A deep learning effect correlation mining method based on children's animation viewing behavior data

Abstract

The present application relates to the technical field of educational big data analysis and personalized education, and discloses a deep learning effect correlation mining method based on children's animation watching behavior data, which specifically comprises the following method steps: step one, constructing a three-dimensional data model, the three dimensions including behavior dimension, content dimension and effect dimension; step two, through a group difference analysis model, an effective animation watching behavior path is identified for children groups of different ages. The present application can comprehensively and accurately evaluate the learning process and effectiveness of children by constructing a three-dimensional data model (behavior dimension, content dimension and effect dimension). Experimental data shows that after using the method, the vocabulary standard rate of the small child group (0-6 years old) is improved to 68%, and the long-time immersion mode conversion rate of the big child group (7-12 years old) is improved by 75%. This data-driven optimization mechanism ensures that the learning path is highly matched with the cognitive development law of children, thereby significantly improving the English enlightenment effect.

Description

Technical Field

[0001] This invention relates to the field of educational big data analysis and personalized education technology, specifically a method for mining the correlation of deep learning effects based on children's animation viewing behavior data. Background Technology

[0002] The following technical deficiencies exist in the current field of early childhood English education: 1. Insufficient data utilization: Existing technologies only use animation playback data for basic recommendations (such as playback statistics and simple interest tag generation), without establishing a correlation model between playback behavior sequences (such as viewing order and repetition patterns) and language ability development. This results in key learning paths not being identified (such as the difference in the effect of "high-frequency short time" and "low-frequency long time" modes on children with zero foundation). Parents only receive playback records and lack scientific behavioral guidance suggestions.

[0003] 2. The guidance strategy is crude: the general education suggestion template lacks age- and level-based design (the difference in needs between young children and children with no foundation is not reflected), and it is not related to the specific characteristics of the animation content (such as vocabulary density and sentence complexity), and lacks a dynamic adjustment mechanism (it cannot optimize suggestions based on the learning effect at each stage).

[0004] 3. Mixed user groups: Traditional behavioral analysis does not screen high-value users and mixes active and inactive user data, which causes key success paths to be masked by noise. It lacks a benchmark model of successful user behavior and cannot extract reusable high-quality paths.

[0005] Therefore, this invention provides a method for mining the correlation of deep learning effects based on children's animation viewing behavior data. Summary of the Invention

[0006] The purpose of this invention is to provide a method for mining the correlation of deep learning effects based on children's animation viewing behavior data, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for mining the correlation of deep learning effects based on children's animation viewing behavior data, specifically including the following method steps: Step 1: Construct a three-dimensional data model, which includes behavioral dimension, content dimension, and effect dimension; Step 2: Using a group difference analysis model, identify effective animation viewing behavior paths for children of different age groups; Step 3: Based on the identified effective behavioral paths, generate age-appropriate and phase-specific actionable parent guidance strategies.

[0008] Preferably, the behavioral dimension focuses on specific behavioral patterns of children watching animation, and is constructed by collecting and analyzing the following key indicators: A1.1 Single viewing duration: Record the duration of each time a child watches an animation to assess their attention span and learning interest; A1.2 Daily Frequency: This counts the number of times children watch animation each day, reflecting their learning frequency and habits; A1.3 Continuous Learning Days: Track the number of consecutive days a child watches animation to reflect their learning persistence and perseverance; A1.4. Active repetitive behavior: Monitor whether children actively and repeatedly watch the same or related animations to determine their interest in and understanding of the content.

[0009] Preferably, the content dimension focuses on the characteristics of the animation content itself, revealing its impact on children's learning outcomes through in-depth analysis of the following: A2.1 Vocabulary Growth Rate: Calculate the number of new words added in each episode of the animation to evaluate the animation's contribution to vocabulary expansion; A2.2 Sentence Complexity: Analyze the proportion of clauses used in the animation to measure their effect on improving children's grammar skills; A2.3 Thematic Continuity: Examine the thematic relevance between the series of animations to ensure the continuity and systematic nature of the learning content.

[0010] Preferably, the effectiveness dimension is directly related to children's learning outcomes and is quantified and evaluated using the following indicators: A3.1 Regular assessment scores: Regularly assess children's language abilities and adjust learning paths and content recommendations based on the assessment results.

[0011] A3.2 Language Output Frequency Feedback from Parents: The frequency of children's language output feedback from parents is collected through the data collection interface, such as the use of new vocabulary and sentence expression in daily conversations, in order to objectively reflect the learning effect.

[0012] Preferably, the group difference analysis model includes a group of young children with no prior knowledge (0-6 years old) and a group of older children with no prior knowledge (7-12 years old). Specifically targeting young children (0-6 years old) with no prior knowledge of cognitive abilities, who have shorter attention spans and are in a rapid cognitive development phase, an effective model focusing on identifying "short-duration, high-frequency + strong thematic association" is adopted: Short and frequent: It is recommended to limit each animation viewing time to about 8 minutes and encourage 3 short learning sessions per day. This arrangement is in line with the characteristic that young children's attention is easily distracted. Through high-frequency and short-duration learning stimulation, memory association can be effectively strengthened and learning results can be improved. Strong thematic relevance: It is recommended to choose animated series with strong content relevance, such as series revolving around themes like "animal world" or "daily necessities"; by continuously exposing children to the same theme, they can build a complete knowledge system in their minds and deepen their understanding of vocabulary and sentence structure; For older children (7-12 years old) with no prior knowledge of vocabulary, who already possess a certain level of attention span and cognitive ability, an optimized learning path is adopted that focuses on "long-term immersion + cross-thematic vocabulary repetition": Extended Immersion: It is recommended to extend the time for watching animation at one time to about 25 minutes, so that older children can immerse themselves in the animation situation and deeply understand complex sentence structures and abstract concepts; this extended immersion learning method helps to improve older children's language comprehension and thinking depth. Cross-theme vocabulary repetition: It is recommended to choose animated content that includes two or more related themes, such as a series that combines "natural science" with "history and culture". By naturally repetitive vocabulary across different themes, older children can expand their vocabulary while improving their practical application of vocabulary, forming a more flexible and richer language network.

