Aigc-based personalized tourism scenic spot publicity copy automatic generation method and system

Abstract

The application discloses a tourism scenic spot personalized publicity copy automatic generation method and system based on AIGC, and relates to the technical field of artificial intelligence and natural language processing. Through multi-modal data fusion and tourist state dynamic modeling, the problem that traditional methods cannot realize real-time perception of tourist interest and emotional change is solved, and accurate input is provided for personalized generation; fine-grained knowledge graph and graph neural network retrieval are used to realize deep integration and accurate calling of scenic cultural knowledge, and overcome the defects of the prior art, such as empty content and lack of cultural connotation; through reinforcement learning strategy control and multi-task hierarchical generation model, adaptive and accurate regulation of copy style, emotion and information is realized, the cultural accuracy and emotional resonance are ensured, the content generation efficiency and personalized matching degree are greatly improved, and the quality, adaptability and communication value of the tourism publicity content are significantly improved.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence and natural language processing technology, specifically to a method and system for automatically generating personalized promotional copy for tourist attractions based on AIGC. Background Technology

[0002] As a crucial medium connecting scenic spots and tourists, promotional copywriting for tourist attractions plays a key role in stimulating tourism interest, disseminating cultural values, and enhancing the tourist experience. Its quality directly impacts a destination's attractiveness and market competitiveness. However, the production of current tourism promotional content generally faces systemic challenges: on the one hand, traditional manual creation methods are inefficient and costly, failing to meet the massive demand for personalized content; on the other hand, while template-based automated solutions exist, these methods often result in highly homogenized content, lacking deep adaptation to individual tourist characteristics, real-time contexts, and cultural connotations.

[0003] In practical applications, existing technologies exhibit significant limitations. Most automated generation systems rely solely on simple user tags or historical behavior for static recommendations, failing to dynamically perceive real-time changes in tourists' states during their visits, such as shifts in interest due to geographical location changes, emotional fluctuations caused by environmental factors (weather, season), and preference evolution reflected in social media behavior. Furthermore, these systems often struggle to deeply integrate the unique historical and cultural knowledge of scenic spots, resulting in generated content lacking cultural accuracy and emotional resonance, and failing to reflect the unique value of the stories, customs, and other aspects behind the landscape. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for automatically generating personalized promotional copy for tourist attractions based on AIGC. By sensing the status of tourists in real time through multimodal data, and combining fine-grained scenic spot knowledge graphs with reinforcement learning strategies, personalized promotional copy is dynamically generated and optimized to deeply adapt to individual tourist characteristics, real-time context and cultural background.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] This application provides a method for automatically generating personalized promotional copy for tourist attractions based on AIGC, including the following steps:

[0007] S1. By deploying IoT sensors in the scenic area and user terminals, heterogeneous tourist data is acquired in real time. Multimodal deep neural networks are used to align and fuse the heterogeneous data to obtain a quantitative representation of the tourist status.

[0008] S2. By using a pre-constructed fine-grained scenic area knowledge graph, combined with the spatiotemporal coordinates and interest tags in the quantitative representation of real-time tourist status, a semantic retrieval algorithm based on graph neural networks is adopted to generate a personalized set of contextual knowledge.

[0009] The pre-constructed fine-grained scenic area knowledge graph includes scenic area landscape entities, historical events, cultural elements, and their spatiotemporal semantic relationships.

[0010] S3. Analyze the quantitative representation of tourist status through reinforcement learning policy network, establish a mapping model from tourist characteristics to copywriting style control parameters, and automatically generate the optimal style control parameters.

[0011] S4. Integrate the quantitative representation of tourist status, contextual knowledge set and optimal style control parameters, input the multi-task generation model based on the Transformer architecture, and output multiple versions of personalized candidate copy by implementing a three-layer progressive decoding strategy.

[0012] S5. Through the content evaluation model, the candidate texts of multiple versions are comprehensively scored from multiple dimensions. Combined with the format specifications of the target publishing channels, the adaptive conversion algorithm is used to restructure and adapt the format to generate the final content for different media.

[0013] S6. Through the tracking module, real-time data on user exposure, interaction and conversion behavior of the generated final content is collected. Attribution analysis model is used to correlate user behavior data with corresponding generation control parameters and visitor characteristics to form a traceable generation effect evaluation record.

[0014] S7. Based on the constructed reinforcement learning model infrastructure, the generated performance evaluation data is used as training samples to optimize the policy gradient of the mapping model of the generation control parameters, while the multi-task generation model is incrementally trained.

[0015] This application provides an AIGC-based automatic generation system for personalized promotional copy for tourist attractions, used to implement the aforementioned AIGC-based automatic generation method for personalized promotional copy for tourist attractions, including:

[0016] The tourist feature perception module collects tourist location, environment, interaction and social data in real time through multi-source IoT sensors, and uses multimodal deep neural networks to align and fuse the data, outputting a quantitative feature vector containing spatiotemporal preferences, interest intensity and emotional state.

[0017] The contextual knowledge retrieval module, based on a pre-built fine-grained scenic area knowledge graph, combines tourists' real-time location, time, and interest tags, and uses a graph neural network semantic retrieval algorithm to output a personalized knowledge set that is highly matched with the current tourist context.

[0018] The style parameter generation module uses an Actor-Critic reinforcement learning network to map tourist feature vectors into style control parameters such as cultural concentration coefficient, language style code, emotional tone vector, and information density threshold, and outputs the optimal parameter combination to guide copywriting generation.

[0019] The content generation optimization module integrates tourist characteristics, contextual knowledge, and style parameters. It implements three-layer progressive decoding through a Transformer-based multi-task generation model and outputs a set of personalized candidate texts with multiple versions that have undergone quality filtering.

[0020] The multi-dimensional assessment and adaptation module uses a three-dimensional assessment model of cultural fit, personalization matching and dissemination potential to score candidate copy in parallel, and intelligently reconstructs the format in combination with the target channel specifications to output the final content that is ready for the channel and the corresponding release tracking mark.

[0021] The effect monitoring and attribution module collects user exposure, interaction and conversion behavior data in real time through the tracking system, establishes causal relationships between behavior and generated parameters using a deep attribution analysis model, and outputs a structured evaluation record that includes parameter impact analysis and optimization suggestions.

