expansion of the time-series propagation trajectory map, and visualization of the propagation density and coverage of content in different regions through regional distribution mapping, to obtain regional propagation heat distribution;

[0076] (4) Construct a network topology for the content-related link library according to the propagation relationship, and identify the key nodes and propagation paths of content propagation through node connection analysis to obtain a mesh propagation relationship network;

[0077] (5) Conduct an influence assessment on the network of propagation relationships. Quantitatively analyze the radiation range and influence of each node in the propagation network through influence diffusion calculation to obtain content influence assessment data.

[0078] (6) Integrate the time-series propagation trajectory map, regional propagation heat distribution and content influence assessment data in multiple dimensions to construct a three-dimensional model of propagation influence and obtain a three-dimensional content influence map.

[0079] Specifically, high-weight content tags refer to content elements whose weight values ​​exceed a preset threshold. These elements typically represent content features with high regulatory value. The extraction process sorts the dynamic weighted content scoring table, rearranging the weight values ​​from highest to lowest. Then, a filtering threshold is set, generally selecting the top 30% of weight values ​​as high-weight tags. These high-weight tags are extracted, retaining their complete attribute information. After obtaining the high-weight content tags, similarity calculations are used to match and analyze related content in different film and television works. The similarity calculation uses a cosine similarity algorithm to calculate the cosine of the angle between the feature vectors of high-weight tags in different works; the closer this value is to 1, the higher the content similarity. A similarity matrix is ​​used to construct the relationships between content elements, forming a content association link library. This library records the degree and type of association between content elements in different film and television works. Based on the content association link library, a timeline propagation sequence of film and television works is constructed. This step uses time-series data association technology to arrange the content propagation process in chronological order. The temporal data correlation method collects content publication and dissemination time information from various platforms; then, these time points are arranged chronologically to form a dissemination timeline; next, content-related link information is mapped onto the timeline, marking the time nodes of content appearance, dissemination peaks, and declines; through time window sliding analysis, the changes in dissemination popularity within each time period are calculated, generating a continuous dissemination trend curve. This data is integrated to form a temporal dissemination trajectory map, which visually displays the dissemination patterns and evolution trends of content over time, providing a temporal benchmark for subsequent spatial analysis.

[0080] The temporal propagation trajectory map is spatially expanded by using geographic distribution mapping technology to visualize the propagation data in geographic space. Geographic distribution mapping divides the target area into geographic grids, typically based on administrative divisions or population density. Then, content propagation data for each region is collected, including views, shares, and interactions. This data is then mapped onto the geographic grid according to location information. A heatmap rendering algorithm highlights data-dense areas, generating a propagation heatmap. This process creates a geographic propagation heat map, visually demonstrating the propagation density and coverage of content in different geographic regions, revealing regional differences in propagation and the distribution of hotspots. The content-related link library is then used to construct a network topology based on propagation relationships. This process employs complex network modeling techniques, treating content elements as network nodes and the propagation relationships between content as network connections. The network topology construction defines node attributes, including content type, influence level, and timestamp; then, it defines connection weights based on association strength, with higher strength resulting in greater weights; next, it establishes connections according to the actual propagation relationships between nodes, forming the initial network structure; through node connection analysis techniques, it calculates network characteristics such as node connectivity, centrality, and clustering coefficients to identify key nodes and main propagation paths in the network. This process forms a mesh-like propagation relationship network, which reveals the structural characteristics and key influence points of content propagation.

[0081] Influence assessment is performed on a network of propagation relationships, and the reach and impact of each node are quantitatively analyzed through influence diffusion calculation. The influence diffusion calculation is based on an information cascade model, and its core formula can be expressed as:

[0082]

[0083] in, This represents the influence index of node v; Weights are the inherent attributes of the node. This represents the content weight value of node v; The weights are passed on to the nodes; N(v) is the set of nodes directly connected to node v. Indicates the connection weight from node u to v; Indicates the out-degree of node u; The attenuation coefficient; The network distance between nodes; Weighting based on content feedback; This indicates the intensity of user feedback received by node v; The maximum feedback intensity value; K is the number of influencing factors; This represents the correction coefficient of the j-th external influence factor for node v. Represents nodes Directly connected upstream nodes The influence index is calculated using a formula that comprehensively considers various factors such as inherent node attributes, network location, connectivity, and user feedback to fully assess a node's influence within the dissemination network. By applying this calculation model to each node, content influence assessment data is generated.