[0013] Preferably, the parent guidance strategy generation includes the following two mechanisms to achieve precise guidance: 1. Output a visual learning path diagram This roadmap uses a timeline as its basis and incorporates the principles of children's cognitive development, breaking down long-term learning goals into quantifiable, phased tasks, specifically including: 1.1 Stage Goal Design: Adopting a "theme-based progression" model, for example: Week 1 focuses on "basic vocabulary input on animal themes," using the "Animal Kingdom" animated series to learn 20 core vocabulary words; Week 3 was upgraded to "thematic question and answer animation", introducing simple question patterns such as "Where is the cat?" to strengthen language output ability; 1.2 Quantifying Success Criteria: Set verifiable achievement indicators for each stage, such as: Vocabulary achievement rate (≥80% of new words can be used correctly in context); Sentence structure imitation accuracy (≥70% of sentences are structurally complete and grammatically correct); 1.3 Parental Guidance Strategies: Provide scenario-specific operation guidelines, such as: Before watching: Guide children to focus on the core content using "vocabulary preview cards"; During viewing: Parents are advised to use the "Interactive Questioning Manual" for real-time language stimulation; After watching: We recommend the "Theme Extension Game Pack" to reinforce learning outcomes; 2. Establish a dynamic early warning mechanism This mechanism achieves intelligent identification and intervention of abnormal states by monitoring learning behavior data in real time and building individualized learning models using machine learning algorithms. 2.1 Multi-dimensional monitoring system: Behavioral metrics: duration of a single viewing session, time period distribution, and frequency of interaction; Content metrics: vocabulary coverage, sentence complexity matching degree; Performance indicators: accuracy rate of real-time assessments and frequency of parental feedback.

[0014] 2.2 Intelligent Early Warning Trigger Rules: When the system detects that "single viewing time > 30 minutes but vocabulary absorption rate < 10%", it determines the state as "passive viewing" and automatically triggers adjustment suggestions. If the theme jump rate is greater than 50% for three consecutive days (such as a sudden switch from an animal theme to a math theme), an "attention distraction" warning will be activated. Optimize suggestion push logic: Regarding the suggestion for "passive viewing": insert a 3-minute "active recall" segment, requiring children to retell key plot points of the animation; Regarding the suggestion for "distraction": provide a "theme transition animation pack" to achieve smooth transitions through character linking.

[0015] Preferably, it also includes a high-value user screening mechanism, which establishes scientific user segmentation standards to accurately identify key factors influencing learning outcomes. This mechanism comprises the following three core modules: 1. Define the criteria for "long-term loyal users" This standard quantifies user stickiness through two dimensions to ensure that the samples included in the analysis possess continuous learning behavior characteristics: 1.1 Time Dimension Requirement: Users must meet the hard requirement of continuous system use for ≥90 days. This period is set based on research in child behavioral psychology, which shows that regular learning for 90 consecutive days can form stable learning habits (refer to "Research on the Habit Formation Cycle of Children"). 1.2 Activity Requirements: During the continuous use period, users must maintain an active number of ≥5 days per week. This setting aims to exclude users who engage in "cramming" learning and ensure that the sample reflects real learning patterns rather than short-term behavioral fluctuations. 2. Define the "higher-level" standard This standard objectively quantifies users' language proficiency development level through a tiered competency certification system, specifically including: 2.1 Advanced Test Level Design: Set at least 3 progressive evaluation nodes, each level corresponding to a specific ability development milestone, for example: Level 1: Basic vocabulary recognition (≥200 words); Level 2: Simple sentence structure (able to construct a "subject + verb + object" structure); Level 3: Complex context comprehension (vocabulary ≥ 500 words, able to understand subjunctive mood); 2.2 Dynamic evaluation mechanism: Adaptive assessment technology is adopted to adjust the difficulty of questions based on the user's real-time performance, ensuring the scientific nature of the evaluation results; 3. Successfully build user-specific datasets After completing user segmentation, the system builds a high-quality analysis sample library through the following steps: 3.1 Data Cleaning Rules: Exclude "zero-evaluation users": samples that did not participate in any performance evaluation. Filter out “abnormal data segments”: such as irrational data with a single-day viewing time exceeding 180 minutes; Exclude "test users": Suspected test accounts that were used more than 20 times per day within 7 days of registration; 3.2 Data Augmentation Processing: Complete the behavioral trajectories of valid samples and predict missing data segments using machine learning models; Establish a user profile tagging system to label derivative characteristics such as learning style and content preference.

