Artificial intelligence-based personalized book recommendation method and system for picture book live broadcast

Abstract

The application discloses a picture book live broadcast personalized book recommendation method and system based on artificial intelligence, relates to the technical field of live broadcast book recommendation, and comprises the following steps: acquiring multi-modal interaction data of a user in a picture book live broadcast, generating a three-dimensional ability portrait vector and a psychological portrait vector representing cognitive ability and psychological tendency of the user based on the multi-modal interaction data; performing time series analysis, calculating an evolution direction vector and an evolution rate scalar; when the evolution direction is positive, comparing the evolution rate scalar with an adaptive threshold range, updating a preset coordination coefficient within the threshold range, adjusting an ability fit degree weight and an emotional resonance degree weight when the evolution rate scalar exceeds the threshold range, and dynamically updating parameters based on the weight adjustment count; and calculating a matching total score by using the dynamically adjusted parameters through a matching decision formula and outputting a picture book recommendation. The application realizes real-time adaptation to a user state in live broadcast interaction, adaptively optimizes a recommendation strategy, and significantly improves the accuracy of the instant personalization degree of the recommendation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of live book recommendation, more specifically, the present application relates to a picture book live personalized book recommendation method and system based on artificial intelligence. BACKGROUND

[0002] In the live picture book interactive scene, the existing recommendation technology cannot realize the personalized recommendation deeply integrated with real-time interaction. The core disadvantage is that the recommendation process is disconnected from the high-fidelity real-time interaction data stream, and cannot capture and respond to the changes in cognitive ability and fluctuations in emotional state exhibited by children in real-time conversations, resulting in delayed, static and lack of educational relevance.

[0003] Traditional recommendation methods rely heavily on users' past static historical data or preset tags, and their effectiveness is significantly reduced in the dynamic and spontaneous scene of live interaction. Since they cannot access and process real-time voice, expressions and conversation text and other multi-dimensional interaction data generated during the live interaction, these methods lose the best window to assess the real-time state of children. They often make recommendations based on historical behavior that is weakly related to the current interaction, which makes it difficult for the recommended results to match the immediate interest focus, language organization ability or emerging emotional needs exhibited by children in the just-ended conversation. Essentially, it is a kind of after-the-fact remedy that is out of context, rather than a synchronous guide that integrates into the interaction process.

[0004] Further, the evaluation model of the existing technology is usually single-dimensional and fixed as a static framework. They may only measure the matching degree from a single angle such as language difficulty or content theme, lacking a comprehensive evaluation system that can simultaneously quantify the cognitive expression level and psychological tendency of children. For example, they cannot analyze the emotional richness, identification of specific values and complexity of language logic exhibited by children when telling stories, and convert these multi-dimensional information into a computable recommendation basis. Therefore, their recommended results often only meet the needs of one aspect, may meet the cognitive difficulty but fail to alleviate the current mood, or cater to the surface interest but fail to challenge the ability short board, making it difficult to achieve the educational goal of coordinating ability development and emotional support. In addition, such recommendations usually lack an explainable educational development direction, and only present a list of books without explaining the specific relationship between the recommended books and the children's performance in this interaction, making the recommendation more like an isolated product exposure rather than a logically coherent educational intervention step.

[0005] To address the problem, the technical field urgently needs a technical solution that can deeply embed real-time live interactive processes, make full use of data from the live-mic process to conduct immediate multi-dimensional state assessment of children, and generate personalized picture book recommendations with clear development orientation. In short, the key defect of the prior art is that it cannot dynamically assess the cognitive and psychological state of children through the analysis of real-time interaction data in the dynamic scenario of picture book live-mic, and achieve immediate personalized recommendations based on the assessment. SUMMARY

[0006] To solve the technical problem, the present application provides a picture book live personalized book recommendation method based on artificial intelligence, which solves the problems raised in the background art.

[0007] In a first aspect, the present application provides a picture book live personalized book recommendation method based on artificial intelligence, comprising the following steps: obtaining interaction data of a target user, the interaction data including text data, speech feature data and expression feature data; based on the interaction data, generating a three-dimensional ability portrait vector sequence representing the current cognitive expression ability of the target user and a three-dimensional psychological portrait vector sequence representing the current psychological tendency of the target user; performing time series analysis on the three-dimensional ability portrait vector sequence and the three-dimensional psychological portrait vector sequence to calculate an evolution direction vector and an evolution rate scalar; when each component of the evolution direction vector is positive, determining whether the evolution rate scalar is within an adaptive threshold range, if yes, calculating the absolute difference between the evolution rate scalar and the midpoint of the adaptive threshold and updating the preset coordination coefficient based on the range limiting function boundary value comparison result; if not, and if the evolution rate scalar is lower than the lower limit of the adaptive threshold range, then no adjustment is made; if not, and if the evolution rate scalar exceeds the upper limit of the adaptive threshold range, then the ability fit degree weight and the emotional resonance degree weight are adjusted; recording the adjustment count of the ability fit degree weight and the emotional resonance degree weight, and updating the boundary value of the range limiting function accordingly; calculating the matching decision total score by matching the decision formula according to the updated preset coordination coefficient, the adjusted ability fit degree weight and the emotional resonance degree weight; and selecting at least one target picture book according to the matching decision total score and outputting it.

[0008] In a second aspect, the embodiments of the present application provide a picture book live broadcast personalized book recommendation system based on artificial intelligence, comprising: a data acquisition module for acquiring interaction data of a target user, the interaction data including text data, voice feature data and expression feature data; a vector construction module for generating a three-dimensional ability portrait vector sequence representing the current cognitive expression ability of the target user and a three-dimensional psychological portrait vector sequence representing the current psychological tendency of the target user based on the interaction data; a time series analysis module for performing time series analysis on the three-dimensional ability portrait vector sequence and the three-dimensional psychological portrait vector sequence to calculate an evolution direction vector and an evolution rate scalar; a coordination coefficient acquisition module for determining whether the evolution rate scalar is within an adaptive threshold range when each component of the evolution direction vector is positive, and if so, calculating the absolute difference between the evolution rate scalar and the midpoint of the adaptive threshold and updating the preset coordination coefficient in combination with the boundary value comparison result of the range limiting function; a judgment module for determining that no adjustment is needed if the result is negative and the evolution rate scalar is lower than the lower limit of the adaptive threshold range; a weight adjustment module for adjusting the ability fit degree weight and the emotional resonance degree weight if the result is negative and the evolution rate scalar exceeds the upper limit of the adaptive threshold range; a statistical module for recording the adjustment count of the ability fit degree weight and the emotional resonance degree weight and updating the boundary value of the range limiting function accordingly; a total score calculation module for calculating a matching decision total score by a matching decision formula according to the updated preset coordination coefficient, the adjusted ability fit degree weight and the emotional resonance degree weight; and an output module for selecting at least one target picture book according to the matching decision total score and outputting the same.

[0009] The one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0010] 1. By processing the text data, voice feature data and expression feature data generated in the live streaming link in real time and generating the three-dimensional ability portrait vector sequence and the three-dimensional psychological portrait vector sequence in real time, the recommendation decision is completely based on the instant performance of the user under the current live streaming theme. Further analysis of the evolution direction vector and the evolution rate scalar of the vector sequence can realize real-time perception of whether the user state is steadily involved or emotionally fluctuated, and dynamically adjust the preset coordination coefficient or the ability fit degree weight and the emotional resonance degree weight within seconds, so that the recommendation strategy can evolve synchronously with the interactive process, realize dynamic recommendation deeply coupled with real-time live streaming interaction, and solve the problem of recommendation lag and context disconnection.

[0011] 2. Not only the three-dimensional ability image vector is constructed from the narrative structure features, emotional vocabulary features and syntactic complexity features, but also the three-dimensional psychological image vector is constructed through the multi-modal emotion recognition model, value proposition library matching and role classification model, forming a complete quantitative description of the user's cognitive and psychological state. In the matching stage, the matching degree in ability difficulty and psychological theme is measured by the fit degree coefficient and the resonance degree coefficient respectively, and the preset synergy coefficient is introduced to strengthen the recommendation gain of both. This multi-dimensional matching mechanism can take into account the cognitive development needs and emotional resonance needs of users at the same time, and the recommendation result is more comprehensive and three-dimensional, constructing a multi-dimensional collaborative evaluation and matching model integrating cognitive ability and psychological state, solving the problem of single evaluation dimension.

[0012] 3. The adjustment events of the ability fit degree weight and the emotional resonance degree weight are recorded, and the dynamic upper limit value of the range limiting function is dynamically calculated according to the up-regulation count through the saturation growth function. When the user frequently needs emotional support, the upper limit of the weight adjustment will be appropriately relaxed, giving the strategy greater flexibility; when the user's state is stable, the upper limit will gradually recover to normal with the time decay factor. This design enables learning and adapting to the user's long-term interaction mode, preventing excessive parameter adjustment shock, while maintaining sensitivity to the immediate state, ensuring the robustness and individualization depth of the long-term recommendation strategy. BRIEF DESCRIPTION OF DRAWINGS

[0013] Figure 1 The step schematic diagram of the AI-based personalized book recommendation method for picture book live broadcast provided by the embodiments of the present application;

[0014] Figure 2 The logic flow schematic diagram of the AI-based personalized book recommendation method for picture book live broadcast provided by the embodiments of the present application;

[0015] Figure 3 The logic schematic diagram of the update of the boundary of the range limiting function provided by the embodiments of the present application;

[0016] Figure 4 The structure schematic diagram of the AI-based personalized book recommendation system for picture book live broadcast provided by the embodiments of the present application. DETAILED DESCRIPTION

[0017] The AI-based personalized book recommendation method and system for picture book live broadcast provided by the embodiments of the present application solve the technical problem of the prior art that the user's state is adapted in real time in live interaction, and the recommendation strategy is optimized adaptively, significantly improving the accuracy of the immediate individualization degree of the recommendation.

