Fiological and creative element data analysis method based on natural language processing technology

Abstract

The invention discloses a natural language processing technology-based cultural and creative element data analysis method, which comprises the following steps of: based on AI intelligent plush doll interaction data, generating a cultural and creative structure embedding degree index through named entity recognition and semantic network topology analysis; an impulse response function is constructed based on the cultural creative structure embedding degree, and a cultural creative memory cycle index is calculated in combination with the path length and information backflow; fusing the multi-modal features and the ideal innovation region, and outputting a perception novelty score; screening high-potential-energy elements to construct a semantic sub-graph, simulating a thermodynamic process, and identifying a text-creative fusion hotspot chain; and decoding the hotspot chain into a design semantic tag, matching a material database to generate an appearance scheme, and converting the appearance scheme into an interaction template, thereby realizing a customized design closed loop from a user dialogue to a doll form and behavior. Through data driving and cross-modal analysis, implicit culture preferences in user dialogues are dominated, and creativity of the appearance of the AI intelligent plush doll and humanization of interaction are improved.

Description

Technical Field

[0001] This invention relates to the field of culture, and in particular to a method for analyzing cultural and creative element data based on natural language processing technology. Background Technology

[0002] Against the backdrop of the rapid development of the cultural and creative industries, users' demand for personalized, emotional, and culturally resonant intelligent products is growing. Traditional cultural and creative products are mostly static and mass-produced, lacking deep interaction with individual users' cultural memories and emotional experiences. Existing AI dolls, in scenarios involving intangible cultural heritage transmission, regional cultural experiences, and literary IP interaction, often merely serve as carriers of cultural symbols, mechanically reusing pre-set content. They fail to dynamically capture users' cultural preferences, emotional connections, and cognitive evolution revealed in dialogue, resulting in superficial cultural expression lacking the ability to generate deep meaning and personalized interpretation. Current technology lacks an effective path to systematically extract cultural elements from user language, assess their cognitive potential, and transform them into multimodal expressions. Based on this, this invention proposes a data analysis method for cultural and creative elements based on natural language processing technology. Summary of the Invention

[0003] This invention provides a method for analyzing cultural and creative element data based on natural language processing technology, characterized by comprising: S10. Based on the natural language data collected by AI smart plush toys in cultural and creative theme interaction, cultural and creative elements are extracted through named entity recognition, and combined with their frequency, betweenness centrality and clustering coefficient in the semantic network, a cultural and creative structure embedding degree index is generated. S20. Based on the embedding degree of cultural and creative structure, an impulse response function is constructed to characterize the cognitive dynamics of cultural and creative themes. Combined with weighted propagation path length and information feedback analysis, a cultural and creative memory cycle index is generated. S30. Tensor fusion is performed between the cultural and creative memory cycle index and the multimodal expression of each cultural and creative element to construct a three-dimensional feature space. Manifold projection is then performed in this space to calculate the probability density overlap between the behavior distribution of each element and the ideal innovation area, and output the cultural and creative perception novelty score. S40. Based on the novelty score of cultural and creative perception, select high-potential cultural and creative elements, construct a co-occurrence semantic subgraph and simulate a non-equilibrium thermodynamic process, assign initial cultural temperature to nodes, iteratively calculate entropy productivity and free energy gradient, identify the path with the steepest energy gradient, and extract cultural integration hotspot chains. S50 decodes the hot topic chain of cultural and creative integration into design semantic tags, and generates a doll appearance scheme with a consistent style by matching the material aesthetic database through Wasserstein distance; at the same time, it transforms its element sequence and emotional dynamics into interactive templates to guide the customized design of the doll's form and interactive content.

[0004] The above-described method for analyzing cultural and creative element data based on natural language processing technology involves extracting cultural and creative elements from natural language data collected using AI-powered plush toys during interactive activities on cultural and creative themes. This data is then combined with named entity recognition to determine the frequency, betweenness centrality, and clustering coefficient of these elements in the semantic network, generating a cultural and creative structure embedding degree index. Specifically, this method comprises the following sub-steps: By using named entity recognition and cultural and creative ontology knowledge graph, cultural and creative elements are accurately extracted and classified from interactive doll texts, ensuring the semantic accuracy and consistency of the recognition results. Based on dependency parsing, the adjectives, verbs and metaphors in the context of cultural and creative elements are weighted according to their weights, and a syntactic distance decay factor is introduced to calculate the context expansion dimension score. A weighted semantic network is constructed based on the co-occurrence relationship of cultural and creative elements. The embedding degree of cultural and creative structures is generated by combining topological indicators such as betweenness centrality, clustering coefficient and feature vector centrality with occurrence frequency and context expansion score.

[0005] The above-described method for analyzing cultural and creative element data based on natural language processing technology accurately extracts and classifies cultural and creative elements from interactive doll texts through named entity recognition and a cultural and creative ontology knowledge graph, ensuring the semantic accuracy and consistency of the recognition results. Specifically, it consists of the following sub-steps: Construct a knowledge graph of cultural and creative ontology to form a standardized semantic reference system for cultural and creative products; Fine-grained entity recognition of user dialogue text is performed using a hybrid model of bidirectional encoder-representation converter-bidirectional long short-term memory network-conditional random field. By using knowledge graph embedding technology, entities and relationships in the knowledge graph are mapped to a low-dimensional vector space. The semantic distance between the candidate entities output by the model and their potential matching nodes in the graph is calculated, and only high-confidence matching results with a distance less than a preset threshold are retained.

[0006] The above-described data analysis method for cultural and creative elements based on natural language processing technology involves constructing an impulse response function based on the embedding degree of cultural and creative structures to characterize the cognitive dynamics of cultural and creative themes, and generating a cultural and creative memory cycle index by combining weighted propagation path length and information feedback analysis. Specifically, it consists of the following sub-steps: Using the embedding degree of cultural and creative structures as the initial weight, a dynamic cognitive activation function formula based on the superposition of impulse responses is constructed. The exponential decay kernel is used to simulate memory decay, and the periodic oscillation term is used to capture the echo pattern, so as to comprehensively reflect the evolution process of the activation intensity of cultural and creative elements after they are mentioned in user dialogue. By introducing the geometric mean of the embedding degree of cultural and creative structures as the denominator, the weighted length of the cultural and creative dissemination path in the dialogue flow is calculated. By identifying information feedback events and assessing cognitive reconstruction by combining semantic similarity and sentence variation, the system integrates feedback quality, dissemination efficiency, and memory volatility to quantify the continuous activation capacity of cultural and creative themes using a formula, thereby generating a cultural and creative memory cycle index.