[0022] The closed-loop optimization and evolution module constructs a reinforcement learning environment based on the effect evaluation data, optimizes the style parameter generation model online, and performs incremental training on the content generation model, outputting continuously iterative model parameters and performance indicators.

[0023] The beneficial effects of this invention are as follows:

[0024] This invention utilizes IoT sensors and multimodal deep neural networks to acquire and integrate multi-source heterogeneous data such as tourists' spatiotemporal trajectories and emotional states in real time, constructing a dynamically updated quantitative representation of tourists' states. This effectively solves the problem in the background technology that traditional methods cannot dynamically perceive changes in tourists' real-time states (such as geographical location movement and environmental factors). It achieves accurate capture of tourists' real-time interests, emotions, and behavioral evolution, thus laying a precise data foundation for the subsequent generation of personalized content.

[0025] By introducing a fine-grained scenic area knowledge graph and a graph neural network semantic retrieval algorithm, this invention can automatically retrieve and generate a personalized contextual knowledge set that integrates cultural allusions and historical events based on the real-time status of tourists. This overcomes the limitations of existing systems that are difficult to deeply integrate the unique historical and cultural knowledge of scenic areas, resulting in a lack of cultural accuracy and emotional resonance in the content. As a result, the generated promotional copy can not only dynamically adapt to tourists' interests, but also deeply reflect the cultural value and unique connotation behind the landscape, significantly improving the accuracy and appeal of the content.

[0026] By employing a three-layer progressive decoding strategy that combines reinforcement learning strategy networks with a multi-task generation model, this invention achieves adaptive mapping from tourist characteristics to copywriting style control parameters and can output multiple versions of personalized candidate copywriting. This breaks through the bottlenecks of low efficiency and severe content homogenization caused by relying on manual or template-based methods. It not only significantly improves the automation and efficiency of content production, but also ensures that the copywriting is highly matched with individual tourists and real-time scenarios in terms of style, emotion, and information density, thereby comprehensively enhancing the attractiveness and dissemination effect of promotional content. Attached Figure Description

[0027] To better understand and implement this application, the technical solution is described in detail below with reference to the accompanying drawings.

[0028] Figure 1 This is a flowchart illustrating the method for automatically generating personalized promotional copy for tourist attractions based on AIGC, as provided in Embodiment 1 of this application.

[0029] Figure 2 This is a flowchart illustrating step S2 in the AIGC-based method for automatically generating personalized promotional copy for tourist attractions, as provided in Embodiment 1 of this application.

[0030] Figure 3 This is a flowchart illustrating step S4 of the AIGC-based method for automatically generating personalized promotional copy for tourist attractions, as provided in Embodiment 1 of this application.

[0031] Figure 4 This is a schematic diagram of the structure of the AIGC-based automatic generation system for personalized promotional copy for tourist attractions provided in Embodiment 2 of this application. Detailed Implementation

[0032] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, exemplary embodiments will be described in detail below, examples of which are illustrated in the accompanying drawings. In the following description, when referring to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application.

[0033] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used herein are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0034] The following detailed description of the specific implementation methods, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided in detail.

[0035] Example 1

[0036] Please see Figures 1-3 This embodiment provides a method for automatically generating personalized promotional copy for tourist attractions based on AIGC (Artificial Intelligence Generated Content). The AIGC technology includes the following steps:

[0037] S1. By deploying IoT sensors in the scenic area and user terminals, heterogeneous data such as tourist location trajectory, environmental parameters, interactive behavior and social media data are acquired in real time. Multimodal deep neural networks are used to align and fuse the heterogeneous data, and dynamic feature vectors containing spatiotemporal preferences, interest intensity and emotional state are extracted to obtain a quantitative representation of tourist status.

[0038] Further, step S1 specifically includes:

[0039] The scenic area collects tourists' spatiotemporal trajectory data and environmental parameter data through cameras, WiFi probes, and environmental sensors; it also collects tourists' interactive behavior data through mobile terminal application interfaces; and it collects text, images, and location check-in data posted by tourists through authorized open social media platform interfaces.

[0040] The collected heterogeneous data is cleaned, denoised, and standardized in format: trajectory data is processed for coordinate unification and path smoothing; environmental parameter data is synchronized in time and normalized in units; text data is segmented, stop words are removed, and sentiment polarity is labeled; and image data is extracted and classified for scene labeling.

[0041] The preprocessed multimodal data is input into a multimodal deep neural network based on an attention mechanism. The network includes a text encoder, an image encoder, and a temporal data encoder. The alignment and weighted fusion of feature vectors from different modalities are achieved through a cross-modal attention layer to generate a unified fused feature representation.

[0042] Based on a unified fusion feature representation, feature components of three dimensions—spatiotemporal preference intensity, interest label weight, and emotional state value—are calculated through a fully connected neural network layer, ultimately generating a dynamic feature vector with fixed dimensions as a quantitative representation of tourist status.

[0043] The spatiotemporal preference intensity is calculated based on the length of stay and frequency of visits by tourists in different scenic areas; the interest tag weight is determined by the frequency of interaction with various cultural themes and deep learning; and the emotional state value is calculated by combining sentiment analysis of tourists' social media texts with environmental parameters.

[0044] Furthermore, a unified fusion feature representation is generated, including: a three-way parallel processing architecture comprising a text encoder, an image encoder, and a temporal data encoder, designed to meet the specific needs of personalized copywriting generation for tourist attractions; the text encoder uses a BERT model finely tuned on tourism corpora to specifically extract interest preferences, sentiment tendencies, and cultural semantic features from tourists' social media texts; the image encoder is based on the ResNet-50 architecture and adds a scene classification module to focus on identifying landscape types, seasonal features, and activity scenes in tourist-taken images; the temporal encoder uses a bidirectional LSTM network to specifically process the spatiotemporal patterns of tourists' movement trajectories, stay sequences, and interactive behaviors; through a cross-modal attention fusion mechanism, the system establishes a bidirectional alignment relationship between the text, image, and temporal modalities: the text-image attention layer calculates the relevance weights between semantic descriptions and visual scenes, the temporal-spatial attention layer associates behavioral sequences with spatiotemporal patterns of geographic coordinates, and the global multi-head attention layer integrates all modal features for adaptive weighted fusion. This fusion mechanism specifically introduces tourism knowledge-guided attention constraints to ensure that culturally relevant features receive higher weight during the fusion process. The final generated joint feature representation is transformed by a fully connected layer to output a dynamic feature vector containing three dimensions: spatiotemporal preference intensity, cultural interest weight, and real-time emotional state. This provides accurate and interpretable representations of tourist states for subsequent personalized copywriting generation.