[0018] To solve the problem that existing picture book live broadcast recommendation technology is difficult to respond to children's dynamic interaction state in real time, the fundamental solution of this scheme lies in building a closed-loop processing flow that can tightly couple live data stream, quantify children's multi-dimensional state in real time, and adaptively adjust the recommendation strategy according to the state change.

[0019] This idea starts from the data source. It must directly process high-fidelity data generated in live microphone that can reflect children's immediate reaction. Therefore, it is necessary to obtain and synchronously process text data from speech stream, speech feature data extracted from audio, and expression feature data extracted from video in real time.

[0020] After obtaining the original data, the core task is to convert unstructured interaction information into machine-calculable and comparable standardized state indicators. This requires designing two parallel quantification systems.

[0021] On the one hand, from the perspective of cognitive development, analyze the language organization logic, emotional vocabulary use in text data, and sentence complexity in speech data, and through a pre-set feature dimension mapping model, quantify them into scores on three orthogonal dimensions of language ability, emotional expression, and cultural cognition, forming a three-dimensional ability portrait vector representing cognitive expression ability.

[0022] On the other hand, from the perspective of psychology and emotion, fuse and analyze the intonation and speech rate features in speech and facial action features in expression to identify the dominant emotion category; at the same time, analyze the value tendency keywords contained in the text; then, combined with the live broadcast theme, judge the children's role identification in interaction. Finally, map the emotion category, value tendency, and role type into three dimensions of emotional state intensity, value tendency intensity, and role projection intensity, forming a three-dimensional psychological portrait vector representing psychological tendency. These two vectors together form the digital twin of children's current comprehensive state.

[0023] However, a static snapshot is not enough to support dynamic recommendation. To capture the evolving trend of children's state, it is necessary to conduct time series analysis on the continuously generated ability and psychological portrait vector sequence. By calculating the change direction and rate of the vector sequence, we get the evolution direction vector representing the overall development trend and the evolution rate scalar representing the change intensity. These two dynamic features become the key basis for determining whether children are in a stable engagement or emotional fluctuation mode.

[0024] Based on the results of dynamic perception, the corresponding picture book recommendation strategy needs to be adapted in real time and accurately. If the child shows a positive and stable progress trend, enter the state adaptation adjustment mode, at this time the optimization goal is to consolidate the learning state, so the synergy coefficient in the matching decision formula will be calculated and appropriately increased to enhance the emphasis on the synergistic effect of ability matching and interest resonance. If the child shows a positive but too intense emotional fluctuation, which may be accompanied by attention dispersion, enter the emotional adaptation adjustment mode, at this time emotional support needs to be provided first, so the linkage adjustment of the weight of ability fit degree and emotional resonance degree will be started, and the decision weight of emotional factors will be increased, so that the recommendation will tilt towards the content that can arouse the current emotional resonance.

[0025] In order to have long-term adaptability and stability, and avoid strategy oscillation, a self-optimization mechanism at the strategy level is further introduced. The historical frequency of emotional weight adjustment is recorded, and the amplitude boundary allowed by the weight adjustment itself is dynamically adjusted according to the frequency. When the child frequently needs emotional attention, the upper limit of adjustment will be appropriately relaxed, giving the strategy more flexibility; when the child's state returns to stable, the boundary will gradually shrink to normal over time. This dynamic boundary mechanism based on historical behavior feedback enables the learning and adaptation of the long-term interaction mode of the user.

[0026] Finally, using the parameters adjusted dynamically and bounded, combined with the child's current ability and psychological portrait vector, a comprehensive matching degree score is calculated for each candidate reading in the picture book database through the matching decision formula, and the picture book most suitable for the child's immediate state and development needs is output. The whole process realizes a complete closed loop from real-time data perception, state quantification, to strategy dynamic adjustment and self-optimization, so that the recommendation can be deeply integrated into the interaction process and have a clear educational development orientation.

[0027] In order to better understand the technical solutions, the technical solutions will be described in detail below in conjunction with the drawings in the specification and specific embodiments.

[0028] Figure 1The step schematic diagram of the personalized book recommendation method based on the AI picture book live broadcast provided by the embodiment of the application includes the following steps: obtaining interactive data of a target user, the interactive data including text data, voice feature data, and expression feature data; based on the interactive data, generating a three-dimensional ability portrait vector sequence representing the current cognitive expression ability of the target user and a three-dimensional psychological portrait vector sequence representing the current psychological tendency of the target user; performing time series analysis on the three-dimensional ability portrait vector sequence and the three-dimensional psychological portrait vector sequence to calculate an evolution direction vector and an evolution rate scalar; when each component of the evolution direction vector is positive, determining whether the evolution rate scalar is within a self-adaptive threshold range, if yes, calculating an absolute difference between the evolution rate scalar and a midpoint of the self-adaptive threshold and updating a preset coordination coefficient according to a range limiting function boundary value comparison result; if no, and the evolution rate scalar is lower than a lower limit of the self-adaptive threshold range, no adjustment is performed; if no, and the evolution rate scalar exceeds an upper limit of the self-adaptive threshold range, the ability fit degree weight and the emotional resonance degree weight are adjusted; the adjustment count of the ability fit degree weight and the emotional resonance degree weight is recorded, and the boundary value of the range limiting function is updated accordingly; according to the updated preset coordination coefficient, the adjusted ability fit degree weight and the emotional resonance degree weight, a matching decision total score is calculated through a matching decision formula; and at least one target book is selected according to the matching decision total score and output.

[0029] Figure 2 The logic flow schematic diagram of the personalized book recommendation method based on the AI picture book live broadcast provided by the embodiment of the application.

[0030] By obtaining the multi-modal interactive data of the target user in real time, and dynamically generating the three-dimensional portrait vector sequences representing the cognitive expression ability and the psychological tendency of the target user, the evolution direction and rate of the user state can be captured through time series analysis on these sequences, and the coordination coefficient and the weights in the recommendation strategy are adaptively adjusted accordingly.

[0031] Therefore, the deep fusion evaluation and instant response of the cognitive and psychological state of children in the picture book live broadcast intercommunication are realized, the limitations of the traditional recommendation lag, staticity, and lack of education targeting are overcome, and more dynamic, collaborative, and development-oriented personalized book recommendations can be provided.

[0032] Further, the specific generation process of the three-dimensional ability portrait vector is: extracting narrative structure features and emotional vocabulary features from the text data; extracting syntactic complexity features from the speech feature data; inputting the narrative structure features, the emotional vocabulary features and the syntactic complexity features into a preset feature dimension mapping model, the feature dimension mapping model outputs a score quantity representing the target user in the preset three orthogonal ability dimensions, the three orthogonal ability dimensions include language ability dimension, emotional expression dimension and cultural cognition dimension; the score quantity in the three orthogonal ability dimensions is normalized, and a three-dimensional vector is constructed in the preset orthogonal three-dimensional space and used as the three-dimensional ability portrait vector.

[0033] In the embodiment, the narrative structure feature refers to the mode of the text content in logical organization, plot development and information presentation. Its role is to reflect the logical thinking ability, information organization ability and understanding ability of complex concepts of the target user.

[0034] In specific implementation, natural language processing techniques can be used, such as identifying the dependency relationship between sentences through syntactic analysis, identifying the connection and transition between paragraphs through discourse analysis, identifying the evolution of the core topic of the text through theme chain, or tracking the consistency of entities in the text through reference resolution technology. These technologies can quantify the coherence, completeness and complexity of the text, thereby representing the narrative ability of the user.

[0035] The emotional vocabulary feature refers to the words or phrases expressing emotional tendency in the text. Its role is to reflect the emotional understanding ability, emotional expression richness and empathy ability for others' emotions of the target user.

[0036] In specific implementation, a preset emotion dictionary can be constructed or used to match the words in the text, and the appearance frequency and intensity of positive, negative or neutral emotional vocabulary are counted. In addition, a deep learning-based emotion classification model can also be used to judge the overall emotional tendency of the text and extract key emotional expression words as features.

[0037] The syntactic complexity feature refers to the complexity of sentence structure, vocabulary use and grammar rules in speech expression. Its role is to reflect the language organization ability, cognitive load and fluency and accuracy of spoken language expression of the target user.

[0038] In specific implementation, the speech feature data can be first converted into text data through speech recognition technology, and then the converted text is subjected to syntactic analysis, such as counting average sentence length, number of clauses, vocabulary diversity, and frequency of complex sentence patterns. In addition, prosodic features can also be directly extracted from the speech signal as an indirect indicator of syntactic complexity, because these prosodic features are often associated with the syntactic structure and cognitive load of spoken language expression.

[0039] The preset feature dimension mapping model is a trained machine learning model, which can be a multilayer perceptron, a support vector machine or a more complex neural network model. The role of this model is to effectively integrate and map the low-level and diverse features extracted from the original interaction data to higher-level and more interpretable ability dimensions.