[0007] The above-described data analysis method for cultural and creative elements based on natural language processing technology involves tensor fusion of the cultural and creative memory cycle index with the multimodal expressions of each element to construct a three-dimensional feature space. Manifold projection is then performed within this space to calculate the probability density overlap between the behavioral distribution of each element and the ideal innovation region, outputting a cultural and creative perceived novelty score. Specifically, this method comprises the following sub-steps: Based on the cultural and creative memory cycle index, three types of multimodal behavioral data were collected: voice emotion fluctuation, cultural and creative emotion jump, and interactive action activity, to construct a three-dimensional cross-modal behavioral tensor. The three-dimensional cross-modal behavior tensor is reduced to a two-dimensional manifold by isometric mapping, the response of cultural and creative elements is represented as a continuous trajectory, and its first principal curvature and Gaussian curvature are calculated based on smooth fitting and differential geometric analysis. The ideal innovation region is defined as a Gaussian distribution. The overlap between the behavior distribution of each cultural and creative element in the manifold space and this region is calculated. The geometric dynamic characteristics of its first principal curvature and Gaussian curvature are integrated to calculate the perceived novelty score of cultural and creative products.

[0008] The above-described method for analyzing cultural and creative element data based on natural language processing technology involves: screening high-potential cultural and creative elements based on perceived novelty scores; constructing a co-occurrence semantic subgraph and simulating a non-equilibrium thermodynamic process; assigning initial cultural temperatures to nodes; iteratively calculating entropy productivity and free energy gradients; identifying the path with the steepest energy gradient; and extracting cultural integration hotspot chains. Specifically, it comprises the following sub-steps: High-potential elements are selected based on the novelty score of cultural and creative perception to construct a high-potential cultural and creative subgraph, and the edge weights between nodes are determined by integrating co-occurrence frequency and semantic vector cosine similarity. Based on the embedding degree of cultural and creative structures, the node temperature is initialized, the heat conduction process in the weighted network is simulated, and the cognitive roles are identified by iteratively calculating the temperature evolution, entropy production rate and free energy decrease rate, revealing the dynamic evolution path of the influence of cultural and creative industries. The calculation of path strength is based on a weighted product of temperature gradient, average entropy yield, maximum local instability, and total entropy change of the path. It focuses on identifying paths with the steepest energy gradient and the most intense information flow, and captures hot chains of cultural and creative integration.

[0009] The above-described method for analyzing cultural and creative element data based on natural language processing technology involves decoding the cultural and creative integration hotspot chain into design semantic tags, matching the material aesthetics database using Wasserstein distance, and generating a coherent doll appearance scheme. Simultaneously, it transforms the element sequence and emotional dynamics into interactive templates to guide the customized design of the doll's form and interactive content. Specifically, it consists of the following sub-steps: By normalizing the temperature gradient and entropy change accumulation value of the cultural and creative integration hotspot chain, we construct the potential energy transition intensity and information evolution density index, and map them into design semantic tags based on the sensible engineering dictionary. Based on design semantic tags, features are matched in the material aesthetics database through optimal transmission theory, the composite semantic decomposition sub-targets are decomposed and the Cesarstein distance is minimized to generate an AI doll appearance scheme with a unified style and progressive meaning. Based on the structure and dynamic characteristics of the cultural and creative integration hotspot chain, an interactive template containing personality, dialogue logic and feedback strategies is generated to systematically guide the interactive design of AI dolls and achieve accurate expression of emotions and culture.

[0010] This invention also provides a data analysis system for cultural and creative elements based on natural language processing technology, comprising: Extraction and Embedding Degree Calculation Module: Based on the natural language data collected by AI smart plush toys in cultural and creative theme interactions, cultural and creative elements are extracted through named entity recognition, and combined with their frequency, betweenness centrality and clustering coefficient in the semantic network, a cultural and creative structure embedding degree index is generated. The memory cycle index module: Based on the embedding degree of cultural and creative structures, an impulse response function is constructed to characterize the cognitive dynamics of cultural and creative themes. Combined with weighted propagation path length and information feedback analysis, a cultural and creative memory cycle index is generated. Fusion and Novelty Calculation Module: Tensor fusion is performed between the cultural and creative memory cycle index and the multimodal expression of each cultural and creative element to construct a three-dimensional feature space. In this space, manifold projection is performed to calculate the probability density overlap between the behavior distribution of each element and the ideal innovation area, and output the cultural and creative perceived novelty score. Screening and Hotspot Chain Extraction Module: Based on the novelty score of cultural and creative perception, high-potential cultural and creative elements are screened, a co-occurrence semantic subgraph is constructed and a non-equilibrium thermodynamic process is simulated, nodes are given an initial cultural temperature, entropy yield and free energy gradient are iteratively calculated, the path with the steepest energy gradient is identified, and cultural integration hotspot chains are extracted. Decoding and Solution Generation Module: Decodes the hot topic chain of cultural and creative integration into design semantic tags, and generates a doll appearance scheme with a coherent style by matching the material aesthetic database through Wasserstein distance; at the same time, it transforms its element sequence and emotional dynamics into interactive templates to guide the customized design of the doll's form and interactive content.

[0011] The beneficial effects achieved by this invention are as follows: This invention makes implicit cultural preferences in user dialogue explicit through data-driven and cross-modal analysis, thereby improving the creativity of the appearance of AI intelligent plush toys and the humanization of their interaction. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0013] Figure 1 This is a flowchart of a data analysis method for cultural and creative elements based on natural language processing technology, provided in Embodiment 1 of this application.

[0014] Figure 2 This is a schematic diagram of a data analysis system for cultural and creative elements based on natural language processing technology, provided in Embodiment 2 of this application. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] Example 1

[0017] like Figure 1 As shown, Embodiment 1 of this application provides a method for analyzing cultural and creative element data based on natural language processing technology, including: S10. Based on the natural language data collected from AI-powered smart plush toys in cultural and creative themed interactions, cultural and creative elements are extracted through named entity recognition. Combining their frequency, betweenness centrality, and clustering coefficient in the semantic network, a cultural and creative structure embedding degree index is generated.

[0018] S11. By using named entity recognition and cultural and creative ontology knowledge graph, cultural and creative elements are accurately extracted and classified from interactive doll texts, ensuring the semantic accuracy and consistency of the recognition results.

[0019] To accurately extract core cultural and creative elements from the natural language interactions between AI-powered smart plush toys and users, a fusion method combining named entity recognition and a cultural and creative ontology knowledge graph is employed to perform structured identification and classification of these elements. First, an authoritative cultural and creative ontology knowledge graph is constructed, comprising three main categories of nodes: traditional symbols, regional imagery, and historical narratives. Each node is labeled with multilingual names, semantic descriptions, hierarchical relationships, and synonym variants, forming a standardized semantic reference system for cultural and creative products.