[0045] Specifically, by deploying a sensor network and using a specially designed multimodal deep neural network, real-time data on tourists' trajectories, environment, interactions, and social media is fused to accurately extract dynamic feature vectors containing spatiotemporal preferences, interest intensity, and emotional state. This directly solves the problem in the background technology that existing systems cannot dynamically perceive the real-time state evolution of tourists caused by changes in geographical location, environmental factors, etc. during their visit. It achieves deep, accurate, and quantifiable dynamic perception of individual tourist characteristics and real-time context, laying a reliable data foundation for generating copy that is truly adapted to individuals and scenarios. This overcomes the shortcomings of traditional methods that rely on static tags, resulting in superficial adaptation and rigid content.

[0046] S2. By using a pre-constructed fine-grained scenic area knowledge graph, combined with the spatiotemporal coordinates and interest tags in the real-time generated quantitative representation of tourist status, a semantic retrieval algorithm based on graph neural networks is adopted to dynamically match landscape nodes, cultural allusions and activity information related to the current tourist situation from the knowledge graph, and generate a personalized set of contextual knowledge.

[0047] The pre-constructed fine-grained scenic area knowledge graph includes scenic area landscape entities, historical events, cultural elements, and their spatiotemporal semantic relationships.

[0048] Further, step S2 specifically includes:

[0049] S21. Extract the real-time geographic location coordinates, timestamps, and interest tag set from the quantitative representation of tourist status, and transform them into a knowledge graph query language. Construct a composite query graph pattern that combines spatiotemporal constraints and semantic interest constraints. The spatiotemporal constraints limit the temporal validity and spatial proximity of nodes, and the interest constraints are based on the semantic similarity between tourist interest tags and knowledge nodes.

[0050] S22. Input the composite query into the pre-constructed fine-grained scenic area knowledge graph, use a retrieval algorithm based on graph attention network to calculate the matching degree between the query conditions and the nodes and relationships in the knowledge graph, traverse the nodes in the graph that meet the conditions, dynamically expand the retrieval scope through the graph propagation mechanism, and finally extract the connected knowledge subgraph containing relevant entities, relationships and attributes.

[0051] S23. Perform structured analysis on the retrieved knowledge subgraphs, extract information fragments such as landscape descriptions, historical events, and cultural elements, sort the knowledge fragments according to the weight of tourist interest tags, and adaptively adjust the knowledge content in combination with real-time environmental parameters (such as weather and season) to generate a structured personalized contextual knowledge set.

[0052] Specifically, the structured contextual knowledge set is converted into a multimodal representation, including text summaries, keyword sets, associated image labels, and spatial coordinate sequences, forming a knowledge representation vector that can be directly input into downstream generative models.

[0053] The pre-constructed fine-grained scenic area knowledge graph includes: extracting entities, attributes, and relationships from official scenic area materials, local chronicles, academic literature, and authoritative tour guide texts to establish a basic graph structure; performing knowledge completion and semantic relationship enhancement based on visitor-generated content (UGC) and expert-annotated data; and mapping entities and relationships to a low-dimensional vector space through a graph embedding algorithm to support efficient semantic similarity calculation and graph neural network operations.

[0054] The semantic retrieval algorithm based on graph neural networks specifically adopts a heterogeneous graph attention network, which can simultaneously process multiple node types (landscape, event, person) and relationship types (spatiotemporal relationship, semantic association, subordinate relationship) in the knowledge graph, and achieve accurate context-aware retrieval.

[0055] Furthermore, the specific implementation of constructing the composite query graph pattern and performing the retrieval in step S2 includes:

[0056] Construction of a composite query graph pattern: Real-time geographic coordinates of tourists are mapped to spatial constraints, timestamps are converted into time validity constraints, and interest tag sets are transformed into semantic query vectors. Specifically, a two-dimensional "spatiotemporal-interest" constraint model is established: spatial constraints limit the retrieval radius (e.g., relevant nodes within a 500-meter range) using a geographic coordinate buffer algorithm; temporal constraints filter seasonal landscapes and time-sensitive activities using a time window function; and interest constraints assign retrieval weights to different interest tags based on semantic similarity calculations, forming a weighted multi-constraint query graph pattern.

[0057] Matching Degree Calculation Based on Graph Attention Network: A heterogeneous attention network is used to calculate the matching degree between a composite query and a knowledge graph. This includes: firstly, encoding the spatiotemporal constraints and semantic interest constraints in the composite query graph pattern into multi-dimensional query vectors. Spatial constraints are converted into coordinate feature vectors through geocoding, temporal constraints extract periodic features through a time encoder, and interest constraints generate interest embedding vectors through a semantic encoder. The algorithm employs a heterogeneous graph attention network architecture and designs a node type-aware attention mechanism, calculating attention weights for different node types such as landscape entities, historical events, and cultural elements. For landscape nodes, the spatial proximity weight is emphasized, using a distance-decay-based spatial attention function; for historical event nodes, the temporal relevance weight is emphasized. The algorithm employs a temporal attention module to analyze time window matching. For cultural element nodes, it emphasizes semantic similarity weighting, comparing query interest vectors with node semantic vectors through a cross-modal attention layer. In terms of relation matching calculation, the algorithm uses a relation-aware message passing mechanism to reason along three core relation paths: "spatial-temporal association, semantic association, and subordinate association." Spatial relation paths calculate coordinate distance and topological connection strength; temporal relation paths analyze event temporal logic and seasonal matching; and semantic relation paths evaluate concept relevance. Finally, a multi-head attention aggregation layer integrates node-level and relation-level matching to generate a comprehensive matching score for each candidate node. An adaptive threshold is set to dynamically control the retrieval range, achieving accurate and efficient context-aware knowledge retrieval.