[0040] In specific implementation, the model establishes a nonlinear relationship between features and ability dimension scores by learning a large amount of labeled data. The language ability dimension refers to the comprehensive level of the target user in vocabulary mastery, grammar use, expression fluency and language organization. This dimension score quantifies the user's ability to effectively communicate in language. The emotional expression dimension refers to the target user's ability to understand others' emotions, express their own emotions, and show emotional empathy in communication.

[0041] The dimension score quantifies the level of the user's emotional intelligence. The cultural cognition dimension refers to the target user's understanding and use of specific cultural backgrounds, social customs, allusions and values. This dimension score quantifies the user's breadth and depth in cross-cultural communication and understanding. The three dimensions are designed to be orthogonal, meaning they are conceptually independent of each other and can comprehensively evaluate the user's cognitive expression ability from different perspectives.

[0042] The present scheme can also accurately capture the cognitive expression ability of the target user from multi-modal interaction data. Specifically, by extracting narrative structure features and emotional vocabulary features from text data, and extracting syntactic complexity features from voice feature data, it ensures comprehensive consideration of the user's language organization ability, emotional understanding and expression ability, and cognitive complexity. These features are processed by the preset feature dimension mapping model to output quantified scores in the three orthogonal dimensions of language ability, emotional expression and cultural cognition, avoiding the limitations of single-dimensional evaluation. The scores are normalized and constructed into a three-dimensional vector, making the user ability portrait have a unified dimension and intuitive geometric representation, greatly improving the accuracy and interpretability of the three-dimensional ability portrait vector, providing a solid and refined foundation for subsequent picture book matching decisions, and significantly improving the personalization and precision of picture book recommendations.

[0043] Further, the specific generation process of the three-dimensional psychological portrait vector is as follows: the tone and speed parameter sequence in the voice feature data is fused with the facial action unit parameter sequence in the expression feature data, and is input into a preset multi-modal emotion recognition model to output the current emotional state category of the target user; the text data is subjected to semantic analysis and keyword extraction, and is matched with a preset value proposition vocabulary to identify the core value tendency category implied in the expression of the target user; the emotional state category, the core value tendency category, and the theme tag data associated with the live streaming and microphone session are collectively input into a preset role classification model to output the role projection type of the target user in the current interaction; the emotional state category, the core value tendency category, and the role projection type are respectively mapped to the initial values of the emotional state intensity, the value tendency intensity, and the role projection intensity according to a preset quantitative mapping table; and the initial values of the three dimensions are normalized to construct a three-dimensional vector in a preset orthogonal three-dimensional space as the three-dimensional psychological portrait vector.

[0044] In the present embodiment, the preset multi-modal emotion recognition model is a trained machine learning or deep learning model for receiving the fused multi-modal features and identifying the current emotional state category of the user therefrom. The model is usually trained based on a large multi-modal emotion annotation data set and can map the fused features to preset emotional categories such as joy, sadness, surprise, calmness, etc. Through the model, the instantaneous emotional state of the user in the live streaming interaction can be objectively judged.

[0045] Semantic analysis and keyword extraction are processes for in-depth understanding of text data. Semantic analysis aims to reveal the deep meaning and contextual relationship of the text, while keyword extraction focuses on identifying the most important words or phrases in the text. These techniques can be implemented using word embedding, topic modeling, dependency syntax analysis, etc. in natural language processing. They are used to extract key information from user input text that represents the core ideas and concerns of the user.

[0046] The preset value proposition vocabulary is a pre-constructed set of words or phrases related to different core value tendencies. For example, the vocabulary may include value labels such as courage, exploration, kindness, responsibility, and their corresponding synonyms, near-synonyms, or related expressions. By matching the keywords extracted from the user's text with the vocabulary, the core value tendency category implied in the user's expression or valued by the user can be identified.

[0047] The theme tag data associated with the live streaming and microphone session refers to the preset label information related to the content of the interactive session of the live streaming. These labels can describe the theme, content type, educational goal, etc. of the live streaming, such as science enlightenment, emotional intelligence cultivation, historical stories, art appreciation, etc. These labels provide important contextual information for understanding the psychological activities of the user in a specific situation.

[0048] The preset role classification model is a model used to determine the role type played or projected by the user in the current interaction based on the user's multi-dimensional information (emotional state category, core value tendency category, and theme tag data associated with the live streaming segment). For example, the user can be classified as an active participant, a curious explorer, an emotional resonator, a critical thinker, etc. This model is usually trained through a supervised learning method, using interaction data with role labels to learn classification rules.

[0049] The preset quantitative mapping table is a lookup table or function that converts qualitative emotional state categories, core value tendency categories, and role projection types into quantitative numerical values. For example, different emotional categories can be mapped to different emotional intensity values, different value tendencies can be mapped to intensity values between 0 and 1, and different role types can also be mapped to corresponding intensity values. This mapping table ensures that the psychological characteristics of different categories can be uniformly quantified for subsequent mathematical operations.

[0050] After these multi-dimensional and refined psychological characteristics are quantified and normalized, a three-dimensional psychological portrait vector with more representativeness and discriminability is constructed. This vector can more accurately reflect the user's real psychological state and potential needs, providing a solid foundation for subsequent picture book matching decisions, significantly improving the accuracy of personalized recommendations and user satisfaction.

[0051] Further, the specific acquisition process of the target picture book is as follows: for the candidate picture book in the picture book database, the pre-stored difficulty feature vector and the theme feature vector corresponding to the candidate picture book are acquired; the difficulty feature vector is composed of preset difficulty level values of the candidate picture book in three dimensions of language ability, emotional expression and cultural cognition; the theme feature vector is composed of preset theme intensity values of the candidate picture book in three dimensions of emotional state, value tendency and role projection; the absolute value of the difference between the cosine similarity and the module length of the three-dimensional ability portrait vector and the difficulty feature vector is calculated, and the cosine similarity and the absolute value of the difference between the module lengths are weighted and averaged according to a preset first weight to generate a fit coefficient; the dot product between the three-dimensional psychological portrait vector and the theme feature vector is calculated as a psychological theme fit degree, and the vector components representing the emotional state in the three-dimensional psychological portrait vector and the theme feature vector are extracted respectively, the standardized absolute difference value is calculated and recorded as an emotional matching degree; the psychological theme fit degree and the emotional matching degree are weighted and averaged according to a preset second weight to generate a resonance coefficient; a first weighted component obtained by multiplying the fit coefficient by the ability fit degree weight, a second weighted component obtained by multiplying the resonance coefficient by the emotional resonance weight, and a synergy gain component obtained by multiplying the product of the fit coefficient and the resonance coefficient by a preset synergy coefficient, are added together to obtain a matching decision total score; the candidate picture book whose matching decision total score exceeds a recommendation threshold is selected as the target picture book.

[0052] In the present embodiment, in the process of acquiring the target picture book, the pre-stored difficulty feature vector and the theme feature vector of the candidate picture book need to be acquired from the picture book database first. The difficulty feature vector is used to quantify the complexity of the picture book in the cognitive level, which can be manually annotated by experts according to the content of the picture book, or automatically extracted by a natural language processing model to analyze the text complexity or by a machine learning model.

[0053] In the language ability dimension, the vocabulary difficulty and syntactic complexity of the picture book can be evaluated; in the emotional expression dimension, the richness of emotional vocabulary and the fluctuation of emotional curve in the picture book can be evaluated; in the cultural cognition dimension, the background knowledge required by the picture book and the depth of cultural metaphor can be evaluated.

[0054] These values are usually standardized to facilitate subsequent calculation and comparison. The theme feature vector is used to describe the content tendency of the picture book in the psychological and emotional level, which can be obtained by manual annotation, keyword extraction, theme model analysis or emotion analysis model based on the content of the picture book. For example, the emotional state dimension can evaluate the emotion mainly conveyed by the picture book; the value tendency dimension can evaluate the values advocated by the picture book; the role projection dimension can evaluate the type of the role in the picture book and its attractiveness to the reader. These values are also usually standardized.

[0055] After obtaining the feature vector of the picture book, the absolute value of the difference between the cosine similarity and the module length of the three-dimensional ability portrait vector and the difficulty feature vector is calculated.

[0056] The cosine similarity is used to measure the similarity of the directions of the two vectors, that is, the relative matching degree of the user's ability portrait and the difficulty feature of the picture book in each dimension. A higher cosine similarity indicates that the user's ability and the difficulty of the picture book are highly consistent in structure.

[0057] The absolute value of the difference in module length is used to measure the absolute intensity difference of the two vectors in each dimension. A smaller difference in module length indicates that the user's ability and the difficulty of the picture book are close in overall level. Then, the cosine similarity and the absolute value of the difference in module length are weighted and averaged according to a preset first weight, thereby generating a fit coefficient. The preset first weight is used to balance the contribution of the cosine similarity and the difference in module length in the calculation of the fit coefficient, which can be set according to experience or obtained through machine learning model training to optimize the recommendation effect. The fit coefficient comprehensively reflects the matching degree between the user's cognitive expression ability and the difficulty of the picture book, and a higher value indicates a better matching degree.

[0058] The dot product between the three-dimensional psychological portrait vector and the theme feature vector is also calculated as the psychological theme fit. The dot product is used to measure the comprehensive matching degree of the two vectors in direction and size. When the directions of the two vectors are consistent and the module length is large, the dot product value is large, indicating that the user's psychological tendency and the theme of the picture book are highly consistent in the whole.