[0016] Preferably, it also includes the abstraction and transfer of successful learning paths, which enables the reuse of learning experiences through a data-driven approach, significantly improving the learning start and effectiveness for new users. This step includes the following two key components: 1. Extract common behavioral patterns of successful users This step uses cluster analysis and pattern recognition techniques to extract universally applicable learning path templates from successful user datasets, specifically including: 1.1 Pattern Feature Engineering: For young children (0-6 years old), a typical pattern was identified: "3 times a day for 10 minutes of the same theme animation for the first 30 days + 1 nursery rhyme reinforcement every 7 days". This pattern combines children's short-term memory reinforcement mechanism, establishes basic cognition through high-frequency input of the same theme (such as watching the "Marine Life" series for 3 consecutive days), and then uses nursery rhyme reinforcement (such as the theme song of "Finding Nemo") to achieve multimodal memory encoding. 1.2 For older children (7-12 years old), a "dual-theme alternating reinforcement cycle" model is abstracted, which is a cyclical structure of "5 days of specialized training (such as science theme) + 2 days of comprehensive application (such as science experiment animation)". This design is in line with the cognitive development characteristics of older children. Through the alternation of specialized breakthroughs and comprehensive applications, the deep internalization of knowledge is promoted. Parametric representation: Transform behavioral patterns into a computable set of parameters, such as "single duration = 10 minutes", "topic switching cycle = 7 days", "reinforcement frequency = 1 time / week", etc., to provide a quantitative basis for subsequent path matching; 2. By calculating path similarity, match new users with the initial path of the most similar successful users. This step employs the Dynamic Time Warping (DTW) algorithm to achieve precise matching of the paths of new users and successful users, specifically including: 2.1 Feature Vector Construction: Behavioral data generated during the initial registration period of new users (such as viewing records in the first 3 days) are transformed into multi-dimensional feature vectors, including indicators such as "single duration distribution", "topic preference coefficient", and "interaction response speed". 2.2 Similarity Calculation: The DTW algorithm is used to calculate the similarity matrix between the feature vector of new users and the path template of successful users. The algorithm optimizes path alignment through dynamic programming, which effectively solves the matching bias caused by differences in user behavior rhythm. For example, even if the viewing time of new users on the second day is 20% shorter than that of the template, the algorithm can still achieve accurate matching through time axis compression. 2.3 Recommended Initial Path: Based on similarity ranking, the system recommends the top 3 candidate paths to new users and sets an "adaptive observation period". During the observation period, the system continuously monitors the matching degree between the user's actual behavior and the recommended path. If the deviation exceeds the threshold (e.g., behavioral similarity <70% for 3 consecutive days), the path rematch mechanism is automatically triggered.

[0017] Preferably, the method is implemented through the following system architecture: The data layer includes a behavior log library and a knowledge graph of successful user paths; The analysis layer includes a path mining engine and a real-time similarity calculation module; The recommendation layer includes a dynamic path generator, anomaly detection, and calibrator. The interaction layer includes a visual report for parents and a one-click optimization button.

[0018] Preferably, the method is implemented through the following specific embodiments: Data pre-screening and cleaning, defining the criteria for "successful users" and constructing a dataset of successful users; Behavioral path modeling and analysis: the FP-Growth algorithm is used to mine high-frequency behavioral sequences for young children, and cluster analysis is used to identify effective patterns for older children; Dynamic recommendations and parental guidance are generated by matching new user characteristics with a historical successful user prototype library and calculating behavioral sequence similarity. The system provides initial recommendations and deviation warnings through examples of parent-side commands.

[0019] This invention provides a deep learning-based method for mining correlations of children's animation viewing behavior data. It offers the following advantages: (1) By constructing a three-dimensional data model (behavioral dimension, content dimension, and effect dimension), this invention can comprehensively and accurately evaluate children's learning process and effectiveness. Experimental data shows that after adopting this method, the vocabulary achievement rate of young children (0-6 years old) increased to 68%, and the conversion rate of long-term immersion mode of older children (7-12 years old) increased by 75%. This data-driven optimization mechanism ensures that the learning path is highly matched with the cognitive development law of children, thereby significantly improving the effect of English enlightenment.

[0020] (2) This invention provides parents with scientific and operable guidance solutions by outputting a visual learning path map and establishing a dynamic early warning mechanism. For example, the “short-term high-frequency + strong theme association” mode recommended for young children and the “long-term immersion + cross-theme vocabulary repetition” mode recommended for older children have been empirically verified to be effective. Parent end instruction examples (such as “play Peppa Pig S1 EP1 at 9:00 / 15:00 / 19:00 every day”) make the guidance strategy more concrete and significantly improve parents’ willingness to adopt it (up to 90%), thus realizing the precision and personalization of parent guidance strategies.

[0021] (3) This invention extracts universal learning path templates (such as the "3 times a day for 10 minutes of the same theme animation for the first 30 days" pattern) from the successful user dataset through a high-value user screening mechanism and success path abstraction technology. After adopting the Dynamic Time Warping (DTW) algorithm, the accuracy of new user path matching is increased to 85%. This "experience reuse" mechanism enables new users to quickly enter an efficient learning state. The vocabulary achievement rate in 3 months is 68% higher than that of traditional recommendation systems, significantly shortening the learning adaptation cycle, constructing a transferable success path knowledge base, and optimizing the learning starting point for new users. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the overall process of the present invention; Figure 2 This is a view of the data acquisition and preprocessing process of the present invention; Figure 3 This is a view of the population difference analysis of the present invention; Figure 4 This is an abstract view of the successful path of this invention; Figure 5 This is the new user path matching view of the present invention; Figure 6 Generate a view for the parent guidance strategy of this invention; Figure 7This is a view for evaluating and providing feedback on the effectiveness of the invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the invention, and should not be construed as limiting the invention.