[0084] By integrating time-series dissemination trajectory maps, regional dissemination heatmaps, and content influence assessment data, a multi-dimensional model of dissemination influence is constructed. This multi-dimensional integration employs data fusion technology to unify data from different dimensions onto a single coordinate system. The integration process establishes a three-dimensional coordinate system for time, space, and network influence; then, data from each dimension is mapped onto the corresponding coordinate axes; next, interpolation algorithms are used to fill in data gaps to ensure data continuity; finally, a three-dimensional data visualization model is constructed, generating a three-dimensional content influence map. This map provides a comprehensive display of the three dimensions of time, space, and influence.

[0085] In one specific embodiment, the process of executing step S105 may specifically include the following steps:

[0086] (1) Extract high-risk regulatory points from the three-dimensional content impact map, identify and screen high-risk regulatory points through risk weight calculation, and obtain a list of key regulatory points;

[0087] (2) The list of key regulatory points is sorted by hierarchical priority, and the processing order and resource allocation ratio of different regulatory points are determined by the urgency assessment to obtain the regulatory priority sequence table;

[0088] (3) Based on the regulatory priority sequence list, for high-priority content nodes, the differences between non-compliant content and compliance standards are accurately located through content comparison analysis, and a precise list of modification suggestions is obtained;

[0089] (4) Based on the content propagation characteristics in the list of key regulatory points, a tiered restriction scheme is formulated for content with different impact ranges through a gradient control strategy, resulting in a gradient restriction measure scheme;

[0090] (5) Based on the content characteristics and audience information of the list of key regulatory points, the scope of the target audience for the content is accurately defined by mapping the grading standards, and a personalized grading requirement table is obtained.

[0091] (6) Integrate and link the precise modification suggestion list, the gradient restriction measures plan and the personalized graded requirements table to obtain the intelligent supervision plan.

[0092] Specifically, risk weighting calculation technology is used to assess and screen content nodes for risk. This process sets multi-dimensional risk assessment indicators for each content node in the three-dimensional content impact map, including dimensions such as content sensitivity, dissemination speed, target audience characteristics, and influence intensity. Then, weight coefficients are assigned to each dimension to differentiate the risk factors for different types of content. Next, a comprehensive risk score is calculated by multiplying each dimension's score by its corresponding weight and summing the results. A risk threshold is set, and content nodes with risk scores exceeding the threshold are selected, forming a set of high-risk monitoring points. These high-risk points are further analyzed for their network location and scope of influence, focusing on identifying nodes at the core of the dissemination network with broad influence, generating a list of key monitoring points. This list includes content nodes requiring priority handling and their risk characteristics. The key monitoring point list is then prioritized hierarchically, and an urgency assessment technique is used to scientifically rank the processing order of different monitoring points. The urgency assessment establishes a multi-factor scoring standard, considering the timeliness of the content, the speed of its spread, the potential harm, and resource consumption. Each regulatory point is then scored according to the scoring standard, generating a quantifiable urgency value. Next, they are ranked from highest to lowest urgency, categorized into four levels: emergency handling, priority handling, routine handling, and low priority. Based on the ranking, processing resources are allocated, and the processing timeframe and resource allocation ratio for each regulatory point are determined. This process forms a regulatory priority sequence table.

[0093] Based on a regulatory priority sequence list, content comparison analysis is performed on high-priority content nodes. This process uses precise comparison technology to identify discrepancies between non-compliant content and compliance standards. The content comparison analysis constructs a regulatory standard template library, covering regulatory requirements and compliance standards for different content types. Then, high-priority content nodes are compared with their corresponding standard templates to identify violations and non-compliant content. Next, each violation is analyzed in depth to determine the degree, type, and direction of adjustment. Specific modification suggestions are generated, including detailed operational guidelines such as content deletion, expression replacement, and scenario adjustment. This process forms a precise list of modification suggestions, providing content creators with clear guidance for compliance adjustments and improving the accuracy and operability of regulation. Based on the content dissemination characteristics in the key regulatory point list, a tiered restriction scheme is developed for content with different impact ranges using gradient control strategies. The tiered control strategy categorizes content into control levels based on its scope of influence, typically into four levels: full restriction, partial restriction, guided viewing, and no restriction. Differentiated restriction measures are then designed for each level, including playback permission control, time-based restrictions, geographical restrictions, and user screening techniques. Next, the most suitable restriction level and specific measures are matched to each monitoring point, forming a complete tiered restriction scheme that provides targeted control strategies for content with different characteristics, achieving refined supervision.