[0059] In addition, the vector components representing emotional states in the three-dimensional psychological portrait vector and the theme feature vector are also extracted respectively, and the standardized absolute difference is calculated and recorded as the emotional matching degree. The vector component of the emotional state refers to the component in the three-dimensional psychological portrait vector and the theme feature vector that is specifically used to describe the emotional dimension. The standardized absolute difference measures the absolute difference between the user's current emotional state and the emotional state expressed by the picture book, and is standardized to be between 0 and 1, with 0 indicating complete matching and 1 indicating complete mismatch.

[0060] The psychological theme fit reflects the overall matching degree between the user's psychological tendency and the theme of the picture book, while the emotional matching degree specifically reflects the matching degree between the user's emotional state and the emotional expression of the picture book. Then, the psychological theme fit and the emotional matching degree are weighted and averaged according to a preset second weight to generate a resonance coefficient. The preset second weight is used to balance the contribution of the psychological theme fit and the emotional matching degree in the calculation of the resonance coefficient, which can be adjusted according to user feedback or expert experience to emphasize the user's identification of the overall theme or resonance of specific emotions. The resonance coefficient comprehensively reflects the emotional resonance degree between the user's psychological tendency and the theme of the picture book, and a higher value indicates a stronger resonance.

[0061] The matching degree coefficient is multiplied by the ability matching degree weight to obtain a first weighted component, and the resonance degree coefficient is multiplied by the emotional resonance degree weight to obtain a second weighted component. The ability matching degree weight and the emotional resonance degree weight are used to adjust the matching degree of the cognitive ability of the user and the difficulty of the picture book and the relative importance of the matching degree of the psychological tendency of the user and the theme of the picture book in the final recommendation decision. They are dynamic adjustment parameters according to the user interaction data and behavior evolution, to adapt to the different needs of the user at different stages for cognitive challenge and emotional satisfaction.

[0062] For example, when it is judged that the user is more inclined to cognitive development, the ability matching degree weight can be adjusted higher; when the user pays more attention to emotional experience, the emotional resonance degree weight can be adjusted higher. In addition, the product of the matching degree coefficient and the resonance degree coefficient is multiplied by a preset synergy coefficient to obtain a synergy gain component. The preset synergy coefficient is used to quantify and amplify the positive synergy effect when both cognitive matching and emotional resonance exist. When the user is matched with the picture book in cognition and also resonates in emotion, the coefficient can further improve the priority of the recommendation, so as to identify those picture books that can provide a comprehensive and high-quality experience. The coefficient is also dynamically adjusted according to the evolution trend of the user state.

[0063] Finally, the first weighted component, the second weighted component and the synergy gain component are all added together to obtain a matching decision total score. The matching decision total score comprehensively considers the matching of the user's ability and the difficulty of the picture book, the resonance of the user's psychology and the theme of the picture book, and the synergy between the two, and is the basis for the final recommendation decision. Finally, the candidate picture books with a matching decision total score exceeding a recommendation threshold are selected as the target picture books. The recommendation threshold is a preset value used to filter out picture books that are sufficient to meet the user's needs. The threshold can be dynamically adjusted or optimized according to historical recommendation effects, user satisfaction and other indicators.

[0064] The scheme also provides a fine and dynamic picture book recommendation decision mechanism. By introducing the difficulty feature vector and the theme feature vector, the internal attributes of the picture book are quantitatively matched with the user portrait, ensuring the objective basis of the recommendation. By calculating the matching degree coefficient and the resonance degree coefficient, the matching degree of the user and the picture book is evaluated from two core dimensions of cognitive ability and psychological emotion, avoiding the limitations of single-dimensional recommendation.

[0065] The dynamically adjusted ability matching degree weight, emotional resonance degree weight and preset synergy coefficient are integrated into the calculation of the matching decision total score, so that the recommendation can flexibly adjust the recommendation strategy according to the real-time changes of the user state. For example, when the cognitive ability of the user improves, the ability matching degree weight can be appropriately increased to recommend more challenging picture books; when the emotional needs of the user are strong, the emotional resonance degree weight can be increased to recommend picture books that can touch the heart of the user.

[0066] In particular, the introduction of the synergistic gain component can capture the chemical reaction when cognitive and emotional double matching, effectively identify those high-quality picture books that can meet the user's cognitive development and cause strong emotional resonance, thereby significantly improving the accuracy of the recommendation and user satisfaction, solving the problem that static matching based on user portraits and picture book features may lead to insufficient personalization and inability to adapt to dynamic changes in users.

[0067] Further, the specific updating process of the preset synergistic coefficient is as follows: within the evolution time window, the three-dimensional ability portrait vector and the three-dimensional psychological portrait vector are obtained every preset time interval, and a time series of three-dimensional ability portrait vectors and a time series of three-dimensional psychological portrait vectors are obtained; the time series of three-dimensional ability portrait vectors and the time series of three-dimensional psychological portrait vectors are subjected to first-order difference operation respectively, and the ability change direction sub-vector and the psychological change direction sub-vector are obtained; the weighted average vector of the ability change direction sub-vector and the psychological change direction sub-vector in the evolution time window is calculated, and the evolution direction vector is obtained; the Euclidean distance between the three-dimensional ability portrait vectors of adjacent preset time intervals in the time series of three-dimensional ability portrait vectors and the Euclidean distance between the three-dimensional psychological portrait vectors of adjacent preset time intervals in the time series of three-dimensional psychological portrait vectors are calculated respectively, and two rate sub-sequences are obtained; the values of the two rate sub-sequences are normalized and added, and the evolution rate scalar is obtained; when the vector direction of the evolution direction vector is in the positive direction and the evolution rate scalar is within the adaptive threshold range, it is judged as the state adaptation adjustment mode, and the preset synergistic coefficient is updated accordingly.

[0068] In this embodiment, within the evolution time window, the three-dimensional ability portrait vector and the three-dimensional psychological portrait vector of the target user at different times are continuously obtained at a preset time interval.

[0069] The evolution time window can be a preset fixed time length, for example, 10 minutes or the duration of a live session, for capturing the short-term dynamics of the user state.

[0070] The preset time interval determines the frequency of data sampling, for example, once every 5 seconds or 1 minute, to balance real-time performance and computational overhead. In this way, sequence data reflecting the changes of user ability and psychological state over time, i.e., the time series of three-dimensional ability portrait vectors and the time series of three-dimensional psychological portrait vectors, can be constructed, laying a foundation for subsequent dynamic analysis.

[0071] On this basis, in order to capture the instantaneous changes of the user state, the time series of three-dimensional ability portrait vectors and the time series of three-dimensional psychological portrait vectors are subjected to first-order difference operation respectively.

[0072] The first-order difference operation can obtain a series of ability change direction sub-vectors and psychological change direction sub-vectors by calculating the difference between adjacent vectors in the sequence.

[0073] Each capability change direction sub-vector represents the change direction and magnitude of the user's language ability, emotional expression, and cultural cognition at adjacent time points.

[0074] Similarly, each psychological change direction sub-vector reflects the instantaneous change of the user in the dimensions of emotional state, value inclination, and role projection. These sub-vectors provide detailed information on the micro-dynamics of the user's state.

[0075] To obtain the overall evolution trend of the user's state, the capability change direction sub-vectors and the psychological change direction sub-vectors are weighted and averaged over the evolution time window, thereby obtaining an evolution direction vector. The weighted average can give different sub-vectors different weights according to the time distance or other strategies, to more accurately reflect the current trend. This evolution direction vector integrates the micro-changes of the user's ability and psychological state, and represents the macro-development trend of the user's cognitive expression ability and psychological inclination within the entire evolution time window. For example, when all components of the evolution direction vector are positive, it usually means that the user presents positive growth or interest improvement in multiple dimensions.

[0076] To quantify the speed of the user's state change, the Euclidean distances between adjacent three-dimensional capability portrait vectors in the sequence of time-series three-dimensional capability portrait vectors are calculated, as well as the Euclidean distances between adjacent three-dimensional psychological portrait vectors in the sequence of time-series three-dimensional psychological portrait vectors. The Euclidean distance can intuitively measure the difference between two points in a multi-dimensional space, which is used here to represent the change speed of the user's ability and psychological state within the adjacent time interval, thereby obtaining two rate sub-sequences. Subsequently, the values of these two rate sub-sequences are normalized to eliminate dimensional differences, and they are added together to finally obtain a single evolution rate scalar. This scalar comprehensively reflects the overall change speed of the user's ability and psychological state within the evolution time window.

[0077] When the vector direction of the evolution direction vector is all in the positive direction, and the evolution rate scalar is within the adaptive threshold range, it is judged that the current state is in the state adaptation adjustment mode. The components of the evolution direction vector are all in the positive direction, indicating that the user presents positive development in all key ability and psychological dimensions. The adaptive threshold range is a dynamically adjusted interval used to determine whether the evolution rate scalar is in a moderate, neither too fast nor too slow, stable state. Only when the user's state meets such positive and stable conditions, is it considered that this is the best time to update the preset coordination coefficient, and the coordination coefficient is updated accordingly.

[0078] The scheme also provides a more detailed and dynamic preset synergy coefficient updating mechanism. By continuously monitoring and generating a time series vector sequence within an evolution time window, and performing a first-order difference operation to capture instantaneous changes, the evolution direction vector and evolution rate scalar of the user's ability and psychological state are accurately quantified through weighted average and Euclidean distance calculation.