[0020] Based on this, a hybrid model of bidirectional encoder representation converter-bidirectional long short-term memory network-conditional random field is used to perform fine-grained entity recognition on user dialogue text: the bidirectional encoder representation converter provides context-sensitive semantic vectors to effectively understand non-standard expressions; the bidirectional long short-term memory network captures long-distance dependencies to accurately define entity boundaries; and the conditional random field layer ensures the logical consistency of the label sequence, thereby achieving high-precision recognition of colloquial, vague or metaphorical descriptions.

[0021] Subsequently, knowledge graph embedding technology is used to map entities and relationships in the knowledge graph to a low-dimensional vector space. The semantic distance between the candidate entities output by the model and their potential matching nodes in the graph is calculated. Only high-confidence matching results with a distance less than a preset threshold are retained, thereby resolving ambiguity issues such as polysemy and synonymy, and filtering out misidentifications due to semantic deviation. The entire process is supplemented by hierarchical consistency verification, contextual semantic verification, and confidence scoring mechanisms to ensure that the finally extracted cultural and creative elements are not only correctly identified in form, but also consistent with the cultural and creative ontology in terms of deep semantics.

[0022] S12. Based on dependency parsing, adjectives, verbs and metaphorical expressions in the context of cultural and creative elements are weighted according to their weights, and a syntactic distance decay factor is introduced to calculate the context expansion dimension score.

[0023] For each confirmed cultural and creative element, its semantic expansion ability in the dialogue context is analyzed using the context expansion dimension score to measure the depth of association and emotional richness evoked by the element in user expression. Specifically, a context window of ±3 sentences is formed by expanding three sentences forward and three sentences backward from the sentence containing the target element, and all language components directly or indirectly related to it are extracted within this window. Dependency parsing is used to identify words that modify or dominate the cultural and creative element, focusing on capturing three types of semantic expansion components: adjectives modifying its attributes, verbs describing its behavior or state, and metaphorical expressions formed through rhetorical devices such as metaphor and metonymy. To reflect the different contributions of different components to semantic expansion, adjectives, verbs, and metaphorical expressions are assigned differentiated weights. An exponential decay factor based on syntactic path length is introduced. Where d is the shortest path length between the word and the target element in the dependency syntax tree. The attenuation constant is used to simulate the natural law that semantic influence weakens with increasing distance. The weighted values ​​of all related words are multiplied by their corresponding attenuation coefficients and then summed to obtain the contextual expansion dimension score of the cultural and creative element.

[0024] S13. Construct a weighted semantic network based on the co-occurrence relationship of cultural and creative elements. Generate the embedding degree of cultural and creative structure by combining topological indicators such as betweenness centrality, clustering coefficient and feature vector centrality with occurrence frequency and context expansion score.

[0025] All identified and aligned cultural and creative elements are treated as nodes in a semantic network. A weighted co-occurrence semantic network is constructed, using the frequency with which they appear in the same contextual window during user dialogues as edge weights. This network characterizes the strength of associations and structural relationships between different cultural and creative elements. Based on this network, three key topological indicators are calculated for each node: betweenness centrality (calculated by calculating the proportion of shortest paths between all node pairs passing through the node, representing its role as a hub in information dissemination); local clustering coefficient (measured by the ratio of the actual number of connections between the node's neighbors to the theoretically maximum possible number of connections, reflecting the density of its neighborhood and its tendency to cluster locally); and eigenvector centrality (calculated by solving the principal eigenvectors of the network adjacency matrix, ensuring that each node's score is proportional to the scores of its neighbors, thus reflecting its influence and importance in the global network).

[0026] These network topology indicators, along with the frequency of occurrence of cultural and creative elements in the corpus and the scores of contextual expansion dimensions, are combined through standardization and weighting to generate a cultural and creative structure embedding index. This comprehensively quantifies the organizational core role of each cultural and creative element in the user's cognitive structure, i.e., whether it is a high-frequency, semantically rich, and key cultural and creative cognitive anchor in the network.

[0027] S20. Based on the embedding degree of cultural and creative structures, an impulse response function is constructed to characterize the cognitive dynamics of cultural and creative themes. Combined with weighted propagation path length and information feedback analysis, a cultural and creative memory cycle index is generated.

[0028] S21. Using the embedding degree of cultural and creative structures as the initial weight, a dynamic cognitive activation function formula based on the superposition of impulse responses is constructed. The exponential decay kernel is used to simulate memory decay, and the periodic oscillation term is used to capture the echo pattern, so as to comprehensively reflect the evolution process of the activation intensity of cultural and creative elements after they are mentioned in user dialogue.

[0029] Using the embedding degree of cultural and creative structures as the initial weight of each cultural and creative element node, the mention of a certain element in each round of dialogue between the user and the AI ​​doll is regarded as a pulse-like diffusion event of cultural and creative information. A dynamic response function for cultural and creative dissemination is constructed. This response function forms a continuous and dynamically changing cognitive activation curve by superimposing the pulse responses of all historical mention events. The specific formula is as follows ,in, Representing cultural and creative elements The intensity of comprehensive cognitive activation at any time t. (Summation range) This represents the set of all historical points in time when the element is mentioned, each... This corresponds to a single pulse input. The embedding degree of the element in the cultural and creative structure serves as the basis for the amplitude of each pulse, reflecting its fundamental importance in the user's cognitive structure: elements that occupy a core position in the semantic network, appear frequently, and have rich context will trigger a stronger propagation effect every time they are mentioned. It is an exponentially decaying kernel, used to simulate the natural weakening of memory over time. The decay rate controls the rate at which the activation intensity decreases. The larger the value, the faster the topic is forgotten; conversely, a smaller value indicates a more lasting impact. It is a periodic oscillation term used to capture the periodic echo phenomenon of cultural themes in users' cognition. Under the guidance of specific topics, emotional resonance, or situational triggers, users may refocus on the same theme after several rounds of dialogue, forming a fluctuating recurrence of memory. The oscillation frequency determines the speed of the echo. The initial phase represents the semantic starting state at the time of the first activation.

[0030] S22. By introducing the geometric mean of the embedding degree of cultural and creative structures as the denominator, calculate the weighted length of the cultural and creative dissemination path in the dialogue flow.