[0058] Specifically, by using a pre-constructed fine-grained scenic area knowledge graph and a graph attention network algorithm, combined with the dynamic retrieval and integration of relevant landscape, historical, and cultural information based on tourists' real-time status, a personalized set of contextual knowledge is generated. This solves the core problem in the background technology that existing automated generation systems struggle to deeply integrate the unique historical and cultural knowledge of scenic areas, resulting in a lack of cultural accuracy and emotional resonance in promotional content. It achieves in-depth mining and precise adaptation of the unique value of scenic area anecdotes and folk customs, thereby ensuring that the generated copy is not only highly relevant to tourists' current spatial and temporal location and interests, but also conveys profound cultural connotations, effectively enhancing the attractiveness and value of the content.

[0059] S3. Analyze the quantitative representation of tourist status through reinforcement learning policy network, establish a mapping model from tourist characteristics to copywriting style control parameters, and automatically generate optimal style control parameters including cultural concentration coefficient, language style code, emotional tone vector, and information density threshold, providing accurate guiding parameters for the AIGC generation process.

[0060] Further, step S3 specifically includes:

[0061] Extract preference feature vectors, emotional state indicators, and cultural awareness parameters from the quantitative representation of tourist status to form standardized input features, and construct a multi-dimensional parameter space.

[0062] The multidimensional parameter space includes: a continuous-valued cultural concentration coefficient for quantifying the density of cultural elements, a discrete language style code selected from a preset style vocabulary, an emotional tone vector representing emotional tendency and intensity in the multidimensional emotional space, and an adjustable information density threshold for controlling the saturation of text information.

[0063] A dual-network architecture of Actor-Critic is constructed, in which the policy network (Actor) receives standardized feature vectors as state input and generates cultural concentration coefficient, language style code and sentiment tone vector through three parallel output layers; the value network (Critic) evaluates the expected reward of state-action pairs and uses tourist feature-parameter pairs of historical high-quality copywriting for supervised pre-training to establish basic mapping relationships and complete model initialization.

[0064] The trained model generates basic parameters based on real-time tourist characteristics, and dynamically adjusts them in combination with environmental factors such as time, weather, and holidays, outputting optimal control parameters that are highly adapted to the current scene.

[0065] The generated optimal control parameters are subjected to numerical boundary checks, logical consistency checks, and abnormal pattern recognition. When an abnormal output is detected, the system automatically switches to a rule-based backup parameter generation mode to ensure the stability of the AIGC generation process and records abnormal cases for subsequent model optimization.

[0066] The cultural concentration coefficient is generated using a time-decay-based weighted fusion algorithm. Specifically, based on tourists' interaction records with different cultural dimensions such as history, folklore, and art, the frequency and average depth of interaction for each dimension within a preset time window are statistically analyzed. Combined with preset dimension weight coefficients, the cultural concentration coefficient is output through weighted summation and normalization.

[0067] The generation of the emotional tone vector adopts multi-source emotional fusion technology, specifically: performing sentiment analysis on the social media texts posted by tourists to extract their multi-dimensional emotional features; at the same time, using an environmental sentiment analysis model, analyzing the current weather, season, and scene atmosphere to generate environmental sentiment influencing factors; weighted fusion of textual sentiment features and environmental sentiment factors, and vector projection and standardization in a preset sentiment space to form the final emotional tone vector.

[0068] The determination of the information density threshold adopts a multi-factor dynamic adjustment algorithm, which is as follows: matching a preset reading ability level according to the age characteristics of tourists, determining the screen information carrying parameters according to the terminal device type, and calculating the environmental interference coefficient according to the real-time environmental complexity (such as crowd density and weather conditions); inputting the above three types of parameters into the adjustment function, dynamically calculating and outputting the information density threshold adapted to the current comprehensive conditions.

[0069] Specifically, by using a reinforcement learning strategy network, an adaptive mapping model was constructed from the real-time status of tourists to the copywriting style control parameters. This model can dynamically generate optimal control parameters by comprehensively considering environmental factors. This solves the core defects of existing automated methods in the background technology, which rely on static templates or simple tags, resulting in content homogenization and a lack of in-depth personalization. It achieves precise quantification and dynamic control of the cultural concentration, language style, emotional tone, and information density of the copywriting, thereby ensuring that the generated promotional content closely matches each tourist's real-time preferences, emotional state, and specific tour context, significantly improving the individual adaptability and contextual awareness of the copywriting.

[0070] S4. By integrating the quantitative representation of tourist status, contextual knowledge set and optimal style control parameters, inputting into a multi-task generation model based on the Transformer architecture, and implementing a three-layer progressive decoding strategy through the attention mechanism in the model, outputting multiple versions of personalized candidate copy.

[0071] Further, step S4 specifically includes:

[0072] S41. The tourist state quantification representation vector, the graph embedding vector of the contextual knowledge set, and the optimal style control parameter vector are concatenated and positionally encoded to form a unified input sequence. Through the multi-head self-attention mechanism in the Transformer model, the dynamic semantic association and weight distribution between multi-source information are calculated and established, so as to build a joint representation that integrates personalized needs, domain knowledge and style instructions for the subsequent generation process.

[0073] S42. Based on joint representation, a three-layer progressive decoding strategy is adopted to generate semantically coherent initial copy that meets preset constraints. The first layer of the decoder generates a copy framework containing the theme, structure and core information points based on tourist characteristics. The second layer uses a knowledge gating attention mechanism to accurately inject relevant entities and attributes from the contextual knowledge set into the corresponding positions of the framework. The third layer iteratively optimizes the vocabulary, sentence structure and rhetoric of the text based on style control parameters.

[0074] S43. Based on the joint representation and the initial text after hierarchical decoding, by adjusting the temperature parameters of the decoding process and applying kernel sampling technology, the model is driven to generate multiple candidate text versions with diverse expressions but consistent content. Then, basic quality filtering is performed on the generated multiple versions, including grammatical correctness checks, basic logical consistency verification, and preset constraint compliance checks, to eliminate seriously unqualified texts and output a high-quality personalized candidate text set.

[0075] The preset constraints specifically include: cultural accuracy constraint: requiring that the historical facts, cultural allusions, etc. involved in the generated content must be consistent with the knowledge graph data; personalization adaptation constraint: requiring that the text content must be highly matched with the tourists' preferences and real-time context; style consistency constraint: requiring that the language style and emotional tone of the generated text must be consistent with the target set by the style control parameters; and information density constraint: requiring that the information content of each unit of text must meet the set information density threshold range.