[0094] Based on the content characteristics and audience information of the key regulatory point list, a tiered standard mapping technology is used to accurately define the target audience for the content. The tiered standard mapping establishes a content tiering system, including multiple levels such as suitable for all ages, parental guidance, caution for teenagers, and adult-only. Then, it analyzes the matching degree between content characteristics and audience psychological characteristics, assessing the acceptance ability of different age groups and cultural backgrounds. Next, based on the matching results, it assigns the most suitable audience range to each content node, generating a personalized tiered requirement table that clearly defines the target audience, viewing conditions, and warning requirements for various types of content, providing scientific guidance for user selection. The precise modification suggestion list, tiered restriction measures, and personalized tiered requirement table are integrated and linked to form a complete intelligent regulatory solution. The integration and linking process involves structuring the three lists to ensure data format uniformity and field standardization. Then, it establishes a mapping relationship between the three lists to ensure consistency in modification suggestions, restriction measures, and tiered requirements for each regulatory point. Next, it prioritizes the integrated solution according to regulatory priorities, forming an orderly execution plan. Finally, it generates a complete intelligent regulatory solution document, including regulatory objectives, specific measures, execution timelines, and expected results, providing comprehensive guidance for regulatory implementation.

[0095] Taking a popular romantic reality show as an example, risk nodes were extracted from its multi-dimensional content impact map, revealing high-risk content characteristics such as "excessive arguments among guests," "disclosure of emotional privacy," and "biased value orientation." Through risk weight calculation, considering that "biased value orientation" possesses both high content sensitivity and strong dissemination power, reaching a broad youth audience, its comprehensive risk score reached 1.8 times the risk threshold, thus identifying it as a key regulatory point and adding it to the list of key regulatory points. An urgency assessment of the list was conducted; due to the show's peak broadcast period and high social media discussion, "biased value orientation" was marked as an emergency and placed at the top of the regulatory priority list. Through content comparative analysis, the show's expressions of values ​​such as "money worship" and "gamification of relationships" were compared with the "positive and healthy emotional view" in the regulatory standard template, precisely identifying six core problematic passages. Specific modification suggestions were generated, including "removing guests' statements about judging love based on material conditions" and "adding expert commentary with positive emotional guidance," forming a precise list of modification suggestions. Based on the program's dissemination characteristics, it was found to spread rapidly and exhibit strong imitation tendencies among teenagers. A tiered control strategy was adopted, classifying it into a "guided viewing" level and implementing restrictions such as "adding opening prompts" and "setting broadcast times after 10 PM," forming a tiered restriction plan. Through a rating standard mapping, the program was defined as "suitable for viewers aged 15 and above, with parental accompaniment required for ages 12-15," with detailed explanations and precautions provided in a personalized rating requirement table. Integrating these three lists generated a complete intelligent regulatory solution encompassing modification requirements, broadcast restrictions, and age ratings. This provided program producers with clear directions for compliance adjustments and set specific restrictions for broadcasting platforms, achieving precise content supervision of the variety show.

[0096] In one specific embodiment, the process of executing step S106 may specifically include the following steps:

[0097] (1) Implement and deploy the intelligent supervision plan, and collect and record the time nodes and effect data of the implementation process of the measures through execution tracking to obtain the supervision execution data flow;

[0098] (2) Extract key performance indicators from the regulatory execution data stream, and quantitatively compare the difference between the expected regulatory effect and the actual execution result through target deviation measurement to obtain regulatory effect deviation data;

[0099] (3) Root cause analysis is conducted on the data of deviations in regulatory effectiveness. By tracing the source, the key factors and influencing conditions that cause the deviations are identified and located, and a deviation factor attribution table is obtained.