[0079] This detailed dynamic analysis enables accurate identification of whether the user is in an active and stable state adaptation adjustment mode. When the user is in this mode, i.e., both their ability and psychological state are showing positive trends and the change rate is moderate, the preset synergy coefficient is updated. This avoids inappropriate coefficient adjustment when the user's state is unstable or changes too fast / slow, ensuring that the synergy coefficient is updated more accurately and effectively. Ultimately, this helps improve the accuracy of the matching decision total score, enabling the picture book recommendation to respond more timely and accurately to the user's current learning and psychological needs, significantly improving the personalization of the recommendation and the user experience.

[0080] Further, the specific updating process of the preset synergy coefficient also includes: taking the included angle between the evolution direction vector and the preset positive reference vector as input, processing it through the inverse tangent function to obtain a basic adjustment factor; calculating the absolute difference between the evolution rate scalar and the midpoint of the adaptive threshold, and processing the difference through a negative exponential function to obtain a rate stability decay factor; multiplying the basic adjustment factor and the rate stability decay factor, and the product is the increment of the synergy coefficient in this calculation; after obtaining the increment, add the preset synergy coefficient and the increment to obtain a preliminary update value; if the preliminary update value is lower than the lower limit of the range limiting function, the lower limit value of the range limiting function is output as the output result, if the preliminary update value is higher than the upper limit of the range limiting function, the upper limit value of the range limiting function is output as the output result; if the preliminary update value is between the upper and lower limits of the range limiting function, the preliminary update value is output as the output result; update the preset synergy coefficient with the output result.

[0081] In this embodiment, the included angle between the evolution direction vector and the preset positive reference vector is taken as input, and the inverse tangent function is used to process it to obtain a basic adjustment factor. This is used to quantify the degree of agreement between the evolution direction of the three-dimensional ability portrait vector sequence representing the target user's cognitive expression ability and the three-dimensional psychological portrait vector sequence representing the psychological tendency, and the expected positive growth direction.

[0082] A positive reference vector is preset, which represents an ideal trend of user ability and mental health development, for example, a vector with all positive components in a three-dimensional space. By calculating the included angle between the current evolution direction vector and the positive reference vector, the consistency of the two directions can be intuitively reflected. The smaller the included angle, the closer the evolution direction to the ideal positive trend. The included angle is input into an arctangent function, which can map the angle value into a continuous factor with specific gain characteristics, i.e., a basic adjustment factor. For example, when the included angle is 0, the arctangent function outputs the maximum value, indicating complete positivity; as the included angle increases, the arctangent function output value gradually decreases, so that when the user evolution direction is more consistent with the ideal positive trend, the basic adjustment factor is larger, providing stronger positive driving force for subsequent adjustment of the synergy coefficient.

[0083] At the same time, the absolute difference between the evolution rate scalar and the midpoint of the adaptive threshold is calculated, and the difference is processed by a negative exponential function to obtain a rate stability decay factor, which is used to evaluate the stability of the target user ability and mental evolution rate, i.e., whether its change speed is in an ideal and stable interval. First, the absolute difference between the evolution rate scalar and the preset adaptive threshold midpoint is calculated. The adaptive threshold midpoint represents the most suitable user evolution speed. The smaller the absolute difference, the closer the user evolution rate to the ideal stable state. Subsequently, the absolute difference is input into a negative exponential function for processing. The negative exponential function has the characteristic of rapid decay with the increase of the input value. Therefore, when the evolution rate scalar is closer to the adaptive threshold midpoint, the absolute difference is smaller, and the rate stability decay factor output by the negative exponential function is larger, indicating that the evolution rate is more stable; on the contrary, when the evolution rate deviates from the midpoint more, the decay factor is smaller, thereby introducing an inhibitory effect in the subsequent adjustment of the synergy coefficient to avoid excessive adjustment under unstable or too fast / slow evolution rate.

[0084] On this basis, the basic adjustment factor and the rate stability decay factor are multiplied, and the product is used as the increment of the synergy coefficient in this calculation. This step comprehensively considers the positivity of the user evolution direction and the stability of the evolution rate to determine the adjustment amplitude of the preset synergy coefficient in this calculation. The basic adjustment factor reflects the degree of fit between the user evolution direction and the ideal positive trend, while the rate stability decay factor reflects the stability of the evolution rate. Multiplying these two factors can ensure that only when the user evolution direction is positive and the evolution rate is stable, the synergy coefficient can obtain a larger increment.

[0085] For example, even if the evolution direction is very active, but if the evolution rate is extremely unstable, the increment will be correspondingly reduced, avoiding aggressive adjustment of the synergy coefficient when the user state fluctuates greatly. Conversely, if the evolution direction is not clear, even if the rate is stable, the increment will be smaller. This multiplication combination mechanism makes the adjustment of the synergy coefficient more prudent and reasonable.

[0086] After obtaining the increment, the preset synergy coefficient is added to the increment to obtain a preliminary update value. After calculating the increment of the synergy coefficient, it is directly added to the current preset synergy coefficient to obtain a preliminary update value. This step is the basis for the actual adjustment of the synergy coefficient, and embodies the strategy of dynamic adaptation according to the current evolution state of the user.

[0087] If the preliminary update value is lower than the lower limit of the range limiting function, the range limiting function outputs the lower limit value as the output result, and if the preliminary update value is higher than the upper limit of the range limiting function, the range limiting function outputs the upper limit value as the output result; if the preliminary update value is between the upper and lower limits of the range limiting function, the preliminary update value is taken as the output result; update the preset synergy coefficient with the output result. This is to ensure that the update of the preset synergy coefficient always remains within a reasonable and effective range, preventing it from being too high or too low, thereby avoiding unreasonable bias on the picture book recommendation result.

[0088] The range limiting function sets the minimum and maximum values of the synergy coefficient. If the preliminary update value exceeds this preset range, it will be forcibly limited to the corresponding boundary value. For example, if the preliminary update value is less than the lower limit, the lower limit value is taken; if it is greater than the upper limit, the upper limit value is taken. If the preliminary update value is between the upper and lower limits, the preliminary update value is directly adopted. Through this limiting processing, the dynamic range of the synergy coefficient can be effectively controlled, ensuring its stability and effectiveness in recommendation, and avoiding the imbalance of the recommendation effect caused by extreme values.

[0089] The scheme also provides a more refined and intelligent preset synergy coefficient updating mechanism. By introducing the angle between the evolution direction vector and the preset positive reference vector to quantify the direction fit degree, and using the arctangent function to generate a basic adjustment factor, the adjustment of the synergy coefficient can more sensitively respond to the degree of the user's active growth trend. At the same time, by calculating the absolute difference between the evolution rate scalar and the midpoint of the adaptive threshold, and combining the negative exponential function to generate a rate stability attenuation factor, the stability of the user's evolution rate is effectively evaluated, avoiding inappropriate adjustments when the user's state fluctuates greatly. Multiplying these two factors as the increment of the synergy coefficient ensures that the update of the synergy coefficient is based on the comprehensive consideration of the user's active and stable evolution, so that the adjustment is more prudent and reasonable. Finally, the preliminary update value is constrained by the range limiting function, ensuring that the synergy coefficient is always within an effective and reasonable interval. This refined updating strategy enables more accurate capture of subtle changes in the target user's cognitive expression ability and psychological tendency, thereby achieving more accurate and dynamic synergy effects in the personalized book recommendation process during the picture book live broadcast, significantly improving the degree of personalization and user satisfaction.

[0090] Further, the specific adjustment process of the ability fit degree weight and the emotional resonance degree weight is: through clustering analysis on the time-series three-dimensional psychological portrait vector sequence, the average Euclidean distance of all vector points to the clustering centroid is calculated, and after normalization processing, the average Euclidean distance is taken as the state dispersion scalar; the two components corresponding to the language ability and emotional expression dimensions of the three-dimensional ability portrait vector in the evolution direction vector are extracted, the square sum of the two components is calculated, and the root mean square is obtained, to obtain the ability active change intensity; the two components corresponding to the value tendency and role projection dimensions of the three-dimensional psychological portrait vector in the evolution direction vector are extracted, the sum of the absolute values of the two components is calculated, to obtain the psychological directional change intensity; the ability active change intensity and the psychological directional change intensity are multiplied, and the product is square root operated to obtain the expression coordination factor representing the coordinated change of cognition and psychology; the ratio of the state dispersion scalar and the expression coordination factor is calculated to obtain the attention density coefficient; the emotional resonance degree weight, the state dispersion scalar and the attention density coefficient are multiplied to obtain the adjustment amount; the emotional resonance degree weight is added to the adjustment amount to obtain the new emotional resonance degree weight; the new emotional resonance degree weight and the ability fit degree weight are normalized again to obtain the adjusted ability fit degree weight and the adjusted emotional resonance degree weight.

[0091] In the embodiment, by clustering analysis on the time-series three-dimensional psychological portrait vector sequence, the average Euclidean distance of all vector points to the clustering centroid is calculated, and after normalization processing, the average Euclidean distance is taken as the state dispersion scalar, aiming to quantify the stability and volatility of the target user's psychological state within a period of time.