[0031] To reveal the diffusion trajectory and structural characteristics of cultural and creative themes in multi-turn dialogues, the dialogue sequence between users and AI dolls is modeled as a dynamically evolving dialogue flow graph when analyzing the propagation path of cultural and creative themes. Each node represents a dialogue turn or semantic unit containing a specific cultural and creative element, while edges represent the semantic continuity between adjacent or related dialogues. To measure the efficiency and breadth of the propagation of a particular cultural and creative theme in multi-turn interactions, the weighted length of its propagation path needs to be calculated. For a cultural and creative propagation path from the starting point to the ending point... Its weighted propagation path length is defined as , represents the total weighted length of path P; a smaller value indicates more efficient information propagation. n represents the total number of nodes on path P; the summation term iterates through each pair of consecutive nodes in the path. and That is, each transition from the current dialogue state to the next state. It is the semantic distance between two adjacent nodes, which can be calculated by the cosine distance of sentence vectors. It reflects the degree of topic jump between two mentions. The larger the distance, the more drastic the semantic shift. and These are nodes and The score for the embedding degree of the corresponding cultural elements into the cultural and creative structure. (Denominator) As a weighting factor, it reflects that there is less resistance to information transmission between nodes with high embedding degree: when the cultural and creative elements involved in two connected nodes are both core content in the user's cognition, their transmission potential is higher and the transmission cost per unit distance of the path is lower, so the overall weighted length is compressed.

[0032] S23. By identifying information return events, combining semantic similarity and sentence variation to assess cognitive reconstruction, and comprehensively considering return quality, dissemination efficiency and memory volatility, the continuous activation capacity of cultural and creative themes is quantified by formula to generate a cultural and creative memory cycle index.

[0033] In constructing the cognitive activation curve of cultural and creative elements Subsequently, by detecting its secondary peaks after initial activation and combining this with text sentence variation analysis, a series of information backflow events were identified, forming an information backflow event set. Based on this set, a cultural and creative memory cycle index is calculated to quantify the ability of a cultural and creative theme to be continuously reconstructed and cyclically activated in users' cognition. The specific formula is as follows: , The Cultural and Creative Memory Cycle Index represents the intensity of how frequently a cultural and creative element i is noticed, recalled, and reconstructed by users during the interaction process. The higher the value, the more stable the memory loop of the theme has been formed in cognition. is the set of reflow events for element i, that is, in the conversation flow, the set of all events in which the user mentions the element again in a different way or semantically reconstructs it after the user mentions it for the first time. For set The total number of backflow events is used to normalize the average backflow quality. This represents the j-th reflow event belonging to element i. , which is the semantic similarity, measuring the semantic closeness between the j-th reflow expression and the original mention. Sentence variability measures the degree of difference between the reflow expression and the original mention in terms of syntactic structure. A higher value indicates that the user has made more creative reconstructions. The reflow time difference is the time interval between the j-th reflow event and the previous mention. Adjust the time decay control parameter. The exponential decay intensity is such that the larger the value, the smaller the contribution of long-term reflux. This is the score for the quality of a single reflow. For weighted propagation path length, It is the largest weighted path length among all cultural and creative elements. This is the path suppression parameter. Let i be the response function for cultural and creative communication, representing the cognitive activation intensity of element i at time t. The variance of the response function, This is the mean of the response function. The volatility enhancement parameter adjusts the positive contribution of memory volatility to the overall index.

[0034] S30. Tensor fusion is performed between the cultural and creative memory cycle index and the multimodal expression of each cultural and creative element to construct a three-dimensional feature space. Manifold projection is then performed in this space to calculate the probability density overlap between the behavior distribution of each element and the ideal innovation area, and the cultural and creative perception novelty score is output.

[0035] S31. Based on the cultural and creative memory cycle index, collect three types of multimodal behavioral data: voice emotion fluctuation, cultural and creative emotion jump, and interactive action activity, and construct a three-dimensional cross-modal behavioral tensor.

[0036] After calculating the cultural and creative memory cycle index, to further reveal the concrete impact and emotional penetration of the index theme in interactions, it is necessary to go beyond the textual cognition level and capture users' multimodal behavioral responses in real interactions. Therefore, three types of key behavioral data were collected for each cultural and creative element: the variance of voice tone changes, used to quantify the degree of fluctuation in acoustic characteristics when users mention the theme, reflecting the fluctuations in their internal emotional intensity; the number of cultural and creative emotional polarity jumps, that is, the number of times emotions shift from positive to negative or vice versa in the dialogue surrounding the element, reflecting the depth of cognitive conflict or emotional resonance it evokes; and the standard score of interactive action frequency, measuring the significant increase in users' tactile behaviors such as patting and touching compared to their daily levels, characterizing the degree of activation of their physical participation. The three data were aligned and normalized in time to construct a three-dimensional cross-modal behavioral tensor.

[0037] S32. The three-dimensional cross-modal behavior tensor is reduced to a two-dimensional manifold through isometric mapping, the response of cultural and creative elements is represented as a continuous trajectory, and its first principal curvature and Gaussian curvature are calculated based on smooth fitting and differential geometric analysis.

[0038] An isometric mapping algorithm is applied to the 3D cross-modal behavior tensor to reveal the nonlinear intrinsic structure behind the 3D cross-modal behavior data. By constructing a neighborhood graph and calculating geodesic distances, the high-dimensional time series nonlinearity is reduced to a 2D manifold space, preserving its intrinsic geometric structure. Based on this low-dimensional space, the multimodal response of each textual manifold element is represented as a continuous time evolution trajectory. By smoothing and performing differential geometric analysis on this trajectory, the tangent and normal directions are estimated using local polynomial regression, and then the first principal curvature is calculated to reflect the curvature degree of the path and the intensity of behavioral transitions. Simultaneously, based on local surface fitting of neighboring points, the product of curvatures of the two principal directions is solved through principal component analysis to obtain Gaussian curvature, which characterizes the local complexity and stability of the response region. These two curvature features together reveal the nonlinear multimodal dynamic modes excited by textual manifold elements in interaction.

[0039] S33. Define the ideal innovation region as a Gaussian distribution, calculate the overlap between the behavior distribution of each cultural and creative element in the manifold space and this region, and integrate the geometric dynamic characteristics of its first principal curvature and Gaussian curvature to calculate the cultural and creative perceived novelty score.

[0040] Based on the first principal curvature and Gaussian curvature, to evaluate its innovativeness in user perception, the ideal innovative region is defined as a Gaussian distribution region in the manifold space with high emotional fluctuation, low intermodal linear correlation, and moderate curvature complexity, representing an interaction mode that combines emotional impact and cognitive novelty. Then, the overlap between the response distribution of each element in this space and the ideal region is calculated, and its curvature characteristics are weighted together. The perceived novelty score is mapped using the following formula: ,in, To score the perceived novelty of cultural and creative products, To control the steepness, adjust the slope of the S-curve. The comprehensive potential energy function reflects the innovative potential energy of the element in the manifold space. The higher the value, the closer its behavior pattern is to the ideal innovation region and the more significant its curvature characteristics. The center offset parameter determines the position of the midpoint of the S-shaped function. Represents the support domain of the distribution of cultural and creative element i in the two-dimensional manifold space. Distribution of behavioral trajectories The integral overlapping with the probability density of the ideal innovation region is the ideal innovation region as... The mean, Multivariate Gaussian distribution of covariance , where x is the position in the two-dimensional embedding space. To enhance weights geometrically, It is a geometric curvature enhancement term. As the first principal curvature, The infinite norm of the first principal curvature, For Gaussian curvature, It represents the average absolute value of the Gaussian curvature of all cultural and creative elements.