[0076] The knowledge-gated attention mechanism introduces knowledge type gating vectors into cross-attention calculations to achieve differentiated fusion weight control for knowledge fragments of different categories.

[0077] Furthermore, by employing the multi-head self-attention mechanism in the Transformer model, dynamic semantic associations and weight distributions among multi-source information are calculated and established. Specifically, this involves concatenating the tourist state representation vector, contextual knowledge graph embedding vector, and style control parameter vector, and adding positional encoding to form a unified input sequence. Then, the query vector, key vector, and value vector are computed in parallel using the multi-head attention mechanism. Each attention head focuses on semantic associations in different dimensions; some attention heads focus on analyzing the matching relationship between tourist features and knowledge elements, some focus on the fit between style instructions and content frameworks, and others specifically handle the correspondence between spatiotemporal information and event descriptions. Through a soft attention weight matrix, the relevance scores between elements in the input sequence are dynamically calculated, enabling relevant knowledge fragments to automatically receive higher attention weights based on tourist preferences, while style parameters effectively guide the rhetorical direction of content generation. The resulting attention distribution not only reflects the semantic association strength of multi-source information but also preserves the uniqueness of information from different sources through residual connections and layer normalization, providing a comprehensive and accurate joint representation for subsequent decoding and generation.

[0078] Specifically, by integrating visitor status, contextual knowledge, and style control parameters through a Transformer-based multi-task generation model, and utilizing an attention mechanism to achieve three-layer progressive decoding, the system solves the core problems of inefficiency and lack of depth and personalized adaptation in traditional manual creation or template-based methods in background technology. Through knowledge-gated attention and style parameter guidance, the system can simultaneously ensure the cultural accuracy, personalization, and style consistency of the content. This enables the efficient and automatic generation of diverse, high-quality candidate texts while ensuring information accuracy and style adaptation, thereby solving both the efficiency bottleneck and the lack of cultural connotation in traditional content production, and significantly improving the output efficiency and overall quality of promotional content.

[0079] S5. Through the content evaluation model, the candidate texts of multiple versions are comprehensively scored from multiple dimensions. Combined with the format specifications of the target publishing channels, the candidate texts are restructured and format adapted using an adaptive conversion algorithm to generate the final content for different media.

[0080] Further, step S5 specifically includes:

[0081] A content evaluation model is used to evaluate candidate copy in parallel, including a cultural fit module, a personalized matching module, and a dissemination potential module. The semantic similarity between the copy and the contextual knowledge set, the fit between the copy features and the tourist state vector, and the dissemination potential of the copy on the target channel are calculated respectively, and the three-dimensional score vector of each copy is output.

[0082] Dynamic weighting coefficients are configured for the three evaluation dimensions based on the characteristics of the target channel and marketing needs. A comprehensive score is calculated through a weighted fusion algorithm, and tiered screening is performed.

[0083] The tiered screening strategy includes: removing copywriting with cultural fit below a preset safety threshold, sorting by comprehensive score in descending order, and finally selecting the top N copywriting as the preferred result to enter the next processing stage, where the value of N is dynamically adjusted according to the characteristics of the channel.

[0084] To address the technical specifications of different media channels, a transformation system combining a rule engine and a neural network is used to reconstruct candidate text. Multiple rounds of quality verification are then performed on the reconstructed content. Finally, the content and channel-specific metadata are standardized and encapsulated to generate the final content that can be directly published through the platform's API interface, along with the corresponding publication tracking identifier.

[0085] Specifically, calculating the semantic similarity between the text and the contextual knowledge set involves: using a pre-trained semantic encoding model to encode the candidate text and knowledge fragments such as cultural elements and historical allusions in the contextual knowledge set into high-dimensional semantic vectors; calculating the cosine similarity between the two vectors and combining them with the entity overlap and semantic relevance for weighted fusion to finally obtain a semantic similarity score in the 0-1 range.

[0086] The fit between text features and tourist state vectors specifically includes: extracting deep semantic features of candidate texts through a text encoder to form feature vectors, and then performing similarity matching between these vectors and tourist state quantification representation vectors in a multi-dimensional feature space. The matching process uses cosine similarity as the basic metric and combines it with the matching degree of tourist interest tags for weighted correction, outputting a personalized fit score.

[0087] The potential for copywriting to spread on target channels specifically includes: building a dissemination prediction model based on historical dissemination data and channel characteristics. This model analyzes the copywriting's emotional polarity, keyword popularity, sentence structure, readability index, and other characteristics, combined with the user profiles and behavioral characteristics of the target channels, to predict the expected sharing rate, forwarding rate, and interaction depth of the copywriting on the channels, and finally outputs a standardized dissemination potential score.

[0088] Based on the characteristics of the target channels and marketing needs, dynamic weight coefficients are configured for the three evaluation dimensions, including: for official promotional channels (such as official websites and navigation screens), a high weight is assigned to cultural fit (e.g., 0.5), a medium weight to personalized matching (e.g., 0.3), and a low weight to dissemination potential (e.g., 0.2); for social media channels (such as Weibo and Douyin), the weights of dissemination potential and personalized matching are appropriately increased, while the weight of cultural fit is decreased; for personalized recommendation scenarios (such as app push notifications), the weight of personalized matching is significantly increased, forming a dynamically adjustable weight configuration strategy.

[0089] A conversion system combining a rule engine and neural networks is used to reconstruct candidate text. Specifically, this includes: processing structured format requirements through the rule engine, such as automatically adjusting text length to adapt to the character limits of different channels, optimizing paragraph structure to conform to platform reading habits, processing specific format tags such as topic tags and hyperlinks, and injecting necessary metadata; and performing intelligent content reconstruction while preserving semantics through the neural network module, including attention-based summary generation, sentence style transfer optimization, and intelligent adaptation of multimodal elements, ensuring that the converted content fully complies with the technical specifications and user experience requirements of the target channel while maintaining the core semantics.

[0090] Specifically, by constructing a multi-dimensional content evaluation model and combining it with channel-aware adaptive conversion technology, the system performs refined screening and reconstruction of the generated multi-version candidate copy. This solves the problems of inconsistent content quality and difficulty in meeting the diverse format specifications of different media channels in existing automated generation solutions. By conducting quantitative evaluation and dynamic weighting from multiple perspectives such as cultural accuracy, personalization matching, and dissemination potential, and combining rule and intelligent reconstruction technology, the system ensures that the final published copy not only has cultural depth and personalized adaptability in content, but also perfectly matches the dissemination characteristics and technical requirements of different channels in form. This achieves a closed-loop quality process from automated generation to precise and channel-specific delivery, significantly improving the implementation effect and dissemination efficiency of promotional copy.