[0092] In implementation, clustering algorithms such as K-means, DBSCAN, or Gaussian Mixture Model can be employed to cluster the time-series three-dimensional psychological portrait vector sequence within the preset time window, so as to identify the typical patterns or central tendencies of the user's psychological state. Subsequently, the Euclidean distance of each vector point to the centroid of its belonging cluster is calculated, and the average of these distances is obtained to reflect the concentration or dispersion degree of the psychological state. Finally, the average Euclidean distance is mapped to a preset numerical range through methods such as min-max normalization or Z-score normalization, thereby obtaining the state dispersion scalar. The higher the scalar value, the greater the volatility of the user's psychological state.

[0093] The two components corresponding to the language ability and emotional expression dimensions of the three-dimensional ability portrait vector in the evolution direction vector are extracted, and the square sum and root mean square of the two components are calculated to obtain the ability positive change intensity, which is used to evaluate the positive development trend intensity of the user in the two key cognitive dimensions of language ability and emotional expression.

[0094] The evolution direction vector reflects the overall change direction of the user's ability and psychological state. By identifying and extracting the components related to the language ability dimension and the emotional expression dimension from the vector, the specific aspects of the user's cognitive ability development can be focused on. Subsequently, the square sum and root mean square of the two components are calculated, i.e., the Euclidean norm of the sub-vector composed of them is calculated, thereby obtaining a comprehensive quantitative value representing the intensity of positive change of the user in the two ability dimensions.

[0095] The two components corresponding to the value tendency and role projection dimensions of the three-dimensional psychological portrait vector in the evolution direction vector are extracted, and the sum of the absolute values of the two components is calculated to obtain the psychological orientation change intensity, which aims to measure the change amplitude of the user in the two psychological dimensions of value tendency and role projection. Similar to the ability positive change intensity, this process extracts the components corresponding to the value tendency dimension and the role projection dimension from the evolution direction vector.

[0096] Since psychological changes may involve directionality, by calculating the sum of the absolute values of the two components, the overall intensity of the user's change in the two psychological dimensions can be comprehensively reflected, without distinguishing whether the evolution is positive or negative.

[0097] The ability positive change intensity and the psychological orientation change intensity are multiplied, and the product is square-rooted to obtain the expression coordination factor representing the coordinated change of cognition and psychology, which aims to comprehensively evaluate the coordination consistency between the user's cognitive ability development and psychological tendency change.

[0098] By multiplying the capability positive change intensity and the psychological orientation change intensity and taking the square root, a balanced index can be obtained, which can reflect whether the user's development in the cognitive and psychological aspects is synchronized and mutually promoted. A higher expression coordination factor indicates that the user's cognitive and psychological development is more coordinated, which helps more stable and effective learning.

[0099] The ratio of the state dispersion scalar and the expression coordination factor is calculated to obtain the attention density coefficient, which is used to comprehensively consider the stability of the user's psychological state and the cognitive psychological coordination to infer the current concentration degree or learning preparation state of the user. The coefficient provides a comprehensive index by ratio operation of the state dispersion scalar and the expression coordination factor. For example, when the state dispersion is low and the expression coordination factor is high, the attention density coefficient can be high, indicating that the user is in a more focused and more suitable learning state for accepting new information.

[0100] The emotional resonance degree weight, the state dispersion scalar and the attention density coefficient are multiplied to obtain an adjustment amount, which aims to dynamically calculate the specific value of adjusting the emotional resonance degree weight according to the current complex psychological and cognitive state of the user. The adjustment amount comprehensively considers the current emotional resonance degree weight, the volatility of the user's psychological state and the coordination of cognitive and psychological development, making the adjustment process more intelligent and refined.

[0101] The emotional resonance degree weight is added to the adjustment amount to obtain a new emotional resonance degree weight, and the capability fit degree weight and the adjusted emotional resonance degree weight are obtained by re-normalizing operation according to the new emotional resonance degree weight and the capability fit degree weight, ensuring the effectiveness and internal balance of the weight adjustment. After obtaining the new emotional resonance degree weight, in order to maintain the relative importance of the capability fit degree weight and the emotional resonance degree weight in the recommendation decision and ensure that their sum meets the preset proportion relationship, it is necessary to re-normalize the two. This can ensure that while adjusting the emotional resonance degree weight, the capability fit degree weight is also adjusted accordingly, thereby maintaining the internal logic consistency of the recommendation model.

[0102] Through the technical scheme, the capability fit degree weight and the emotional resonance degree weight can be adjusted more finely, overcoming the limitation of rough adjustment only according to the evolution rate scalar. Specifically, by clustering analysis on the time-series three-dimensional psychological portrait vector sequence and calculating the state dispersion scalar, the stability of the user's psychological state can be accurately evaluated, avoiding inappropriate weight adjustment when the user's psychological fluctuation is large. At the same time, by extracting the key components in the evolution direction vector, calculating the capability positive change intensity and the psychological orientation change intensity, and further generating the expression coordination factor, the application can quantify the coordination degree between cognitive and psychological changes.

[0103] The state dispersion scalar is combined with the expression of the coordination factor to generate an attention density coefficient, so that the weight adjustment not only considers the stability of the user's psychology, but also integrates the coordination of cognitive and psychological development, thereby generating a more adaptive adjustment amount. Finally, by accurately adjusting the emotional resonance weight and re-normalizing the ability matching weight, it ensures that the recommendation can more sensitively and accurately respond to the subtle dynamics of the user's cognitive ability development and emotional psychological demand changes, significantly improving the personalization and accuracy of picture book recommendations, enabling users to obtain picture book content that matches their current comprehensive state.

[0104] Further, the specific adjustment process of the ability matching weight and the emotional resonance weight also includes: within a statistical time window, the number of times the ability matching weight and the emotional resonance weight are adjusted together is counted to obtain an up-count; the up-count is input into a preset saturation growth function for mapping, the function outputs a dynamic upper limit value between 1 and a preset maximum allowed value; the dynamic upper limit value is used to update the upper limit of the range limiting function; the time when the ability matching weight and the emotional resonance weight were last adjusted is recorded and timing starts, and from this time, every time a statistical time window elapses, the dynamic upper limit value is multiplied by a preset decay factor until its value decays to 1.

[0105] In the present embodiment, Figure 3 The logic diagram for updating the boundary of the range limiting function provided by the present embodiment.

[0106] Within the statistical time window, the frequency of simultaneous adjustment of the ability matching weight and the emotional resonance weight is continuously monitored and recorded, thereby obtaining the up-count. The statistical time window can be a preset time period, such as the last hour, day or week, and its length can be configured according to the actual application scenario and the frequency of user behavior changes. The up-count reflects the response activity to user behavior changes, i.e., the frequency of considering the need to adjust user preferences within a certain period.

[0107] The up-count is input into a preset saturation growth function for mapping. The saturation growth function is a mathematical model whose output value increases with the increase of the input value, but the growth rate gradually slows down and eventually approaches an upper limit. Here, the up-count is input into the function to obtain a dynamic upper limit value. This dynamic upper limit value is used to limit the amplitude of subsequent weight adjustment, ensuring that the adjustment can respond to user changes and not grow unlimitedly. Between 1 and the preset maximum allowed value means that even if the adjustment is not frequent, there is at least a basic upper limit (1), and when the adjustment is frequent, the upper limit can be moderately relaxed, but it will not exceed a preset maximum value to prevent out of control.

[0108] The dynamic upper limit value is used to update the upper bound of the range clipping function. The range clipping function is used to ensure that the capability fit degree weight and the emotional resonance degree weight remain within a reasonable numerical interval after adjustment. By taking the calculated dynamic upper limit value as the upper bound of the function, the maximum value allowed by the weight can be adaptively adjusted according to the activity of user behavior changes. This means that when user behavior changes frequently and requires a larger adjustment of the weight, the upper limit can be appropriately increased; conversely, when the change is not frequent, the upper limit will be tightened, thereby enhancing stability and adaptability.

[0109] To prevent the dynamic upper limit value from remaining too high after the active period of user behavior ends, resulting in excessive adjustment in the subsequent stable period, the application also introduces a decay mechanism. Specifically, the last time the capability fit degree weight and the emotional resonance degree weight were adjusted is recorded and the time is started. From this time, every time a statistical time window passes, the dynamic upper limit value is multiplied by a preset decay factor until its value decays to 1. The decay factor is a positive number less than 1, and each decay will make the upper limit value closer to 1 until it reaches 1. This ensures that when user behavior tends to be stable, the upper limit of its weight adjustment will gradually return to a conservative state, avoiding unnecessary fluctuations and improving long-term stability.

[0110] The adjustment upper limit of the capability fit degree weight and the emotional resonance degree weight can be dynamically adjusted according to the activity of user behavior changes, thereby providing more adjustment space when user behavior is active and enhancing responsiveness and adaptability. At the same time, by introducing a decay mechanism, it is ensured that when user behavior tends to be stable, the adjustment upper limit can gradually return, avoiding excessive adjustment and instability. This dynamic and adaptive weight adjustment upper limit management mechanism enables the picture book live personalized book recommendation method to better balance the accuracy and stability of the recommendation, thereby providing personalized picture book recommendations that better match the user's current cognitive expression ability and psychological tendencies.

[0111] Further, the saturation growth function is specifically: wherein, is the number of times the capability fit degree weight and the emotional resonance degree weight are adjusted, is a preset maximum allowed value, is a preset growth rate constant, is a natural exponential function, and the function output value is the calculated dynamic upper limit value.