[0041] S40. Based on the novelty score of cultural and creative perception, select high-potential cultural and creative elements, construct a co-occurrence semantic subgraph and simulate a non-equilibrium thermodynamic process, assign initial cultural and creative temperature to nodes, iteratively calculate entropy yield and free energy gradient, identify the path with the steepest energy gradient, and extract the hot spot chain of cultural and creative integration.

[0042] S41. Select high-potential-energy elements based on the cultural and creative perception novelty score to construct a high-potential-energy cultural and creative subgraph, and determine the edge weights between nodes by integrating co-occurrence frequency and semantic vector cosine similarity.

[0043] All cultural and creative elements are ranked based on their perceived novelty scores, and the top Y high-potential elements are selected to form a high-potential cultural and creative subgraph. The choice of Y can be set according to the actual application scenario. To ensure that these high-potential elements are not isolated highlights, but form semantically coherent and logically understandable cultural and creative narrative units, these nodes are mapped back to the originally constructed cultural and creative co-occurrence semantic network. In this network, if two high-potential elements have appeared in the same context window in a user dialogue, the co-occurrence edge between them is retained. The edge weight is determined by the co-occurrence frequency and the cosine similarity of the semantic vector angle, ensuring that the subgraph is both novel and semantically coherent. The co-occurrence frequency is used to measure the frequency with which two cultural and creative elements are jointly mentioned in a user dialogue. The cosine similarity of the semantic vector angle measures the conceptual closeness of two cultural and creative elements in the semantic space, which is obtained by calculating the cosine similarity between the core expressions of the two elements through sentence vectors generated by a pre-trained language model.

[0044] S42. Based on the embedding degree of cultural and creative structures, initialize the node temperature, simulate the heat conduction process in the weighted network, and identify cognitive roles by iteratively calculating temperature evolution, entropy production rate and free energy decrease rate, revealing the dynamic evolution path of cultural and creative influence.

[0045] After constructing a high-potential-energy cultural and creative subgraph, a thermodynamic-like mechanism is introduced to simulate the dynamic propagation and influence evolution of this cultural and creative theme network in user cognition. The diffusion of cultural and creative information is analogized to the conduction of heat in a complex network, thus revealing its inherent evolutionary trend and stability characteristics. Each node is assigned an initial cultural and creative temperature, which does not represent the actual physical temperature but rather a quantitative representation of its influence potential energy within the user's cognitive structure. Its value is obtained by linearly mapping the element's embedding degree in the cultural and creative structure. The heat conduction rule is that heat is conducted non-equilibrium along the edges of the subgraph, driven by temperature difference, with the rate adjusted by edge weights. The temperature distribution is dynamically updated based on the temperature difference between nodes and edge weights. Iterative calculations are used to realize the diffusion and redistribution of cultural and creative influence from core elements to surrounding nodes. During this process, the entropy productivity and free energy decrease rate of each node are calculated to simulate the dynamic evolution of cultural and creative influence. Entropy production rate is the product of the instantaneous temperature difference change at each node during temperature evolution and the heat flux intensity of its neighborhood. Free energy is defined based on the weighted sum of squared temperature differences, and the rate of decrease in free energy is obtained by calculating the difference in free energy between adjacent iteration steps. Finally, through analysis of the temperature evolution trajectory and two dynamic indicators, different cognitive roles are identified, including persistent heat sources, information relays, rapid cooling, and stable convergence.

[0046] S43. Calculate the path strength, which includes the weighted product of temperature gradient, average entropy yield, maximum local instability and total entropy change of the path, and focus on identifying the path with the steepest energy gradient and the most intense information flow to capture the hot chain of cultural and creative integration.

[0047] Building upon the node-level thermodynamic role identification, to uncover coherent paths with strong evolutionary driving forces among cultural and creative elements, it is necessary to move beyond single-point dynamics and shift to path-level collaborative evolution analysis. Since the transmission of influence between nodes is not uniform but concentrated on connection chains with high potential energy differences, high information activity, and structural instability, it is essential to calculate the comprehensive evolution intensity of each path in the high-potential-energy cultural and creative subgraph. All paths are then sorted in descending order of evolution intensity, and the top-scoring paths are selected. Among these high-scoring paths, those with the steepest energy gradients, concentrated entropy production, and prominent structural instability are identified as hotspot chains for cultural and creative integration. The thermodynamic evolution intensity formula is: Where J is a path composed of multiple cultural and creative element nodes. This represents an edge in path J. Let be the absolute temperature difference between node i and node j. Let i be the average entropy productivity of nodes i and j. The maximum value of the Laplace norm for all nodes in path J represents the intensity of the strongest local mutation experienced by the path. This is the instability enhancement factor. The entropy change enhancement coefficient, Let J be the total entropy change along path J, and let y be the sum of the entropy yields of all nodes along the path.

[0048] S50 decodes the hot topic chain of cultural and creative integration into design semantic tags, and generates a doll appearance scheme with a consistent style by matching the material aesthetic database through Wasserstein distance; at the same time, it transforms its element sequence and emotional dynamics into interactive templates to guide the customized design of the doll's form and interactive content.

[0049] S51. By normalizing the temperature gradient and entropy change accumulation value of the cultural and creative integration hotspot chain, we construct the potential energy transition intensity and information evolution density index, and map them into design semantic tags based on the sensible engineering dictionary.

[0050] To transform abstract cultural and creative dynamics into a perceptible design language, a structured semantic mapping process is established for the two core parameters of the cultural and creative integration hotspot chain: temperature gradient and entropy change accumulation. First, the parameters are normalized. The maximum temperature gradient on the path is divided by the upper bound of the temperature difference across the entire network to obtain the potential energy transition intensity index. Simultaneously, the sum of the entropy yields of all nodes on the path is divided by the global average entropy change level to obtain the information evolution density index. Then, based on a pre-constructed perceptual engineering semantic dictionary, these two quantitative dimensions are mapped to human perception and emotional description words, respectively. When the potential energy transition intensity value is high, the path is determined to have a strong cognitive impact, triggering perceptual characteristics such as suddenness, intensity, and release. Combined with the basic emotional tone of warmth, an explosive warmth label is generated, corresponding design strategies including localized splashes of high-saturation colors, instantaneous illumination of LED light strips, or abrupt splicing of plush materials. When the information evolution density value is high, continuous information generation and cognitive uncertainty are identified, corresponding to perceptual characteristics such as gradualism, complexity, and the unknown. Combined with the cultural tendency towards mystery, a gradual mystery label is generated to guide the use of design techniques such as gradient dyeing, hidden embroidery patterns, slow breathing lighting effects, or multi-layered fabric stacking. Through two-dimensional cross-matching, a composite design semantic label is generated, forming a complete mapping chain from data-driven evolution intensity to concrete aesthetic expression, providing interpretable and actionable intuitive guidance for the appearance and interaction design of AI dolls.