[0091] S6. By deploying a tracking module on the content display terminal, real-time data on user exposure, interaction and conversion behavior of the generated final content is collected. Attribution analysis model is used to correlate user behavior data with corresponding generation control parameters and visitor characteristics to form a traceable generation effect evaluation record.

[0092] Further, step S6 specifically includes:

[0093] By deploying standardized tracking components on the content display terminal, the system collects complete behavioral data sequences of users on the final content in real time, including structured raw data such as exposure trigger time, interaction action type, and conversion path nodes. The collected raw data is then cleaned by noise reduction, timestamp alignment, and format standardization to establish a unified set of behavioral data records.

[0094] Establish a reverse correlation mapping from user behavior to content generation process, bind behavioral data to final content through unique content identifiers, trace the correlation of style control parameter set and extracted visitor state feature vector, and adopt a feature alignment module based on attention mechanism to accurately align data from different sources in the time dimension to form a spatiotemporally consistent multi-source dataset.

[0095] We construct a deep attribution analysis model based on the Transformer architecture. By learning the influence weights of different generation parameters in the user behavior sequence, we establish a causal inference relationship from control parameters to behavioral results. We use a multi-head attention mechanism to analyze the interaction effect between parameters and output an attribution analysis matrix that includes the contribution of each style parameter. This quantitatively demonstrates the influence of parameters such as cultural concentration coefficient and emotional tone vector on user behavior.

[0096] The attribution analysis results are integrated with the raw data to generate a structured evaluation record that includes basic behavioral indicators, parameter influence distribution, anomaly pattern identification, and optimization suggestions.

[0097] The attribution analysis model employs a feature contribution calculation method based on Shapley values ​​to ensure the fairness and interpretability of parameter impact assessment; the assessment record contains complete data traceability information, supporting a complete path query from the final behavior back to the original generated parameters.

[0098] Specifically, by deploying tracking and deep attribution analysis models, the system establishes the capability for reverse tracing and causal correlation analysis from user behavior data to content generation parameters. This addresses the pain points of existing automated systems in the background technology, which lack a feedback loop, hindering dynamic optimization and making it difficult to evaluate the effectiveness of content generation. By precisely linking and quantifying user exposure, interaction, and conversion behaviors with specific copywriting style parameters and visitor characteristics, the system achieves for the first time a refined evaluation of personalized copywriting generation effects and a deep analysis of influencing factors. This provides data-driven decision-making support for the continuous iteration and optimization of the generation model, forming a complete closed loop from content generation and publication to effect evaluation and optimization. This effectively enhances the system's adaptive evolution capabilities and the long-term effectiveness of promotional content.

[0099] S7. Based on the constructed reinforcement learning model infrastructure, the generated performance evaluation data is used as training samples to optimize the policy gradient of the mapping model of the generation control parameters. At the same time, new data is used periodically to incrementally train the multi-task generation model to form a complete closed loop.

[0100] Further, step S7 specifically includes:

[0101] A reinforcement learning environment is constructed using the generated attribution analysis results. The initialized Actor-Critic network is used as the agent to be optimized. By calculating the causal relationship strength between user interaction data and each generation parameter, a multi-level reward function is established, including basic conversion rate reward, content quality score reward and dissemination innovation reward. The proximal policy optimization algorithm is used to iteratively update the policy network parameters to achieve online continuous optimization of the generation control parameter mapping policy.

[0102] High-quality new samples are selected from the evaluation records, including content data with a conversion rate higher than the threshold and its corresponding tourist characteristics and generation parameters. A course learning strategy is used to incrementally train the multi-task generation model. During the training process, the updated content of the knowledge graph is integrated synchronously. New cultural elements and historical allusions are embedded into the parameter space of the generation model through the knowledge adaptation layer to maintain the timeliness and accuracy of content generation capabilities.

[0103] Establish a system performance monitoring index system, including generation quality stability index, personalized matching accuracy index and computing resource efficiency index, design an anomaly detection mechanism, and automatically trigger the diagnostic process when performance anomalies are detected, implement model rollback or retraining operations, and ensure the coordinated evolution and stable operation of all modules of the system.

[0104] The weights of each layer of the multi-level reward function are dynamically adjusted according to the training stage. In the early stage, the focus is on basic conversion indicators; in the middle stage, the balance between quality and innovation is achieved; and in the later stage, the focus is on the stability of long-term effects. The incremental training adopts an elastic sample buffer management strategy. The performance monitoring establishes a real-time early warning and automatic recovery mechanism.

[0105] The multi-level reward function is expressed as: R(t) = α(t)⋅Rc + β(t)⋅Rq + γ(t)⋅Ri; where Rc represents the basic conversion rate reward, reflecting the proportion of user conversion behavior; Rq represents the content quality score reward, which comes from the output of the content evaluation model; Ri represents the dissemination innovation reward, which combines the dissemination depth (level) and breadth (forwarding ratio); α(t), β(t), and γ(t) are dynamic weights that change over time and are used to adjust the importance of different reward components at the current stage.

[0106] A proximal policy optimization algorithm is used to iteratively update the policy network parameters. Specifically, this includes: generating a policy distribution based on the current policy network parameters; collecting state-action-reward sequence data through importance sampling; calculating the probability ratio between the old and new policies; and using a shearing objective function to limit the policy update magnitude to ensure training stability; introducing an entropy regularization term to encourage policy exploration and prevent premature convergence to local optima; and constructing an advantage function estimator and using a generalized advantage estimation algorithm to balance immediate rewards and long-term returns to achieve gradient direction optimization and online continuous improvement of the policy mapping to the generated control parameters.

[0107] Step S7 builds upon the model framework established in S3, utilizing the actual data generated after execution to complete iterative optimization of the model and achieve a closed-loop improvement in the overall system performance.