[0100] (4) Formulate correction strategies based on the deviation factor attribution table, determine the correction direction and adjustment range of different deviation factors through correction force calculation, and obtain the regulatory parameter adjustment plan;

[0101] (5) Optimize and verify the regulatory parameter adjustment plan, test and evaluate the effectiveness of the adjusted regulatory parameters through small-scale trials, and obtain a parameter optimization verification report;

[0102] (6) Extract and analyze the effective parameters in the parameter optimization verification report to obtain the self-optimization monitoring parameter set.

[0103] Specifically, execution tracking technology is used to monitor the entire implementation process of regulatory measures. Execution tracking breaks down the intelligent regulatory solution into specific execution tasks, assigning responsibility, deadlines, and acceptance criteria to each task. Then, regulatory execution data collection points are established, with data collection triggers set at key time nodes to record the execution status, progress, and effects of each task. Next, a multi-channel data collection mechanism gathers execution effect data from multiple dimensions, including content platforms, user feedback, and public opinion. This collected time-series data is then organized and stored in a unified format, forming a complete regulatory execution data stream. This data stream contains records of the entire process from the issuance to the implementation of regulatory measures. Key performance indicators are extracted from the regulatory execution data stream, and target deviation measurement technology is used to accurately compare the expected results with the actual execution outcomes. Key performance indicators (KPIs) are extracted to define the evaluation dimensions of regulatory effectiveness, including compliance level, dissemination control effectiveness, and user feedback indicators. Then, data items related to these evaluation dimensions are selected from the regulatory execution data stream to extract core indicator values. Next, these actual indicator values ​​are compared with the preset target values ​​in the intelligent regulatory solution to calculate the deviation rate and direction. The deviation data for each dimension is then structured and organized to form regulatory effectiveness deviation data. This deviation data intuitively demonstrates the differences in the effectiveness of regulatory implementation, providing precise data basis for subsequent optimization and adjustments.

[0104] Root cause analysis was conducted on data regarding deviations in regulatory effectiveness, using source tracing techniques to deeply uncover the key factors contributing to these deviations. The root cause analysis prioritized the deviation data according to their severity, focusing on indicators with larger deviations. Then, the "5W2H" method was used to decompose each significant deviation, clarifying its specific manifestations and scope of impact. Next, causal chain analysis was employed to trace the chain of causes leading to the deviations, identifying the fundamental factors. These key factors were then categorized and organized according to their impact dimensions and degrees, constructing a deviation factor attribution table. This attribution table clearly presents the various factors leading to regulatory effectiveness deviations and their mechanisms of action. Based on the deviation factor attribution table, correction strategies were developed, scientifically planning adjustment schemes for different deviation factors using correction intensity calculation techniques. The correction strategy development prioritized the factors in the attribution table, determining the order of adjustment. Then, for each high-priority factor, targeted adjustment measures were designed, clarifying the direction and means of adjustment. Next, correction intensity calculations determined the magnitude of the adjustment based on the degree of deviation and the factor's influence weight, avoiding excessive or insufficient adjustments. Finally, all adjustment measures were integrated to form a complete regulatory parameter adjustment plan. The plan includes specific parameter correction values, adjustment steps, and expected results.

[0105] The regulatory parameter adjustment plan was optimized and validated, and the effectiveness of the adjusted parameters was scientifically evaluated through small-scale pilot testing. The optimization and validation process involved selecting representative content samples and user groups to construct a pilot environment. The adjusted regulatory parameters were then applied to the pilot samples to simulate real-world regulatory scenarios. Multiple monitoring points were set up to collect data on the effects during the trial, including user reactions, content changes, and dissemination effects. The pilot data was then comprehensively analyzed to evaluate the effectiveness and feasibility of the parameter adjustments, resulting in a parameter optimization and validation report. The report detailed the results of each parameter adjustment trial.

[0106] The effective parameters in the parameter optimization verification report are integrated and refined to construct a self-improving regulatory parameter system through parameter extraction and analysis techniques. Parameter extraction and analysis screens the parameters in the verification report according to their effectiveness, retaining those adjustments that effectively improve regulatory outcomes. These effective parameters are then merged with existing parameters to update the regulatory parameter library. Next, a correlation model between parameters is constructed to clarify the synergistic effects and constraints between different parameters, forming a complete self-optimizing regulatory parameter set. This parameter set has dynamic adjustment capabilities, automatically optimizing parameter configurations based on actual regulatory results to achieve continuous improvement of the regulatory system.