[0025] A preferred embodiment of the deep learning effect correlation mining method based on children's animation viewing behavior data provided by the present invention is as follows: Figure 1-7 The following is an example of a deep learning-based correlation mining method based on children's animation viewing behavior data, which includes the following steps: Step 1: Construct a three-dimensional data model, comprising behavioral, content, and effect dimensions. The behavioral dimension focuses on children's specific behavioral patterns while watching animation, and is constructed by collecting and analyzing the following key indicators: A1.1 Single viewing duration: Record the duration of each time a child watches an animation to assess their attention span and learning interest; A1.2 Daily Frequency: This counts the number of times children watch animation each day, reflecting their learning frequency and habits; A1.3 Continuous Learning Days: Track the number of consecutive days a child watches animation to reflect their learning persistence and perseverance; A1.4. Active repetitive behavior: Monitor whether children actively and repeatedly watch the same or related animations to determine their interest in and understanding of the content.

[0026] The content dimension focuses on the characteristics of the animation content itself, and through in-depth analysis of the following content, reveals its impact on children's learning outcomes: A2.1 Vocabulary Growth Rate: Calculate the number of new words added in each episode of the animation to evaluate the animation's contribution to vocabulary expansion; A2.2 Sentence Complexity: Analyze the proportion of clauses used in the animation to measure their effect on improving children's grammar skills; A2.3 Thematic Continuity: Examine the thematic relevance between the series of animations to ensure the continuity and systematic nature of the learning content.

[0027] The aforementioned effectiveness dimension is directly related to children's learning outcomes and is quantified and evaluated using the following indicators: A3.1 Regular assessment scores: Regularly assess children's language abilities and adjust learning paths and content recommendations based on the assessment results.

[0028] A3.2, Frequency of Language Output Feedback from Parents: The frequency of children's language output feedback from parents is collected through the data collection interface, such as the use of new vocabulary and sentence expression in daily conversations, in order to objectively reflect the learning effect; Step 2: Identify effective animation viewing behavior paths for children of different age groups using a group difference analysis model; the group difference analysis model includes a group of toddlers with no prior knowledge of animation (0-6 years old) and a group of older children with no prior knowledge of animation (7-12 years old). Specifically targeting young children (0-6 years old) with no prior knowledge of cognitive abilities, who have shorter attention spans and are in a rapid cognitive development phase, an effective model focusing on identifying "short-duration, high-frequency + strong thematic association" is adopted: Short and frequent: It is recommended to limit each animation viewing time to about 8 minutes and encourage 3 short learning sessions per day. This arrangement is in line with the characteristic that young children's attention is easily distracted. Through high-frequency and short-duration learning stimulation, memory association can be effectively strengthened and learning results can be improved. Strong thematic relevance: It is recommended to choose animated series with strong content relevance, such as series revolving around themes like "animal world" or "daily necessities"; by continuously exposing children to the same theme, they can build a complete knowledge system in their minds and deepen their understanding of vocabulary and sentence structure; For older children (7-12 years old) with no prior knowledge of vocabulary, who already possess a certain level of attention span and cognitive ability, an optimized learning path is adopted that focuses on "long-term immersion + cross-thematic vocabulary repetition": Extended Immersion: It is recommended to extend the time for watching animation at one time to about 25 minutes, so that older children can immerse themselves in the animation situation and deeply understand complex sentence structures and abstract concepts; this extended immersion learning method helps to improve older children's language comprehension and thinking depth. Cross-theme vocabulary repetition: It is recommended to choose animation content that includes two or more related themes, such as a series that combines "natural science" with "history and culture"; by naturally repetitive vocabulary across different themes, older children can expand their vocabulary while improving their practical application of vocabulary, forming a more flexible and richer language network; Step 3: Based on the identified effective behavioral paths, generate age-appropriate and phase-specific actionable parent guidance strategies.

[0029] The parent guidance strategy is generated through the following two mechanisms to achieve precise guidance: 1. Output a visual learning path diagram This roadmap uses a timeline as its basis and incorporates the principles of children's cognitive development, breaking down long-term learning goals into quantifiable, phased tasks, specifically including: 1.1 Stage Goal Design: Adopting a "theme-based progression" model, for example: Week 1 focuses on "basic vocabulary input on animal themes," using the "Animal Kingdom" animated series to learn 20 core vocabulary words; Week 3 was upgraded to "thematic question and answer animation", introducing simple question patterns such as "Where is the cat?" to strengthen language output ability; 1.2 Quantifying Success Criteria: Set verifiable achievement indicators for each stage, such as: Vocabulary achievement rate (≥80% of new words can be used correctly in context); Sentence structure imitation accuracy (≥70% of sentences are structurally complete and grammatically correct); 1.3 Parental Guidance Strategies: Provide scenario-specific operation guidelines, such as: Before watching: Guide children to focus on the core content using "vocabulary preview cards"; During viewing: Parents are advised to use the "Interactive Questioning Manual" for real-time language stimulation; After watching: We recommend the "Theme Extension Game Pack" to reinforce learning outcomes; 2. Establish a dynamic early warning mechanism This mechanism achieves intelligent identification and intervention of abnormal states by monitoring learning behavior data in real time and building individualized learning models using machine learning algorithms. 2.1 Multi-dimensional monitoring system: Behavioral metrics: duration of a single viewing session, time period distribution, and frequency of interaction; Content metrics: vocabulary coverage, sentence complexity matching degree; Performance indicators: accuracy rate of real-time assessments and frequency of parental feedback.