[0051] S52. Based on design semantic tags, features are matched in the material aesthetics database through optimal transmission theory, the composite semantic decomposition sub-targets are decomposed and the Cesarstein distance is minimized to generate an AI doll appearance scheme with a unified style and progressive meaning.

[0052] After obtaining the design semantic tags, they are used as query conditions and cross-modal matched with a pre-built material aesthetics database to generate AI plush toy appearance designs that conform to both cultural evolution logic and aesthetic consistency. This knowledge base contains structured descriptions of numerous design elements, including materials, colors, forms, and their corresponding emotional and cultural connotations. The matching process employs optimal transport theory, treating each design semantic tag as a source distribution and candidate design element combinations in the knowledge base as target distributions. It searches for the solution that minimizes the Cerestan distance among all candidate combinations, achieving the most faithful mapping from semantics to design. For composite semantic tags… This approach corresponds to the superposition of multiple abstract design dimensions, which are decomposed into several quantifiable sub-objectives. For each sub-objective, corresponding feature patterns are extracted from a material aesthetics database. The performance of each candidate design scheme on these features is matched with the target semantics to evaluate the fit between the design scheme and the composite semantics. Finally, the solution with the smallest overall difference is sought among all candidates. This results in a set of design schemes that are stylistically unified in terms of color gradation, material transition, and form evolution, with progressively deeper meanings.

[0053] S53. Based on the structure and dynamic characteristics of the cultural and creative integration hotspot chain, generate interactive templates that include personality, dialogue logic and feedback strategies, systematically guide the interactive design of AI dolls, and achieve accurate expression of emotions and culture.

[0054] After identifying the hot topics in the integration of cultural and creative industries, their inherent structure and dynamic characteristics—including the evolutionary order of element sequences, the emotional intensity changes reflected by temperature gradients, the cognitive activity reflected by entropy productivity, and the overall narrative tension of the path—are transformed into executable interactive script generation rules. This achieves a systematic mapping from cultural and creative data analysis to AI doll behavior design. The element sequence is transformed into a topic progression path in the dialogue, determining the logical order in which the doll tells the story or guides interaction. The steepness of the temperature gradient is mapped to the intensity change curve of emotional expression, used to set the emotional outburst points of the character at key nodes, and accordingly defining the affinity and emotional sensitivity parameters in the character's personality. Nodes with high entropy productivity are considered key points for cognitive reconstruction, corresponding to active guidance strategies such as asking questions, counter-questions, or silent reflection in the script, shaping the doll's narrative style as exploratory or guiding. The overall free energy decline trend of the path is used to construct the topic switching logic. If the decline is rapid, a direct convergent dialogue strategy is adopted to quickly establish consensus; if the process is slow, gradual transitional topics and metaphorical expressions are introduced to maintain dialogue tension. In addition, the feedback strategy is dynamically adjusted according to the node temperature: high-temperature nodes trigger rapid responses, complex body movements and rich voice tones to reflect a high level of engagement; low-temperature or stable nodes respond with brief responses or gentle actions to maintain energy efficiency and friendliness.

[0055] These rules have been integrated into customized content templates, forming a set of well-structured and emotionally coherent interaction design specifications. This systematically guides the development of the next generation of AI plush toys in terms of semantic understanding, emotional expression, and behavioral feedback, making their interactions not only natural and smooth, but also faithfully conveying the deep cultural resonance and emotional evolution trajectory in user dialogues.

[0056] Example 2 like Figure 2 As shown, Embodiment 2 of this application provides a data analysis system for cultural and creative elements based on natural language processing technology, including: Extraction and Embedding Degree Calculation Module: Based on natural language data collected from AI-powered smart plush toys in cultural and creative themed interactions, this module extracts cultural and creative elements through named entity recognition. Combining these elements with their frequency, betweenness centrality, and clustering coefficients in the semantic network, it generates a cultural and creative structure embedding degree index. Specifically, it is divided into the following sub-modules: The cultural and creative elements classification submodule accurately extracts and classifies cultural and creative elements from interactive doll texts through named entity recognition and cultural and creative ontology knowledge graphs, ensuring the semantic accuracy and consistency of the recognition results.

[0057] The extended dimension module is based on dependency parsing. It weights the adjectives, verbs and metaphors in the context of cultural and creative elements and introduces a syntactic distance decay factor to calculate the extended dimension score of the context.

[0058] The cultural and creative structure embedding degree submodule: Based on the co-occurrence relationship of cultural and creative elements, a weighted semantic network is constructed. The embedding degree of cultural and creative structure is generated by combining topological indicators such as betweenness centrality, clustering coefficient and feature vector centrality with occurrence frequency and context expansion score.

[0059] The memory cycle index module: Based on the embeddedness of cultural and creative structures, it constructs an impulse response function to characterize the cognitive dynamics of cultural and creative themes. Combined with weighted propagation path length and information feedback analysis, it generates a cultural and creative memory cycle index. Specifically, it is divided into the following sub-modules: Impulse Response Submodule: Using the embedding degree of cultural and creative structures as the initial weight, a dynamic cognitive activation function formula based on the superposition of impulse responses is constructed. The exponential decay kernel is used to simulate memory decay, and the periodic oscillation term is used to capture the echo pattern, comprehensively reflecting the evolution process of the activation intensity of cultural and creative elements after they are mentioned in user dialogue.

[0060] The path weighted length submodule calculates the weighted length of the cultural and creative structure embedding degree by introducing the geometric mean of the embedding degree of cultural and creative structure as the denominator.

[0061] The Cultural and Creative Memory Cycle Index submodule identifies information return events, assesses cognitive reconstruction by combining semantic similarity and sentence variation, and comprehensively considers return quality, dissemination efficiency, and memory volatility. It then uses a formula to quantify the continuous activation capacity of cultural and creative themes and generates the Cultural and Creative Memory Cycle Index.