[0107] Taking a certain online animated series as an example, after implementing and deploying its intelligent supervision solution, execution data was collected at key nodes such as platform launch, user interaction, and social media dissemination using execution tracking technology. This recorded the completion status of modified content, the execution status of restrictive measures, and user viewing data, forming a complete regulatory execution data stream. Key performance indicators were extracted from this data stream, including the compliance rate of violent scene modifications, the success rate of access control for underage users, and the proportion of negative discussions on social media. Comparison with expected targets revealed that the success rate of access control for underage users was 20 percentage points lower than the expected target, indicating a deviation in regulatory effectiveness. Root cause analysis was conducted to address this significant deviation. Through source tracing, three key factors were identified: an imperfect age verification bypass mechanism, insufficient platform enforcement, and unclear regulatory prompts. A detailed attribution table of deviation factors was constructed. Based on this attribution table, a correction strategy was developed. A dual verification process was designed for the age verification mechanism, a reward and punishment mechanism was added to the platform's enforcement efforts, and user-friendly improvements were made to the regulatory prompts, resulting in a regulatory parameter adjustment plan. The proposed solution was optimized and validated through small-scale trials using three popular episodes of the animated series and multiple age groups of test users. The effectiveness of the adjusted regulatory parameters was examined, revealing that the dual verification mechanism reduced the bypass rate by underage users by 85%, while the optimized prompts received higher user acceptance. A detailed parameter optimization validation report was generated. Effective parameters were extracted from the report, including optimized age verification parameter configurations, platform execution supervision thresholds, and user-friendly prompt templates. These were integrated with the original regulatory parameters to construct a self-regulating and self-optimizing regulatory parameter set, establishing a continuously optimizing parameter foundation for the long-term regulation of this type of animated content.

[0108] The above describes the film and television regulatory analysis method based on data mining in the embodiments of this application. The following describes the film and television regulatory analysis system based on data mining in the embodiments of this application. Please refer to [link / reference]. Figure 2 One embodiment of the film and television supervision and analysis system based on data mining in this application includes:

[0109] The acquisition module is used to collect panoramic data from film and television works on video platforms. Through cross-media transcoding, video, audio, and text information are processed in a layered and standardized manner to obtain a multi-dimensional film and television content dataset.

[0110] The extraction module is used to extract scene semantics from video frames, perform multi-level emotion recognition on audio, and perform cultural value orientation analysis on text based on the multi-dimensional film and television content dataset, so as to obtain a list of time-related content tags.

[0111] The scoring module is used to cross-score key elements based on the time-related content tag list, according to the intensity of social impact, the depth of dissemination penetration, and the psychological acceptance of the audience, to obtain a dynamic weighted content scoring table.

[0112] The analysis module is used to analyze the network propagation path of film and television works by correlation tracking based on the dynamic weighted content scoring table, and obtain a three-dimensional content influence map.

[0113] The sorting module is used to prioritize the regulatory points according to the three-dimensional content influence map and through decision rules to obtain an intelligent regulatory solution that includes precise modification suggestions, gradient restriction measures and personalized classification requirements.

[0114] The analysis module is used to perform deviation correction analysis on the implementation effect of the regulatory measures based on the intelligent regulatory scheme and through effect feedback, so as to obtain a self-optimizing regulatory parameter set.