[0030] 2.2 Intelligent Early Warning Trigger Rules: When the system detects that "single viewing time > 30 minutes but vocabulary absorption rate < 10%", it determines the state as "passive viewing" and automatically triggers adjustment suggestions. If the theme jump rate is greater than 50% for three consecutive days (such as a sudden switch from an animal theme to a math theme), an "attention distraction" warning will be activated. Optimize suggestion push logic: Regarding the suggestion for "passive viewing": insert a 3-minute "active recall" segment, requiring children to retell key plot points of the animation; Regarding the suggestion for "distraction": provide a "theme transition animation pack" to achieve smooth transitions through character linking.

[0031] Furthermore, this invention provides a method for mining the correlation of deep learning effects based on children's animation viewing behavior data, and also includes a high-value user screening mechanism. This mechanism establishes scientific user segmentation standards to accurately identify key influencing factors of learning effectiveness. This mechanism comprises the following three core modules: 1. Define the criteria for "long-term loyal users" This standard quantifies user stickiness through two dimensions to ensure that the samples included in the analysis possess continuous learning behavior characteristics: 1.1 Time Dimension Requirement: Users must meet the hard requirement of continuous system use for ≥90 days. This period is set based on research in child behavioral psychology, which shows that regular learning for 90 consecutive days can form stable learning habits (refer to "Research on the Habit Formation Cycle of Children"). 1.2 Activity Requirements: During the continuous use period, users must maintain an active number of ≥5 days per week. This setting aims to exclude users who engage in "cramming" learning and ensure that the sample reflects real learning patterns rather than short-term behavioral fluctuations. 2. Define the "higher-level" standard This standard objectively quantifies users' language proficiency development level through a tiered competency certification system, specifically including: 2.1 Advanced Test Level Design: Set at least 3 progressive evaluation nodes, each level corresponding to a specific ability development milestone, for example: Level 1: Basic vocabulary recognition (≥200 words); Level 2: Simple sentence structure (able to construct a "subject + verb + object" structure); Level 3: Complex context comprehension (vocabulary ≥ 500 words, able to understand subjunctive mood); 2.2 Dynamic evaluation mechanism: Adaptive assessment technology is adopted to adjust the difficulty of questions based on the user's real-time performance, ensuring the scientific nature of the evaluation results; 3. Successfully build user-specific datasets After completing user segmentation, the system builds a high-quality analysis sample library through the following steps: 3.1 Data Cleaning Rules: Exclude "zero-evaluation users": samples that did not participate in any performance evaluation. Filter out “abnormal data segments”: such as irrational data with a single-day viewing time exceeding 180 minutes; Exclude "test users": Suspected test accounts that were used more than 20 times per day within 7 days of registration; 3.2 Data Augmentation Processing: Complete the behavioral trajectories of valid samples and predict missing data segments using machine learning models; Establish a user profile tagging system to label derivative characteristics such as learning style and content preference.

[0032] Furthermore, this invention provides a method for mining the correlation of deep learning effects based on children's animation viewing behavior data, which also includes success path abstraction and transfer. This method achieves the reuse of learning experiences through a data-driven approach, significantly improving the learning starting point and effectiveness for new users. This step includes the following two key components: 1. Extract common behavioral patterns of successful users This step uses cluster analysis and pattern recognition techniques to extract universally applicable learning path templates from successful user datasets, specifically including: 1.1 Pattern Feature Engineering: For young children (0-6 years old), a typical pattern was identified: "3 times a day for 10 minutes of the same theme animation for the first 30 days + 1 nursery rhyme reinforcement every 7 days". This pattern combines children's short-term memory reinforcement mechanism, establishes basic cognition through high-frequency input of the same theme (such as watching the "Marine Life" series for 3 consecutive days), and then uses nursery rhyme reinforcement (such as the theme song of "Finding Nemo") to achieve multimodal memory encoding. 1.2 For older children (7-12 years old), a "dual-theme alternating reinforcement cycle" model is abstracted, which is a cyclical structure of "5 days of specialized training (such as science theme) + 2 days of comprehensive application (such as science experiment animation)". This design is in line with the cognitive development characteristics of older children. Through the alternation of specialized breakthroughs and comprehensive applications, the deep internalization of knowledge is promoted. Parametric representation: Transform behavioral patterns into a computable set of parameters, such as "single duration = 10 minutes", "topic switching cycle = 7 days", "reinforcement frequency = 1 time / week", etc., to provide a quantitative basis for subsequent path matching; 2. By calculating path similarity, match new users with the initial path of the most similar successful users. This step employs the Dynamic Time Warping (DTW) algorithm to achieve precise matching of the paths of new users and successful users, specifically including: 2.1 Feature Vector Construction: Behavioral data generated during the initial registration period of new users (such as viewing records in the first 3 days) are transformed into multi-dimensional feature vectors, including indicators such as "single duration distribution", "topic preference coefficient", and "interaction response speed". 2.2 Similarity Calculation: The DTW algorithm is used to calculate the similarity matrix between the feature vector of new users and the path template of successful users. The algorithm optimizes path alignment through dynamic programming, which effectively solves the matching bias caused by differences in user behavior rhythm. For example, even if the viewing time of new users on the second day is 20% shorter than that of the template, the algorithm can still achieve accurate matching through time axis compression. 2.3 Recommended Initial Path: Based on similarity ranking, the system recommends the top 3 candidate paths to new users and sets an "adaptive observation period". During the observation period, the system continuously monitors the matching degree between the user's actual behavior and the recommended path. If the deviation exceeds the threshold (e.g., behavioral similarity <70% for 3 consecutive days), the path rematch mechanism is automatically triggered.