[0108] Specifically, by constructing an adaptive optimization closed loop based on reinforcement learning and incremental training, user behavior feedback and content evaluation results are transformed into a direct driving force for model iteration. This solves the fundamental problem in the background technology where existing methods lack a data-driven continuous learning mechanism, leading to system rigidity and an inability to adapt to the long-term evolution of tourist preferences and the market environment. By utilizing actual interaction data to perform online optimization and incremental updates to the strategy network and generation model, the transformation from static rule execution to dynamic intelligent evolution is achieved. This ensures that the copywriting generation strategy can continuously track and respond to changes in user preferences and knowledge updates, thereby continuously improving the personalization accuracy, cultural timeliness, and overall dissemination effectiveness of the content in long-term operation. Ultimately, this builds an intelligent content generation ecosystem with self-evolutionary capabilities.

[0109] Example 2

[0110] Please see Figure 4 This embodiment provides an AIGC-based automatic generation system for personalized promotional copy for tourist attractions, which is used to implement an AIGC-based method for automatically generating personalized promotional copy for tourist attractions, including:

[0111] The tourist feature perception module collects tourist location, environment, interaction and social data in real time through multi-source IoT sensors, and uses multimodal deep neural networks for alignment and fusion to directly output a quantitative feature vector containing spatiotemporal preferences, interest intensity and emotional state, providing accurate tourist state representation for subsequent processing.

[0112] The contextual knowledge retrieval module, based on a pre-built fine-grained scenic area knowledge graph, combines tourists' real-time location, time, and interest tags, and uses a graph neural network semantic retrieval algorithm to output a personalized knowledge set that is highly matched with the current tourist context, including structured information such as landscape descriptions, historical anecdotes, and cultural elements.

[0113] The style parameter generation module uses an Actor-Critic reinforcement learning network to map tourist feature vectors into style control parameters such as cultural concentration coefficient, language style code, emotional tone vector, and information density threshold, and outputs the optimal parameter combination to guide copywriting generation.

[0114] The content generation optimization module integrates tourist characteristics, contextual knowledge, and style parameters. It implements a three-layer progressive decoding through a Transformer-based multi-task generation model, outputting a set of personalized candidate texts with multiple versions that have undergone quality filtering, ensuring the accuracy, adaptability, and style consistency of the content.

[0115] The multi-dimensional assessment and adaptation module uses a three-dimensional assessment model of cultural fit, personalization matching and dissemination potential to score candidate copy in parallel, and intelligently reconstructs the format in combination with the target channel specifications to output the final content that is ready for the channel and the corresponding release tracking mark.

[0116] The effect monitoring and attribution module collects user exposure, interaction and conversion behavior data in real time through the tracking system, establishes causal relationships between behavior and generated parameters using a deep attribution analysis model, and outputs a structured evaluation record that includes parameter impact analysis and optimization suggestions.

[0117] The closed-loop optimization and evolution module constructs a reinforcement learning environment based on performance evaluation data, optimizes the style parameter generation model online, and performs incremental training on the content generation model, outputting continuously iterative model parameters and performance indicators to achieve the system's self-evolution and capability improvement.

[0118] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for automatically generating personalized promotional copy for tourist attractions based on AIGC, characterized by: Includes the following steps: S1. By deploying IoT sensors in the scenic area and user terminals, heterogeneous tourist data is acquired in real time. Multimodal deep neural networks are used to align and fuse the heterogeneous data to obtain a quantitative representation of the tourist status. S2. By using a pre-constructed fine-grained scenic area knowledge graph, combined with the spatiotemporal coordinates and interest tags in the quantitative representation of real-time tourist status, a semantic retrieval algorithm based on graph neural networks is adopted to generate a personalized set of contextual knowledge. The pre-constructed fine-grained scenic area knowledge graph includes scenic area landscape entities, historical events, cultural elements, and their spatiotemporal semantic relationships. S3. Analyze the quantitative representation of tourist status through reinforcement learning policy network, establish a mapping model from tourist characteristics to copywriting style control parameters, and automatically generate the optimal style control parameters. S4. Integrate the quantitative representation of tourist status, contextual knowledge set and optimal style control parameters, input the multi-task generation model based on the Transformer architecture, and output multiple versions of personalized candidate copy by implementing a three-layer progressive decoding strategy. S5. Through the content evaluation model, the candidate texts of multiple versions are comprehensively scored from multiple dimensions. Combined with the format specifications of the target publishing channels, the adaptive conversion algorithm is used to restructure and adapt the format to generate the final content for different media. S6. Through the tracking module, real-time data on user exposure, interaction and conversion behavior of the generated final content is collected. Attribution analysis model is used to correlate user behavior data with corresponding generation control parameters and visitor characteristics to form a traceable generation effect evaluation record. S7. Based on the constructed reinforcement learning model infrastructure, the generated performance evaluation data is used as training samples to optimize the policy gradient of the mapping model of the generation control parameters, while the multi-task generation model is incrementally trained.

2. The method for automatically generating personalized promotional copy for tourist attractions based on AIGC according to claim 1, characterized in that: Step S1 specifically includes: Tourist spatiotemporal trajectory data and environmental parameter data are collected through cameras, WiFi probes and environmental sensors deployed in the scenic area; tourist interaction behavior data is collected through mobile terminal application interfaces; and text, images and location check-in data posted by tourists are collected through authorized open social media platform interfaces. The collected heterogeneous data is cleaned, denoised, and standardized in format, including coordinate unification and path smoothing for trajectory data, time synchronization and dimensional normalization for environmental parameter data, word segmentation, stop word removal, and sentiment polarity labeling for text data; and feature extraction and scene classification labeling for image data. The preprocessed multimodal data is input into a multimodal deep neural network based on an attention mechanism. The network includes a text encoder, an image encoder, and a temporal data encoder. The alignment and weighted fusion of feature vectors from different modalities are achieved through a cross-modal attention layer to generate a unified fused feature representation. Based on a unified fusion feature representation, feature components of three dimensions—spatiotemporal preference intensity, interest label weight, and emotional state value—are calculated separately through a fully connected neural network layer to generate a dynamic feature vector with fixed dimensions, which serves as a quantitative representation of tourist status.