[0062] The Fusion and Novelty Calculation Module: This module performs tensor fusion of the cultural and creative memory cycle index with the multimodal expressions of various cultural and creative elements to construct a three-dimensional feature space. Manifold projection is then performed within this space to calculate the probability density overlap between the behavioral distribution of each element and the ideal innovation region, outputting a perceived novelty score for the cultural and creative products. Specifically, it is divided into the following sub-modules: Three-dimensional cross-modal tensor module: Based on the cultural and creative memory cycle index, it collects three types of multimodal behavioral data: voice emotion fluctuation, cultural and creative emotion jump, and interactive action activity, and constructs a three-dimensional cross-modal behavioral tensor.

[0063] The manifold projection and feature extraction submodule reduces the dimensionality of the 3D cross-modal behavior tensor to a 2D manifold through isometric mapping, represents the response of cultural and creative elements as a continuous trajectory, and calculates its first principal curvature and Gaussian curvature based on smooth fitting and differential geometric analysis.

[0064] The Cultural and Creative Perceived Novelty Submodule defines the ideal innovation region as a Gaussian distribution, calculates the overlap between the behavioral distribution of each cultural and creative element in the manifold space and this region, and integrates the geometric dynamic characteristics of its first principal curvature and Gaussian curvature to calculate the cultural and creative perceived novelty score.

[0065] The filtering and hotspot chain extraction module filters high-potential cultural and creative elements based on their perceived novelty score, constructs a co-occurrence semantic subgraph and simulates a non-equilibrium thermodynamic process, assigns initial cultural temperature to nodes, iteratively calculates entropy productivity and free energy gradient, identifies the path with the steepest energy gradient, and extracts cultural integration hotspot chains. Specifically, it is divided into the following sub-modules: The sub-module for constructing cultural and creative subgraphs selects high-potential-energy elements based on the novelty score of cultural and creative perception to construct a high-potential-energy cultural and creative subgraph, and determines the edge weights between nodes by fusing co-occurrence frequency and semantic vector cosine similarity.

[0066] Thermodynamic Evolution Identification Submodule: Based on the embedding degree of cultural and creative structures, the node temperature is initialized, the heat conduction process in the weighted network is simulated, and the cognitive role is identified by iteratively calculating the temperature evolution, entropy production rate and free energy decrease rate, revealing the dynamic evolution path of cultural and creative influence.

[0067] The hotspot chain extraction submodule calculates the path strength, which includes the weighted product of temperature gradient, average entropy yield, maximum local instability, and total path entropy change. It focuses on identifying the paths with the steepest energy gradient and the most intense information flow to capture the hotspot chains of cultural and creative integration.

[0068] The decoding and solution generation module decodes the trending topics of cultural and creative integration into design semantic tags. Using Wasserstein distance matching with a material aesthetics database, it generates a coherent doll appearance scheme. Simultaneously, it transforms the element sequences and emotional dynamics into interactive templates to guide the customized design of the doll's form and interactive content. Specifically, it is divided into the following sub-modules: The semantic tag mapping submodule is designed by constructing potential energy transition intensity and information evolution density indicators by normalizing the temperature gradient and entropy change accumulation value of the cultural and creative integration hotspot chain, and mapping them to design semantic tags based on the sensible engineering dictionary.

[0069] Appearance scheme matching submodule: Based on design semantic tags, it matches features in the material aesthetics database through optimal transmission theory, decomposes the composite semantic sub-targets and minimizes the Cesarstein distance, and generates an AI doll appearance scheme with a unified style and progressive meaning.

[0070] Interactive Template Generation Submodule: Based on the structure and dynamic characteristics of the cultural and creative integration hotspot chain, it generates interactive templates that include personality, dialogue logic and feedback strategies, systematically guiding the interactive design of AI dolls and achieving accurate expression of emotions and culture.

[0071] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for analyzing cultural and creative element data based on natural language processing technology, characterized in that, include: S10. Based on the natural language data collected by AI smart plush toys in cultural and creative theme interaction, cultural and creative elements are extracted through named entity recognition, and combined with their frequency, betweenness centrality and clustering coefficient in the semantic network, a cultural and creative structure embedding degree index is generated. S20. Based on the embedding degree of cultural and creative structure, an impulse response function is constructed to characterize the cognitive dynamics of cultural and creative themes. Combined with weighted propagation path length and information feedback analysis, a cultural and creative memory cycle index is generated. S30. Tensor fusion is performed between the cultural and creative memory cycle index and the multimodal expression of each cultural and creative element to construct a three-dimensional feature space. Manifold projection is then performed in this space to calculate the probability density overlap between the behavior distribution of each element and the ideal innovation area, and output the cultural and creative perception novelty score. S40. Based on the novelty score of cultural and creative perception, select high-potential cultural and creative elements, construct a co-occurrence semantic subgraph and simulate a non-equilibrium thermodynamic process, assign initial cultural temperature to nodes, iteratively calculate entropy productivity and free energy gradient, identify the path with the steepest energy gradient, and extract cultural integration hotspot chains. S50 decodes the hot topic chain of cultural and creative integration into design semantic tags, and generates a doll appearance scheme with a consistent style by matching the material aesthetic database through Wasserstein distance; at the same time, it transforms its element sequence and emotional dynamics into interactive templates to guide the customized design of the doll's form and interactive content.

2. The method for analyzing cultural and creative element data based on natural language processing technology as described in claim 1, characterized in that, Based on natural language data collected from AI-powered smart plush toys in cultural and creative themed interactions, cultural and creative elements are extracted through named entity recognition. These elements are then combined with their frequency, betweenness centrality, and clustering coefficients in the semantic network to generate a cultural and creative structure embedding degree index. The specific steps are as follows: By using named entity recognition and cultural and creative ontology knowledge graph, cultural and creative elements are accurately extracted and classified from interactive doll texts, ensuring the semantic accuracy and consistency of the recognition results. Based on dependency parsing, the adjectives, verbs and metaphors in the context of cultural and creative elements are weighted according to their weights, and a syntactic distance decay factor is introduced to calculate the context expansion dimension score. A weighted semantic network is constructed based on the co-occurrence relationship of cultural and creative elements. The embedding degree of cultural and creative structures is generated by combining topological indicators such as betweenness centrality, clustering coefficient and feature vector centrality with occurrence frequency and context expansion score.

3. The method for analyzing cultural and creative element data based on natural language processing technology as described in claim 2, characterized in that, By leveraging named entity recognition and a knowledge graph of cultural and creative ontologies, cultural and creative elements are accurately extracted and categorized from interactive texts about dolls, ensuring the semantic accuracy and consistency of the recognition results. Specifically, this involves the following sub-steps: Construct a knowledge graph of cultural and creative ontology to form a standardized semantic reference system for cultural and creative products; Fine-grained entity recognition of user dialogue text is performed using a hybrid model of bidirectional encoder-representation converter-bidirectional long short-term memory network-conditional random field. By using knowledge graph embedding technology, entities and relationships in the knowledge graph are mapped to a low-dimensional vector space. The semantic distance between the candidate entities output by the model and their potential matching nodes in the graph is calculated, and only high-confidence matching results with a distance less than a preset threshold are retained.