[0112] In this embodiment, the saturating growth function model is used to describe a growth process whose growth rate gradually slows down with the increase of the independent variable and eventually approaches an upper limit value. In this application, it is used to map the number of weight adjustment times x to a dynamic upper limit value f(x) that grows smoothly and nonlinearly with the increase of adjustment times, but does not grow indefinitely, but gradually approaches the preset maximum allowed value MaxLimit. This function form can simulate the diminishing marginal effect phenomenon in the learning or adaptation process, that is, with the accumulation of experience, the influence of each adjustment on the upper limit value gradually decreases.

[0113] x as an independent variable, represents the cumulative frequency or experience of adjusting the ability fit weight and emotional resonance weight. As input for the saturating growth function, it quantifies the degree of validation of the adjustment in a particular direction.

[0114] The more the adjustment times, the more validation of the effectiveness of the current adjustment strategy, allowing the dynamic upper limit value to have more growth space. MaxLimit is the upper limit value of the saturating growth function, representing the maximum value that the dynamic upper limit value can reach.

[0115] It is a preset constant that limits the unlimited growth of the dynamic upper limit value, ensuring a certain stability boundary in the adaptive adjustment process, preventing the upper limit value of weight adjustment from being too high, thereby avoiding overfitting or introducing instability.

[0116] k is a positive parameter in the saturating growth function, which determines the steepness or speed of the function growth. The larger the value of k, the faster the function reaches MaxLimit, that is, the more sensitive the dynamic upper limit value is to the adjustment times; the smaller the value of k, the slower the function grows, the more sluggish the response. It is used to fine-tune the rate of growth of the dynamic upper limit value with the increase of adjustment times to adapt to different application scenarios and requirements for response speed.

[0117] exp is the natural exponential function in mathematics, that is, e to the power of x. In this function, the term exp(−k∗x) ensures that as x increases, the value of this term rapidly decays from close to 1 and approaches 0, making the entire function f(x) gradually grow from close to 1 and approach MaxLimit.

[0118] It is the core mathematical component that realizes the saturating growth feature. f(x) is the output of the saturating growth function, which represents the dynamic upper limit of the range limiting function calculated according to the current weight adjustment times x. This value is dynamic, reflecting the accumulation of confidence in the weight adjustment strategy and providing a wider or more limited range for subsequent weight adjustment.

[0119] The dynamic upper limit value is calculated by a saturation growth function, which can ensure that the upper limit value of the range limiting function increases nonlinearly in a smooth and controlled manner as the number of adjustments of the capability fit weight and the emotional resonance weight increases. This growth pattern avoids linear or uncontrolled rapid expansion of the upper limit value, effectively preventing problems such as insufficient adjustment at the beginning or excessive adjustment at the end.

[0120] Specifically, when the number of adjustments is small, the upper limit value grows rapidly, allowing rapid adaptation to new adjustment strategies; when the number of adjustments increases, the growth rate of the upper limit value gradually slows down and approaches the preset maximum allowed value MaxLimit, which reflects the need for confidence accumulation and stability of the current adjustment strategy, avoiding instability due to excessive reliance on adjustment.

[0121] Meanwhile, by presetting the growth rate constant k, the sensitivity of the upper limit value growth can be flexibly adjusted, allowing the best balance between response speed and stability to be achieved according to the needs of the actual application scenario, thereby improving the robustness and adaptability of the entire picture book recommendation in long-term operation.

[0122] Figure 4 The structure diagram of the picture book live personalized book recommendation system based on artificial intelligence provided by the embodiment of the application, the picture book live personalized book recommendation system based on artificial intelligence provided by the embodiment of the application, comprising: a data acquisition module: for acquiring interactive data of a target user, the interactive data including text data, voice feature data and expression feature data; a vector construction module: for generating a three-dimensional capability portrait vector sequence representing the current cognitive expression ability of the target user and a three-dimensional psychological portrait vector sequence representing the current psychological tendency of the target user based on the interactive data; a time series analysis module: for performing time series analysis on the three-dimensional capability portrait vector sequence and the three-dimensional psychological portrait vector sequence, and calculating an evolution direction vector and an evolution rate scalar; a coordination coefficient acquisition module: for judging whether the evolution rate scalar is within the adaptive threshold range when all components of the evolution direction vector are positive, and if so, calculating the absolute difference between the evolution rate scalar and the midpoint of the adaptive threshold and updating the preset coordination coefficient in combination with the boundary value comparison result of the range limiting function; a judgment module: for if the judgment result is false and the evolution rate scalar is lower than the lower limit of the adaptive threshold range, then no adjustment is made; a weight adjustment module: for if the judgment result is false and the evolution rate scalar exceeds the upper limit of the adaptive threshold range, then the capability fit weight and the emotional resonance weight are adjusted; a statistical module: for recording the adjustment count of the capability fit weight and the emotional resonance weight, and updating the boundary value of the range limiting function accordingly; a total score calculation module: for calculating a matching decision total score by matching a decision formula according to the updated preset coordination coefficient, the adjusted capability fit weight and the emotional resonance weight; an output module: for selecting at least one target picture book according to the matching decision total score and outputting.

[0123] Those skilled in the art will appreciate that embodiments of the present application can be readily used as software, hardware, or a combination of software and hardware. In a software embodiment, the methods can be tangibly embodied in a machine-readable storage medium having stored thereon instructions that can be used to program a computer to perform any of the methods. The software implementation can be for example, in a modulated data signal such as a carrier wave or other transport mechanism, or a computer readable storage medium.

[0124] The present application is described in reference to the drawings using a flowchart and / or a block diagram of the method, apparatus (system) and computer program product according to embodiments of the application. It will be understood that each block of the flowchart and / or block diagram, and combinations of blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 means for performing each of the functions specified in the flowchart and / or block diagram block or blocks.

[0125] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 means for performing each of the functions specified in the flowchart and / or block diagram block or blocks.

[0126] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 means for performing each of the functions specified in the flowchart and / or block diagram block or blocks.

[0127] While preferred embodiments of the application have been described, modifications and alterations thereto will occur to those skilled in the art upon reading the preceding description. In particular, it will be apparent to those skilled in the art that parts can be added to, or substituted for, parts of the described embodiment. It is intended that the application susceptible to alterations and modifications of the preferred embodiments. Therefore, the following claims are intended to cover all such alterations and modifications that fall within the scope of the present application.

[0128] Obviously, a person skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these modifications and variations.

Claims

1. A personalized book recommendation method for live-streamed picture book reading based on artificial intelligence, characterized in that: Includes the following steps: Acquire interaction data from the target user, which includes text data, voice feature data, and facial expression feature data; Based on interactive data, a three-dimensional ability profile vector sequence representing the target user's current cognitive expression ability and a three-dimensional psychological profile vector sequence representing the target user's current psychological tendency are generated. Temporal analysis was performed on the three-dimensional ability profile vector sequence and the three-dimensional psychological profile vector sequence to calculate the evolution direction vector and the evolution rate scalar. When all components of the evolution direction vector are positive, determine whether the evolution rate scalar is within the adaptive threshold range. If so, calculate the absolute difference between the evolution rate scalar and the midpoint of the adaptive threshold and update the preset synergy coefficient by combining the comparison result of the boundary value of the range limiting function. If not, and the evolution rate scalar is below the lower limit of the adaptive threshold range, then no adjustment is made; If not, and the evolution rate scalar exceeds the upper limit of the adaptive threshold range, then adjust the ability fit weight and emotional resonance weight. Record the adjustment counts of the ability compatibility weight and emotional resonance weight, and update the boundary value of the range limiting function accordingly; Based on the updated preset synergy coefficient, the adjusted ability compatibility weight, and the emotional resonance weight, the total matching decision score is calculated using the matching decision formula. Select at least one target picture book based on the total matching decision score and output it; The specific process for obtaining the target picture book is as follows: For candidate picture books in the picture book database, obtain the pre-stored difficulty feature vector and theme feature vector corresponding to the candidate picture books; The difficulty feature vector is composed of preset difficulty level values ​​for candidate picture books in three dimensions: language ability, emotional expression, and cultural cognition; The theme feature vector is composed of preset theme intensity values ​​of candidate picture books in three dimensions: emotional state, value orientation, and role projection; Calculate the absolute value of the difference between the cosine similarity and the modulus length between the 3D ability profile vector and the difficulty feature vector, and then perform a weighted average of the absolute values ​​of the difference between the cosine similarity and the modulus length according to the preset first weight to generate the fit coefficient. The dot product between the 3D psychological profile vector and the topic feature vector is calculated as the psychological topic fit. The vector components representing emotional state in the 3D psychological profile vector and the topic feature vector are extracted respectively, and their standardized absolute difference is calculated and recorded as the emotional matching degree. The psychological theme fit and emotional match are weighted and averaged according to a preset second weight to generate a resonance coefficient. The first weighted component is obtained by multiplying the compatibility coefficient by the ability compatibility weight, the second weighted component is obtained by multiplying the resonance coefficient by the emotional resonance weight, and the synergy gain component is obtained by multiplying the product of the compatibility coefficient and the resonance coefficient by the preset synergy coefficient. The first weighted component, the second weighted component and the synergy gain component are all added together to obtain the total matching decision score. Candidate picture books whose total matching decision score exceeds the recommendation threshold are selected as target picture books.