[0033] Furthermore, this invention provides a method for mining the correlation of deep learning effects based on children's animation viewing behavior data, which is implemented through the following system architecture: The data layer includes a behavior log library and a knowledge graph of successful user paths; The analysis layer includes a path mining engine and a real-time similarity calculation module; The recommendation layer includes a dynamic path generator, anomaly detection, and calibrator. The interaction layer includes a visual report for parents and a one-click optimization button.

[0034] Furthermore, this invention provides a method for mining the correlation of deep learning effects based on children's animation viewing behavior data, which is implemented through the following specific embodiments: Data pre-screening and cleaning, defining the criteria for "successful users" and constructing a dataset of successful users; Define the criteria for a "successful user": Long-term adherence: Use continuously for ≥90 days, with an average daily usage time of ≥15 minutes; Advanced level: Pass the Level 3 (vocabulary ≥ 300 words) or higher advanced test; Valid data: At least 3 phased assessments have been completed (e.g., TPR test once a month). Build a successful user dataset: 12,000 eligible users were selected from the young children group and 8,000 from the older children group; Remove invalid data (such as records of incomplete evaluations or playback interruptions caused by device malfunctions); Behavioral path modeling and analysis: the FP-Growth algorithm is used to mine high-frequency behavioral sequences for young children, and cluster analysis is used to identify effective patterns for older children; Children (0-6 years old): The FP-Growth algorithm was used to mine high-frequency behavior sequences and discover common paths: Path A (62%): [Repeated playback of the same series of animations] × 7 days → [Interspersed with accompanying nursery rhymes] × 3 days → [New animation in the same series] × 7 days Path B (24%): Alternating between animation and interactive games (game reinforcement every 3 days) Key finding: After 3 months, users on Path A showed a 120% improvement in sentence output ability, significantly outperforming Path B (75% improvement).

[0035] Older children (7-12 years old): Two types of effective patterns were identified through cluster analysis: Mode 1: 25-minute immersive viewing session (including 2 related topics) + weekly topic assessment Mode 2: 15 minutes of animation + 5 minutes of repetition training daily Data findings: The growth rate of user vocabulary in Mode 1 was 1.8 times that of Mode 2; Dynamic recommendations and parental guidance generation: The system matches new user characteristics with a historical database of successful user prototypes and calculates behavioral sequence similarity. Initial recommendations and deviation warnings are provided through parent-side instruction examples. New user matching logic: Match historical successful user prototypes based on age and gender; Calculate the DTW similarity between the current action sequence and the target path (formula): Similarity Score = 1 / (1 + DTW_Distance) If the similarity is less than 0.7, a calibration suggestion will be pushed (such as "It was detected that you did not repeat the playback as suggested, which may affect vocabulary consolidation").

[0036] Example of parental instructions: Initial phase (within 7 days of registration): Based on data from 1,026 successful users aged 3 and above, we recommend that you: Peppa Pig S1 EP1 (8 minutes) will be played at 9:00, 15:00 and 19:00 every day. Starting from day 4, add the nursery rhyme "Animal Sounds" once daily. Deviation warning (real-time): "The current single viewing time (18 minutes) exceeds the average time for successful users aged 3 (12 minutes)." It is suggested to split the viewing into two sessions: [9:00-9:10] + [15:00-15:08] (click to adjust).

[0037] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0038] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for mining the correlation of deep learning effects based on children's animation viewing behavior data, characterized in that, Specifically, the methods and steps include the following: Step 1: Construct a three-dimensional data model, which includes behavioral dimension, content dimension, and effect dimension; Step 2: Using a group difference analysis model, identify effective animation viewing behavior paths for children of different age groups; Step 3: Based on the identified effective behavioral paths, generate age-appropriate and phase-specific actionable parent guidance strategies.

2. The method for deep learning effect correlation mining based on children's animation viewing behavior data according to claim 1, characterized in that: The behavioral dimension focuses on specific behavioral patterns of children watching animation and is constructed by collecting and analyzing the following key indicators: A1.1 Single viewing duration: Record the duration of each time a child watches an animation to assess their attention span and learning interest; A1.2 Daily Frequency: This counts the number of times children watch animation each day, reflecting their learning frequency and habits; A1.3 Continuous Learning Days: Track the number of consecutive days a child watches animation to reflect their learning persistence and perseverance; A1.

4. Active repetitive behavior: Monitor whether children actively and repeatedly watch the same or related animations to determine their interest in and understanding of the content.

3. The method for deep learning effect correlation mining based on children's animation viewing behavior data according to claim 1, characterized in that: The content dimension focuses on the characteristics of the animation content itself, and through in-depth analysis of the following content, reveals its impact on children's learning outcomes: A2.1 Vocabulary Growth Rate: Calculate the number of new words added in each episode of the animation to evaluate the animation's contribution to vocabulary expansion; A2.2 Sentence Complexity: Analyze the proportion of clauses used in the animation to measure their effect on improving children's grammar skills; A2.3 Thematic Continuity: Examine the thematic relevance between the series of animations to ensure the continuity and systematic nature of the learning content.