4. The method for analyzing cultural and creative element data based on natural language processing technology as described in claim 1, characterized in that, Based on the embeddedness of cultural and creative structures, an impulse response function is constructed to characterize the cognitive dynamics of cultural and creative themes. Combined with weighted propagation path length and information feedback analysis, a cultural and creative memory cycle index is generated. Specifically, this involves the following sub-steps: Using the embedding degree of cultural and creative structures as the initial weight, a dynamic cognitive activation function formula based on the superposition of impulse responses is constructed. The exponential decay kernel is used to simulate memory decay, and the periodic oscillation term is used to capture the echo pattern, so as to comprehensively reflect the evolution process of the activation intensity of cultural and creative elements after they are mentioned in user dialogue. By introducing the geometric mean of the embedding degree of cultural and creative structures as the denominator, the weighted length of the cultural and creative dissemination path in the dialogue flow is calculated. By identifying information feedback events and assessing cognitive reconstruction by combining semantic similarity and sentence variation, the system integrates feedback quality, dissemination efficiency, and memory volatility to quantify the continuous activation capacity of cultural and creative themes using a formula, thereby generating a cultural and creative memory cycle index.

5. The method for analyzing cultural and creative element data based on natural language processing technology as described in claim 1, characterized in that, The cultural and creative memory cycle index is fused with the multimodal expressions of various cultural and creative elements using tensors to construct a three-dimensional feature space. A manifold projection is then performed within this space to calculate the probability density overlap between the behavioral distribution of each element and the ideal innovation region, outputting a cultural and creative perceived novelty score. Specifically, this involves the following sub-steps: Based on the cultural and creative memory cycle index, three types of multimodal behavioral data were collected: voice emotion fluctuation, cultural and creative emotion jump, and interactive action activity, to construct a three-dimensional cross-modal behavioral tensor. The three-dimensional cross-modal behavior tensor is reduced to a two-dimensional manifold by isometric mapping, the response of cultural and creative elements is represented as a continuous trajectory, and its first principal curvature and Gaussian curvature are calculated based on smooth fitting and differential geometric analysis. The ideal innovation region is defined as a Gaussian distribution. The overlap between the behavior distribution of each cultural and creative element in the manifold space and this region is calculated. The geometric dynamic characteristics of its first principal curvature and Gaussian curvature are integrated to calculate the perceived novelty score of cultural and creative products.

6. The method for analyzing cultural and creative element data based on natural language processing technology as described in claim 1, characterized in that, Based on the novelty score of cultural and creative perception, high-potential cultural and creative elements are selected. A co-occurrence semantic subgraph is constructed and a non-equilibrium thermodynamic process is simulated. Nodes are assigned an initial cultural temperature, and entropy productivity and free energy gradient are iteratively calculated. The path with the steepest energy gradient is identified, and the cultural integration hotspot chain is extracted. Specifically, this involves the following sub-steps: High-potential elements are selected based on the novelty score of cultural and creative perception to construct a high-potential cultural and creative subgraph, and the edge weights between nodes are determined by integrating co-occurrence frequency and semantic vector cosine similarity. Based on the embedding degree of cultural and creative structures, the node temperature is initialized, the heat conduction process in the weighted network is simulated, and the cognitive roles are identified by iteratively calculating the temperature evolution, entropy production rate and free energy decrease rate, revealing the dynamic evolution path of the influence of cultural and creative industries. The calculation of path strength is based on a weighted product of temperature gradient, average entropy yield, maximum local instability, and total entropy change of the path. It focuses on identifying paths with the steepest energy gradient and the most intense information flow, and captures hot chains of cultural and creative integration.

7. The method for analyzing cultural and creative element data based on natural language processing technology as described in claim 1, characterized in that, The trend of cultural and creative integration is decoded into design semantic tags. By matching the material aesthetics database with Wasserstein distance, a coherent doll appearance scheme is generated. Simultaneously, its element sequences and emotional dynamics are transformed into interactive templates to guide the customized design of the doll's form and interactive content. Specifically, this involves the following sub-steps: By normalizing the temperature gradient and entropy change accumulation value of the cultural and creative integration hotspot chain, we construct the potential energy transition intensity and information evolution density index, and map them into design semantic tags based on the sensible engineering dictionary. Based on design semantic tags, features are matched in the material aesthetics database through optimal transmission theory, the composite semantic decomposition sub-targets are decomposed and the Cesarstein distance is minimized to generate an AI doll appearance scheme with a unified style and progressive meaning. Based on the structure and dynamic characteristics of the cultural and creative integration hotspot chain, an interactive template containing personality, dialogue logic and feedback strategies is generated to systematically guide the interactive design of AI dolls and achieve accurate expression of emotions and culture.

8. A data analysis system for cultural and creative elements based on natural language processing technology, characterized in that, include: Extraction and Embedding Degree Calculation Module: Based on the natural language data collected by AI smart plush toys in cultural and creative theme interactions, cultural and creative elements are extracted through named entity recognition, and combined with their frequency, betweenness centrality and clustering coefficient in the semantic network, a cultural and creative structure embedding degree index is generated. The memory cycle index module: Based on the embedding degree of cultural and creative structures, an impulse response function is constructed to characterize the cognitive dynamics of cultural and creative themes. Combined with weighted propagation path length and information feedback analysis, a cultural and creative memory cycle index is generated. Fusion and Novelty Calculation Module: Tensor fusion is performed between the cultural and creative memory cycle index and the multimodal expression of each cultural and creative element to construct a three-dimensional feature space. In this space, manifold projection is performed to calculate the probability density overlap between the behavior distribution of each element and the ideal innovation area, and output the cultural and creative perceived novelty score. Screening and Hotspot Chain Extraction Module: Based on the novelty score of cultural and creative perception, high-potential cultural and creative elements are screened, a co-occurrence semantic subgraph is constructed and a non-equilibrium thermodynamic process is simulated, nodes are given an initial cultural temperature, entropy yield and free energy gradient are iteratively calculated, the path with the steepest energy gradient is identified, and cultural integration hotspot chains are extracted. Decoding and Solution Generation Module: Decodes the hot topic chain of cultural and creative integration into design semantic tags, and generates a doll appearance scheme with a coherent style by matching the material aesthetic database through Wasserstein distance; at the same time, it transforms its element sequence and emotional dynamics into interactive templates to guide the customized design of the doll's form and interactive content.

Images