[0115] Through the collaborative efforts of the aforementioned components, and by employing panoramic data acquisition and cross-media transcoding, the hierarchical and standardized organization of video, audio, and text information is achieved. This overcomes the fragmented data acquisition and inconsistent formatting issues inherent in traditional film and television regulation, establishing a multi-dimensional film and television content dataset and providing a high-quality data foundation for subsequent analysis. Furthermore, by utilizing content feature recognition technology to extract scene semantics from video frames, perform multi-level emotion recognition on audio, and analyze cultural value orientations in text, the limitations of single-media analysis are overcome. This enables semantic association of cross-media content, forming a temporally related content tagging list and enhancing the ability to identify obscure and harmful content. Finally, based on a tiered scoring mechanism, key elements are categorized according to their social impact intensity and dissemination penetration depth. By cross-scoring with audience psychological acceptance, an innovative multi-dimensional evaluation system was introduced, generating a dynamic weighted content scoring table to make regulatory assessment more comprehensive and objective. Through correlation tracking technology, the network dissemination path of film and television works was analyzed, revealing the spatiotemporal dynamic characteristics of content dissemination and constructing a three-dimensional content impact map, providing a basis for precise regulation. Based on decision-making rules, regulatory points were prioritized, forming an intelligent regulatory scheme that includes precise modification suggestions, tiered restrictions, and personalized grading requirements, achieving optimized allocation and differentiated processing of regulatory resources. Through an effect feedback mechanism, deviation correction analysis of the implementation effect of regulatory measures was conducted, generating a self-optimizing regulatory parameter set, enabling the system to have self-learning and continuous optimization capabilities. It is particularly worth emphasizing that this solution fully leverages the technological contributions of artificial intelligence algorithms and models in specific functions and application areas. For example, the deep learning visual recognition model used in scene semantic extraction can accurately identify sensitive content in complex scenes; the sentiment computing algorithm used in multi-level sentiment recognition can capture the emotional changes implied in sound; the natural language processing technology introduced in cultural value orientation analysis can understand the deep semantics of text; the network propagation algorithm used in propagation path analysis can predict the scope of influence and propagation trend of content; and the multi-objective optimization algorithm applied in regulatory decision-making can balance regulatory effectiveness and resource consumption. The organic combination of these algorithmic features makes this solution demonstrate significant technical advantages in the field of massive multimedia content supervision, improving both the accuracy and efficiency of supervision while reducing labor costs.

[0116] Reference Figure 3 This invention also provides a computer device, which can be a server, and its internal structure can be as follows: Figure 3As shown, the computer device includes a processor, memory, display screen, input device, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores the data corresponding to this embodiment. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements the above-described method.

[0117] Those skilled in the art will understand that Figure 3 The structures shown are merely block diagrams of some structures related to the present invention and do not constitute a limitation on the computer devices on which the present invention is applied.

[0118] An embodiment of the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method. It is understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0119] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the present invention and embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0120] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0121] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0122] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A data mining-based film and television supervision analysis method, characterized in that, The film and television supervision analysis method based on data mining comprises: Panoramic data collection is performed on film and television works of a video platform, layered standardized arrangement is performed on video, audio and text information through cross-media transcoding processing, and a multi-dimensional film and television content dataset is obtained; According to the multi-dimensional film and television content dataset, scene semantic extraction is performed on video frames through content feature recognition, multi-level emotion recognition is performed on audio, and cultural value tendency analysis is performed on text, and a time sequence correlation content marking list is obtained; Based on the time sequence correlation content marking list, cross scoring is performed on key elements according to social influence intensity, transmission penetration depth and audience psychological acceptance through hierarchical scoring, and a dynamic weight content scoring table is obtained; According to the dynamic weight content scoring table, a network transmission path analysis is performed on the film and television works through correlation tracking, and a stereoscopic content influence graph is obtained, comprising: extracting high weight content tags from the dynamic weight content scoring table, performing matching analysis on the correlation content in different film and television works through similarity calculation, and obtaining a content correlation link library; constructing a film and television work time axis transmission sequence according to the content correlation link library, tracking the evolution trend of the content in different periods through time sequence data correlation, and obtaining a time sequence transmission trajectory graph; expanding the space dimension of the time sequence transmission trajectory graph, visualizing the transmission density and coverage range of the content in different regions through regional distribution mapping, and obtaining a regional transmission heat distribution; constructing a network topology of the content correlation link library according to the transmission relationship, identifying the key nodes and transmission paths of the content transmission through node connection analysis, and obtaining a network transmission relationship network; performing influence evaluation on the network transmission relationship network, quantitatively analyzing the radiation range and influence degree of each node in the transmission network through influence diffusion calculation, and obtaining content influence evaluation data; multi-dimensionally integrating the time sequence transmission trajectory graph, the regional transmission heat distribution and the content influence evaluation data, constructing a transmission influence stereoscopic model, and obtaining a stereoscopic content influence graph; Wherein, the influence evaluation is performed on the network transmission relationship network, and the radiation range and influence degree of each node are quantitatively analyzed through influence diffusion calculation; the influence diffusion calculation is based on an information cascade model, and the core formula can be expressed as: wherein, represents the influence index of node v; is the node inherent attribute weight; represents the content weight value of node v; is the association transmission weight; N(v) is the node set directly connected with node v; represents the connection weight of node u to v; represents the out-degree of node u; is the attenuation coefficient; is the network distance between nodes; is the content feedback weight; represents the user feedback intensity obtained by node v; is the maximum feedback intensity value; K is the number of influence factors; represents the correction coefficient of the jth external influence factor to node v; represents the influence index of the upstream node directly connected with node v; . According to the stereoscopic content influence graph, a priority order of supervision points is obtained through decision rules, and an intelligent supervision scheme containing accurate modification suggestions, gradient restriction measures and individualized hierarchical requirements is obtained; According to the intelligent supervision scheme, deviation correction analysis is performed on the execution effect of the supervision measures through effect feedback, and a self-optimizing supervision parameter set is obtained.