4. The method for deep learning effect correlation mining based on children's animation viewing behavior data according to claim 1, characterized in that: The aforementioned effectiveness dimension is directly related to children's learning outcomes and is quantified and evaluated using the following indicators: A3.1 Regular assessment scores: Regularly assess children's language abilities and adjust learning paths and content recommendations based on the assessment results. A3.2 Language Output Frequency Feedback from Parents: The frequency of children's language output feedback from parents is collected through the data collection interface, such as the use of new vocabulary and sentence expression in daily conversations, in order to objectively reflect the learning effect.

5. The method for deep learning effect correlation mining based on children's animation viewing behavior data according to claim 1, characterized in that: The group difference analysis model includes a group of young children with no prior knowledge (0-6 years old) and a group of older children with no prior knowledge (7-12 years old). Specifically targeting young children (0-6 years old) with no prior knowledge of cognitive abilities, who have shorter attention spans and are in a rapid cognitive development phase, an effective model focusing on identifying "short-duration, high-frequency + strong thematic association" is adopted: Short and frequent: It is recommended to limit each animation viewing time to about 8 minutes and encourage 3 short learning sessions per day. This arrangement is in line with the characteristic that young children's attention is easily distracted. Through high-frequency and short-duration learning stimulation, memory association can be effectively strengthened and learning results can be improved. Strong thematic relevance: It is recommended to choose animated series with strong content relevance, such as series revolving around themes like "animal world" or "daily necessities"; by continuously being exposed to the same theme, children can build a complete knowledge system in their minds and deepen their understanding of vocabulary and sentence structure; For older children (7-12 years old) with no prior knowledge of vocabulary, who already possess a certain level of attention span and cognitive ability, an optimized learning path is adopted that focuses on "long-term immersion + cross-thematic vocabulary repetition": Extended Immersion: It is recommended to extend the time for watching animation at one time to about 25 minutes, so that older children can immerse themselves in the animation situation and deeply understand complex sentence structures and abstract concepts; this extended immersion learning method helps to improve older children's language comprehension and thinking depth. Cross-theme vocabulary repetition: It is recommended to choose animated content that includes two or more related themes, such as a series that combines "natural science" with "history and culture". By naturally repetitive vocabulary across different themes, older children can expand their vocabulary while improving their practical application of vocabulary, forming a more flexible and richer language network.

6. The method for deep learning effect correlation mining based on children's animation viewing behavior data according to claim 1, characterized in that: The parent guidance strategy is generated through the following two mechanisms to achieve precise guidance:

1. Output a visual learning path diagram This roadmap uses a timeline as its basis and incorporates the principles of children's cognitive development, breaking down long-term learning goals into quantifiable, phased tasks, specifically including: 1.1 Stage Goal Design: Adopting a "theme-based progression" model, for example: Week 1 focuses on "basic vocabulary input on the theme of animals," using the "Animal Kingdom" animated series to learn 20 core vocabulary words; Week 3 was upgraded to "thematic question and answer animation", introducing simple question patterns such as "Where is the cat?" to strengthen language output ability; 1.2 Quantifying Success Criteria: Set verifiable achievement indicators for each stage, such as: Vocabulary achievement rate (≥80% of new words can be used correctly in context); Sentence structure imitation accuracy (≥70% of sentences are structurally complete and grammatically correct); 1.3 Parental Guidance Strategies: Provide scenario-specific operation guidelines, such as: Before watching: Guide children to focus on the core content using "vocabulary preview cards"; During viewing: Parents are advised to use the "Interactive Questioning Manual" for real-time language stimulation; After watching: We recommend the "Theme Extension Game Pack" to reinforce learning outcomes; 2. Establish a dynamic early warning mechanism This mechanism achieves intelligent identification and intervention of abnormal states by monitoring learning behavior data in real time and building individualized learning models using machine learning algorithms. 2.1 Multi-dimensional monitoring system: Behavioral metrics: duration of a single viewing session, time period distribution, and frequency of interaction; Content metrics: vocabulary coverage, sentence complexity matching degree; Performance indicators: accuracy rate of real-time assessments and frequency of parental feedback. 2.2 Intelligent Early Warning Trigger Rules: When the system detects that "single viewing time > 30 minutes but vocabulary absorption rate < 10%", it determines the state as "passive viewing" and automatically triggers adjustment suggestions. If the theme jump rate is greater than 50% for three consecutive days (such as a sudden switch from an animal theme to a math theme), an "attention distraction" warning will be activated. Optimize suggestion push logic: Regarding the suggestion for "passive viewing": insert a 3-minute "active recall" segment, requiring children to retell key plot points of the animation; Regarding the suggestion for "distraction": provide a "theme transition animation pack" to achieve smooth transitions through character linking.

7. The method for deep learning effect correlation mining based on children's animation viewing behavior data according to claim 1, characterized in that, It also includes a high-value user screening mechanism, which establishes scientific user segmentation standards to accurately identify key factors influencing learning outcomes. This mechanism comprises the following three core modules:

1. Define the criteria for "long-term loyal users" This standard quantifies user stickiness through two dimensions to ensure that the samples included in the analysis possess continuous learning behavior characteristics: 1.1 Time Dimension Requirement: Users must meet the hard requirement of continuous system use for ≥90 days. This period is set based on research in child behavioral psychology, which shows that regular learning for 90 consecutive days can form stable learning habits (refer to "Research on the Habit Formation Cycle of Children"). 1.2 Activity Requirements: During the continuous use period, users must maintain an active number of ≥5 days per week. This setting aims to exclude users who engage in "cramming" learning and ensure that the sample reflects real learning patterns rather than short-term behavioral fluctuations.