2. The method for personalized book recommendation in live picture book streaming based on artificial intelligence according to claim 1, characterized in that, The specific generation process of the three-dimensional capability profile vector is as follows: Extract narrative structure features and sentiment lexical features from text data; Extracting syntactic complexity features from speech feature data; Narrative structure features, emotional vocabulary features, and syntactic complexity features are input into a preset feature dimension mapping model. The feature dimension mapping model outputs a score representing the target user on three preset orthogonal ability dimensions, including language ability dimension, emotional expression dimension, and cultural cognition dimension. The scores on the three orthogonal ability dimensions are normalized, and a three-dimensional vector is constructed in the preset orthogonal three-dimensional space to serve as the three-dimensional ability profile vector.

3. The method for personalized book recommendation in live picture book streaming based on artificial intelligence according to claim 1, characterized in that, The specific generation process of the three-dimensional psychological profile vector is as follows: The tone and speed parameter sequences in the speech feature data are fused with the facial action unit parameter sequences in the expression feature data, and then input into a preset multimodal emotion recognition model to output the current emotion state category of the target user. Semantic analysis and keyword extraction are performed on the text data, and it is matched with a preset value proposition thesaurus to identify the core value orientation category contained in the target user's expression; The emotional state category, the core value orientation category, and the topic tag data associated with the live broadcast interaction are input into a preset role classification model to output the role projection type of the target user in the current interaction. The emotional state category, the core value orientation category, and the role projection type are mapped to initial values ​​for the three dimensions of emotional state intensity, value orientation intensity, and role projection intensity, respectively, according to a preset quantitative mapping table. The initial values ​​of the three dimensions are normalized, and a three-dimensional vector is constructed in a preset orthogonal three-dimensional space, which serves as the three-dimensional psychological profile vector.

4. The method for personalized book recommendation in live picture book streaming based on artificial intelligence according to claim 1, characterized in that, The specific update process for the preset coordination coefficient is as follows: Within the evolution time window, three-dimensional capability profile vectors and three-dimensional psychological profile vectors are acquired at preset time intervals to obtain temporal three-dimensional capability profile vector sequences and temporal three-dimensional psychological profile vector sequences. Perform first-order difference operations on the temporal three-dimensional ability profile vector sequence and the temporal three-dimensional psychological profile vector sequence to obtain the ability change direction sub-vector and the psychological change direction sub-vector respectively. The weighted average vector of the sub-vectors of ability change direction and psychological change direction over the evolution time window is calculated to obtain the evolution direction vector; Calculate the Euclidean distance between the three-dimensional ability profile vectors at adjacent preset time intervals and the Euclidean distance between the three-dimensional mental profile vectors at adjacent preset time intervals in the temporal three-dimensional ability profile vector sequence and the temporal three-dimensional mental profile vector sequence respectively to obtain two rate subsequences; normalize the values ​​of the two rate subsequences and add them together to obtain the evolution rate scalar. When all the evolution direction vectors are positive and the evolution rate scalar is within the adaptive threshold range, it is determined to be a state adaptation adjustment mode, and the preset cooperative coefficient is updated accordingly.

5. The method for personalized book recommendation in live picture book streaming based on artificial intelligence according to claim 4, characterized in that, The specific update process of the preset coordination coefficient also includes: The basic adjustment factor is obtained by taking the angle between the calculated evolution direction vector and the preset positive reference vector as input and processing it through the arctangent function. The absolute difference between the evolution rate scalar and the midpoint of the adaptive threshold is calculated, and this difference is processed through a negative exponential function to obtain the rate stability decay factor. The product of the basic adjustment factor and the rate stability decay factor is used as the increment of the synergy coefficient in this calculation. After obtaining the increment, the preset synergy coefficient is added to the increment to obtain the preliminary update value; If the initial update value is lower than the lower limit of the range limiting function, the range limiting function outputs the lower limit value as the output result; if the initial update value is higher than the upper limit of the range limiting function, the range limiting function outputs the upper limit value as the output result. If the initial update value is between the upper and lower limits of the range limiting function, then the initial update value will be used as the output result. Update the preset synergy coefficient with the output results.

6. The method for personalized book recommendation in live picture book streaming based on artificial intelligence according to claim 1, characterized in that, The specific adjustment process for the competence compatibility weight and emotional resonance weight is as follows: By performing cluster analysis on the temporal three-dimensional psychological profile vector sequence, the average Euclidean distance from all vector points to their cluster centroids is calculated. After normalizing the average Euclidean distance, it is used as a scalar of state discreteness. Extract the two components of the evolution direction vector corresponding to the language ability and emotional expression dimensions of the three-dimensional ability profile vector, calculate the root mean square of the sum of the squares of the two components, and obtain the intensity of positive change in ability. Extract the two components of the evolution direction vector corresponding to the value orientation and role projection dimensions of the three-dimensional psychological profile vector, calculate the sum of the absolute values ​​of the two components, and obtain the intensity of the psychological orientation change. Multiply the intensity of positive change in ability by the intensity of change in psychological orientation, and take the square root of the product to obtain the expression coordination factor that represents the coordinating change between cognition and psychology. The ratio of the state discreteness scalar to the expression coordination factor is calculated to obtain the attention density coefficient; The adjustment amount is obtained by multiplying the emotional resonance weight, the state dispersion scalar, and the attention density coefficient; Add the adjustment amount to the emotional resonance weight to obtain the new emotional resonance weight; The normalization operation is performed again based on the new emotional resonance weight and ability compatibility weight to obtain the adjusted ability compatibility weight and the adjusted emotional resonance weight.

7. The method for personalized book recommendation in live picture book streaming based on artificial intelligence according to claim 6, characterized in that, The specific adjustment process for the competency compatibility weight and emotional resonance weight also includes: Within the statistical time window, the number of times the statistical ability compatibility weight and the emotional resonance weight are adjusted together is the upward adjustment count. The upward count is mapped to a preset saturation growth function, which outputs a dynamic upper limit value between 1 and the preset maximum allowable value. Update the upper bound of the range limiting function using a dynamic upper limit value; Record the moment when the ability compatibility weight and emotional resonance weight were last adjusted and start timing. From that moment on, for each statistical time window, the dynamic upper limit value is multiplied by a preset decay factor until its value decays to 1.

8. The method for personalized book recommendation in live picture book streaming based on artificial intelligence according to claim 7, characterized in that, The saturated growth function is specifically: ,in, The number of times the weights for ability compatibility and emotional resonance were adjusted. This is the preset maximum allowed value. The preset growth rate constant, This is the natural exponential function, and its output value is... This is the calculated dynamic upper limit value.

9. A personalized book recommendation system for live-streamed picture book reading based on artificial intelligence, characterized in that: include: Data acquisition module: used to acquire the interaction data of the target user, which includes text data, voice feature data and facial expression feature data; Vector Construction Module: Used to generate a three-dimensional ability profile vector sequence representing the target user's current cognitive expression ability and a three-dimensional psychological profile vector sequence representing the target user's current psychological tendency based on interactive data; The time series analysis module is used to perform time series analysis on the three-dimensional ability profile vector sequence and the three-dimensional psychological profile vector sequence, and calculate the evolution direction vector and the evolution rate scalar. Synergy coefficient acquisition module: When all components of the evolution direction vector are positive, it determines whether the evolution rate scalar is within the adaptive threshold range. If so, it calculates the absolute difference between the evolution rate scalar and the midpoint of the adaptive threshold and updates the preset synergy coefficient by combining the comparison result of the boundary value of the range limiting function. Judgment module: If the judgment result is negative and the evolution rate scalar is lower than the lower limit of the adaptive threshold range, then no adjustment will be made; Weight adjustment module: If not, and the evolution rate scalar exceeds the upper limit of the adaptive threshold range, then adjust the ability fit weight and emotional resonance weight. Statistics module: Used to record the adjustment counts of ability compatibility weight and emotional resonance weight, and update the boundary values ​​of the range limiting function accordingly; Total score calculation module: It is used to calculate the total score of matching decision based on the updated preset synergy coefficient, the adjusted ability compatibility weight and the emotional resonance weight, through the matching decision formula; Output module: Used to select at least one target picture book based on the total matching decision score and output it; The specific process for obtaining the target picture book is as follows: For candidate picture books in the picture book database, obtain the pre-stored difficulty feature vector and theme feature vector corresponding to the candidate picture books; The difficulty feature vector is composed of preset difficulty level values ​​for candidate picture books in three dimensions: language ability, emotional expression, and cultural cognition; The theme feature vector is composed of preset theme intensity values ​​of candidate picture books in three dimensions: emotional state, value orientation, and role projection; Calculate the absolute value of the difference between the cosine similarity and the modulus length between the 3D ability profile vector and the difficulty feature vector, and then perform a weighted average of the absolute values ​​of the difference between the cosine similarity and the modulus length according to the preset first weight to generate the fit coefficient. The dot product between the 3D psychological profile vector and the topic feature vector is calculated as the psychological topic fit. The vector components representing emotional state in the 3D psychological profile vector and the topic feature vector are extracted respectively, and their standardized absolute difference is calculated and recorded as the emotional matching degree. The psychological theme fit and emotional match are weighted and averaged according to a preset second weight to generate a resonance coefficient. The first weighted component is obtained by multiplying the compatibility coefficient by the ability compatibility weight, the second weighted component is obtained by multiplying the resonance coefficient by the emotional resonance weight, and the synergy gain component is obtained by multiplying the product of the compatibility coefficient and the resonance coefficient by the preset synergy coefficient. The first weighted component, the second weighted component and the synergy gain component are all added together to obtain the total matching decision score. Candidate picture books whose total matching decision score exceeds the recommendation threshold are selected as target picture books.

Images