2. The data mining based film and television supervision analysis method according to claim 1, characterized in that, The panoramic data collection is performed on film and television works of a video platform, layered standardized arrangement is performed on video, audio and text information through cross-media transcoding processing, and a multi-dimensional film and television content dataset is obtained, comprising: Deep penetration collection is performed on film and television content of multi-source video platforms through an adaptive capture network, and full-spectrum film and television original data is obtained; Digital fingerprint recognition is performed on the full-spectrum film and television original data, and intelligent detection is performed on video frame rate, audio sampling and text encoding, and a media feature graph is obtained; According to the media feature map, the heterogeneous media format is processed by spatio-temporal synchronization through cross-coding conversion to obtain a unified standard media stream; The unified standard media stream is subjected to semantic segmentation, and the video, audio and text are precisely sliced through key frame positioning technology to obtain a semantic association slice set; The semantic association slice set is subjected to hierarchical classification through multi-dimensional cross mapping to construct a content association network and obtain a structured content matrix; The structured content matrix is written into a data warehouse through distributed storage to establish a multi-dimensional retrieval channel and obtain a multi-dimensional video content data set.

3. The data mining based film and television supervision analysis method according to claim 1, characterized in that, According to the multi-dimensional video content data set, scene semantic extraction is performed on video frames, multi-level emotion recognition is performed on audio, and cultural value tendency analysis is performed on text through content feature recognition to obtain a time sequence associated content marking list, including: Video key frame sequences are extracted from the multi-dimensional video content data set, and edge detection and region division are performed on the video picture through scene segmentation technology to obtain video scene combination features; The video scene combination features are subjected to theme element classification through a semantic understanding engine to extract key visual elements and context relationships in the scene to obtain a scene semantic mapping set; Audio data streams are separated from the multi-dimensional video content data set, and the audio emotion features are subjected to waveform decomposition through frequency spectrum analysis to obtain an emotion audio feature spectrum; The emotion audio feature spectrum is subjected to multi-level marking, and the tone, speed and rhythm changes are subjected to emotion classification through emotion intensity gradient quantization to obtain multi-level emotion marking data; Text content is extracted from the multi-dimensional video content data set, and theme word extraction and value orientation analysis are performed on the text through a semantic network to obtain a cultural value tendency index; The scene semantic mapping set, the multi-level emotion marking data and the cultural value tendency index are integrated and associated according to a time axis to establish a time sequence mapping relationship between content elements to obtain a time sequence associated content marking list.

4. The data mining based movie censorship analysis method of claim 1, wherein, Based on the time sequence associated content marking list, cross scoring is performed on key elements according to social influence intensity, transmission penetration depth and audience psychological acceptance through hierarchical scoring to obtain a dynamic weight content scoring table, including: The time sequence associated content marking list is subjected to element screening, and sensitive elements and core content in the video content are identified and separated through key point extraction technology to obtain a content key element set; The content key element set is quantitatively scored according to the social influence intensity standard, and the social sensitivity and guiding role of the content elements are evaluated through social public opinion reference to obtain a social influence intensity index; A transmission link model is constructed according to the content key element set, and the diffusion speed and coverage range of the content in different populations are calculated through user behavior analysis to obtain a transmission penetration depth value; Psychological acceptance measurement is performed based on the content key element set, and the psychological acceptance degree of different age groups and cultural background populations is analyzed through audience portrait matching to obtain a psychological acceptance level; The social influence intensity index, the propagation penetration depth value and the psychological acceptance level are multi-dimensionally cross-correlated, and the three indexes are dynamically weighted and calculated through a weight distribution matrix to obtain comprehensive score data; The comprehensive score data is graded and mapped, the content score is segmented through threshold adjustment, a variable weight scoring system is established, and a dynamic weight content score table is obtained.