2. Define the "higher-level" standard This standard objectively quantifies users' language proficiency development level through a tiered competency certification system, specifically including: 2.1 Advanced Test Level Design: Set at least 3 progressive evaluation nodes, each level corresponding to a specific ability development milestone, for example: Level 1: Basic vocabulary recognition (≥200 words); Level 2: Simple sentence structure (able to construct "subject + verb + object" structure); Level 3: Complex context comprehension (vocabulary ≥ 500 words, able to understand subjunctive mood); 2.2 Dynamic evaluation mechanism: Adaptive assessment technology is adopted to adjust the difficulty of questions based on the user's real-time performance, ensuring the scientific nature of the evaluation results; 3. Successfully build user-specific datasets After completing user segmentation, the system builds a high-quality analysis sample library through the following steps: 3.1 Data Cleaning Rules: Excluding "zero-evaluation users": samples that did not participate in any performance evaluation; Filter out "abnormal data segments": such as irrational data with a single-day viewing time exceeding 180 minutes; Exclude "test users": Suspected test accounts that were used more than 20 times per day within 7 days of registration; 3.2 Data Augmentation Processing: Complete the behavioral trajectories of valid samples and predict missing data segments using machine learning models; Establish a user profile tagging system to label derivative characteristics such as learning style and content preference.

8. The method for deep learning effect correlation mining based on children's animation viewing behavior data according to claim 1, characterized in that it further includes success path abstraction and transfer, which realizes the reuse of learning experience through a data-driven approach, significantly improving the learning starting point and effect of new users. This step includes the following two key links:

1. Extract common behavioral patterns of successful users This step uses cluster analysis and pattern recognition techniques to extract universally applicable learning path templates from successful user datasets, specifically including: 1.1 Pattern Feature Engineering: For young children (0-6 years old), a typical pattern was identified: "3 times a day for 10 minutes of the same theme animation for the first 30 days + 1 nursery rhyme reinforcement every 7 days". This pattern combines children's short-term memory reinforcement mechanism, establishes basic cognition through high-frequency input of the same theme (such as watching the "Marine Life" series for 3 consecutive days), and then uses nursery rhyme reinforcement (such as the theme song of "Finding Nemo") to achieve multimodal memory encoding. 1.2 For older children (7-12 years old), a "dual-theme alternating reinforcement cycle" model is abstracted, which is a cyclical structure of "5 days of specialized training (such as science theme) + 2 days of comprehensive application (such as science experiment animation)". This design is in line with the cognitive development characteristics of older children. Through the alternation of specialized breakthroughs and comprehensive applications, the deep internalization of knowledge is promoted. Parametric representation: Transform behavioral patterns into a computable set of parameters, such as "single duration = 10 minutes", "topic switching cycle = 7 days", "reinforcement frequency = 1 time / week", etc., to provide a quantitative basis for subsequent path matching; 2. By calculating path similarity, match new users with the initial path of the most similar successful users. This step employs the Dynamic Time Warping (DTW) algorithm to achieve precise matching of the paths of new users and successful users, specifically including: 2.1 Feature Vector Construction: Behavioral data generated during the initial registration period of new users (such as viewing records in the first 3 days) are transformed into multi-dimensional feature vectors, including indicators such as "single duration distribution", "topic preference coefficient" and "interaction response speed". 2.2 Similarity Calculation: The DTW algorithm is used to calculate the similarity matrix between the feature vector of new users and the path template of successful users. The algorithm optimizes path alignment through dynamic programming, which effectively solves the matching bias caused by differences in user behavior rhythm. For example, even if the viewing time of new users on the second day is 20% shorter than that of the template, the algorithm can still achieve accurate matching through time axis compression. 2.3 Recommended Initial Path: Based on similarity ranking, the system recommends the top 3 candidate paths to new users and sets an "adaptive observation period". During the observation period, the system continuously monitors the matching degree between the user's actual behavior and the recommended path. If the deviation exceeds the threshold (e.g., behavioral similarity <70% for 3 consecutive days), the path rematch mechanism is automatically triggered.

9. The method for deep learning effect correlation mining based on children's animation viewing behavior data according to claim 1, characterized in that, The method is implemented through the following system architecture: The data layer includes a behavior log library and a knowledge graph of successful user paths; The analysis layer includes a path mining engine and a real-time similarity calculation module; The recommendation layer includes a dynamic path generator, anomaly detection, and calibrator. The interaction layer includes a visual report for parents and a one-click optimization button.

10. A method for mining the correlation of deep learning effects based on children's animation viewing behavior data according to claims 1-9, characterized in that, The method is implemented through the following specific embodiments: Data pre-screening and cleaning, defining the criteria for "successful users" and constructing a dataset of successful users; Behavioral path modeling and analysis: the FP-Growth algorithm is used to mine high-frequency behavioral sequences for young children, and cluster analysis is used to identify effective patterns for older children; Dynamic recommendations and parental guidance are generated by matching new user characteristics with a historical successful user prototype library and calculating behavioral sequence similarity. The system provides initial recommendations and deviation warnings through examples of parent-side commands.

Images