5. The data mining based movie censorship analysis method of claim 1, wherein, According to the stereoscopic content influence graph, the priority of the supervision points is sorted through decision rules to obtain an intelligent supervision scheme containing precise modification suggestions, gradient restriction measures and individualized classification requirements, including: High-risk supervision points are extracted from the stereoscopic content influence graph, and the high-risk supervision points are identified and screened through risk weight calculation to obtain a key supervision point list; The key supervision point list is hierarchically prioritized, and the processing order and resource allocation ratio of different supervision points are determined through emergency level evaluation to obtain a supervision priority sequence list; Based on the supervision priority sequence list, for high-priority content nodes, the difference between the illegal content and the compliance standard is accurately positioned through content comparison analysis to obtain a precise modification suggestion list; According to the content propagation characteristics in the key supervision point list, a hierarchical restriction scheme is developed for content with different influence ranges through gradient control strategy to obtain a gradient restriction measure scheme; According to the content characteristics and audience group information of the key supervision point list, the content adaptation population range is accurately defined through hierarchical standard mapping to obtain an individualized classification requirement table; The precise modification suggestion list, the gradient restriction measure scheme and the individualized classification requirement table are integrated and associated to obtain an intelligent supervision scheme.

6. The data mining based movie censorship analysis method of claim 1, wherein, According to the intelligent supervision scheme, the execution effect of the supervision measures is deviated and corrected through effect feedback analysis to obtain a self-optimized supervision parameter set, including: The intelligent supervision scheme is implemented and deployed, the time nodes and effect data in the measure landing process are collected and recorded through execution tracking to obtain supervision execution data flow; Key effect indicators are extracted from the supervision execution data flow, and the difference between the expected supervision effect and the actual execution result is quantitatively compared through target deviation measurement to obtain supervision effect deviation data; The root cause analysis is performed on the supervision effect deviation data, the key factors and influence conditions causing the deviation are identified and located through traceability tracking to obtain a deviation factor attribution table; Based on the deviation factor attribution table, a correction strategy is developed, the correction direction and adjustment amplitude of different deviation factors are determined through correction strength calculation to obtain a supervision parameter adjustment scheme; The supervision parameter adjustment scheme is optimized and verified, the effectiveness of the adjusted supervision parameters is tested and evaluated through small-scale experiments to obtain a parameter optimization verification report; The effective parameters in the parameter optimization verification report are extracted and analyzed to obtain a self-optimized supervision parameter set.

7. A data mining based film and television supervision analysis system for implementing the data mining based film and television supervision analysis method according to any one of claims 1 to 6, characterized in that, The movie supervision analysis system based on data mining comprises: The collection module is used for panoramic data collection of the video platform movie and television works, and the video, audio and text information are standardized and arranged through cross-media transcoding processing to obtain a multi-dimensional movie and television content dataset; The extraction module is used for scene semantic extraction of video frames, multi-level emotion recognition of audio and cultural value tendency analysis of text according to the multi-dimensional movie and television content dataset through content feature recognition to obtain a time sequence associated content marking list; The scoring module is used for cross scoring of key elements according to social influence intensity, transmission penetration depth and audience psychological acceptance through hierarchical scoring based on the time sequence associated content marking list to obtain a dynamic weight content scoring table; The analysis module is used for deviation correction analysis of the execution effect of the supervision measures through effect feedback according to the intelligent supervision scheme to obtain a self-optimizing supervision parameter set. The computer readable storage medium has a computer program stored thereon, and the computer program enables the processor to execute the data mining based movie supervision analysis method in any one of claims 1 to 6 when the computer program is run on the processor.

9. A computer readable storage medium having a computer program stored thereon, the computer program enabling the processor to execute the data mining based movie supervision analysis method in any one of claims 1 to 6 when the computer program is run on the processor.

8. A computer device, comprising: ​ ​

Images