3. The method for automatically generating personalized promotional copy for tourist attractions based on AIGC according to claim 1, characterized in that: Step S2 specifically includes: S21. Extract the real-time geographic location coordinates, timestamps, and interest tag set from the quantitative representation of tourist status, and transform them into a knowledge graph query language. Construct a composite query graph pattern that combines spatiotemporal constraints and semantic interest constraints. The spatiotemporal constraints limit the temporal validity and spatial proximity of nodes, and the interest constraints are based on the semantic similarity between tourist interest tags and knowledge nodes. S22. Input the composite query into the pre-constructed fine-grained scenic area knowledge graph, use a retrieval algorithm based on graph attention network to calculate the matching degree between the query conditions and the nodes and relationships in the knowledge graph, traverse the nodes in the graph that meet the conditions, dynamically expand the retrieval scope through the graph propagation mechanism, and finally extract the connected knowledge subgraph containing relevant entities, relationships and attributes. S23. Perform structured parsing on the retrieved knowledge subgraph, extract information fragments, sort the knowledge fragments by importance according to the weight of tourist interest tags, and adaptively adjust the knowledge content in combination with real-time environmental parameters to generate a structured personalized contextual knowledge set.

4. The method for automatically generating personalized promotional copy for tourist attractions based on AIGC according to claim 3, characterized in that: The pre-constructed fine-grained scenic area knowledge graph includes: extracting entities, attributes, and relationships from official scenic area materials, local chronicles, academic literature, and authoritative tour guide texts to establish a basic graph structure; performing knowledge completion and semantic relationship enhancement based on tourist-generated content and expert-annotated data; and mapping entities and relationships to a low-dimensional vector space through a graph embedding algorithm to perform semantic similarity calculation and graph neural network operations.

5. The method for automatically generating personalized promotional copy for tourist attractions based on AIGC according to claim 1, characterized in that: Step S3 specifically includes: Extract preference feature vectors, emotional state indicators, and cultural awareness parameters from the quantitative representation of tourist status to form standardized input features, and construct a multi-dimensional parameter space. The multidimensional parameter space includes: a continuous-valued cultural concentration coefficient for quantifying the density of cultural elements, a discrete language style code selected from a preset style vocabulary, an emotional tone vector representing emotional tendency and intensity in the multidimensional emotional space, and an adjustable information density threshold for controlling the saturation of text information. A dual-network architecture of Actor-Critic is constructed, in which Actor receives standardized feature vectors as state input and generates cultural concentration coefficient, language style code and sentiment tone vector through three parallel output layers; Critic evaluates the expected reward of state-action pairs and uses tourist feature-parameter pairs of historical high-quality copywriting for supervised pre-training to establish basic mapping relationships and complete model initialization. The trained model generates basic parameters based on real-time tourist characteristics, dynamically adjusts them in combination with environmental factors, and outputs optimal control parameters that are highly adapted to the current scene. The generated optimal control parameters are subjected to numerical boundary checks, logical consistency checks, and abnormal pattern recognition. When an abnormal output is detected, the system automatically switches to a rule-based backup parameter generation mode.

6. The method for automatically generating personalized promotional copy for tourist attractions based on AIGC according to claim 1, characterized in that: Step S4 specifically includes: S41. The tourist state quantification representation vector, the graph embedding vector of the contextual knowledge set, and the optimal style control parameter vector are concatenated and positionally encoded to form a unified input sequence. Through the multi-head self-attention mechanism in the Transformer model, the dynamic semantic association and weight distribution between multi-source information are calculated and established, so as to build a joint representation that integrates personalized needs, domain knowledge and style instructions for the subsequent generation process. S42. Based on joint representation, a three-layer progressive decoding strategy is adopted to generate semantically coherent initial copy that meets preset constraints. The first layer of the decoder generates a copy framework containing the theme, structure and core information points based on tourist characteristics. The second layer uses a knowledge gating attention mechanism to accurately inject relevant entities and attributes from the contextual knowledge set into the corresponding positions of the framework. The third layer iteratively optimizes the vocabulary, sentence structure and rhetoric of the text based on style control parameters. S43. Based on the joint representation and the initial text after hierarchical decoding, by adjusting the temperature parameters of the decoding process and applying kernel sampling technology, the model is driven to generate multiple candidate text versions with diverse expressions but consistent content. Then, basic quality filtering is performed on the generated multiple versions to output a high-quality personalized candidate text set.

7. The method for automatically generating personalized promotional copy for tourist attractions based on AIGC according to claim 1, characterized in that: Step S5 specifically includes: A content evaluation model is used to evaluate candidate copy in parallel, including a cultural fit module, a personalized matching module, and a dissemination potential module. The semantic similarity between the copy and the contextual knowledge set, the fit between the copy features and the tourist state vector, and the dissemination potential of the copy on the target channel are calculated respectively, and the three-dimensional score vector of each copy is output. Dynamic weighting coefficients are configured for the three evaluation dimensions based on the characteristics of the target channel and marketing needs. A comprehensive score is calculated through a weighted fusion algorithm, and tiered screening is performed. The tiered screening strategy includes: removing copywriting with cultural fit below a preset safety threshold, sorting by comprehensive score in descending order, and finally selecting the top N copywriting as the preferred result to enter the next processing stage, where the value of N is dynamically adjusted according to the characteristics of the channel. To address the technical specifications of different media channels, a transformation system combining a rule engine and a neural network is used to reconstruct candidate text. Multiple rounds of quality verification are then performed on the reconstructed content. Finally, the content and channel-specific metadata are standardized and encapsulated to generate the final content that can be directly published through the platform's API interface, along with the corresponding publication tracking identifier.

8. The method for automatically generating personalized promotional copy for tourist attractions based on AIGC according to claim 1, characterized in that: Step S6 specifically includes: By deploying standardized tracking components on the content display terminal, the system collects complete behavioral data sequences of users on the final content in real time. The collected raw data is then denoised, cleaned, timestamp aligned, and format standardized to establish a unified set of behavioral data records. Establish a reverse correlation mapping from user behavior to content generation process, bind behavioral data to final content through unique content identifiers, trace the correlation of style control parameter set and extracted visitor state feature vector, and adopt a feature alignment module based on attention mechanism to accurately align data from different sources in the time dimension to form a spatiotemporally consistent multi-source dataset. We construct a deep attribution analysis model based on the Transformer architecture. By learning the influence weights of different generation parameters in the user behavior sequence, we establish a causal inference relationship from control parameters to behavioral results. We use a multi-head attention mechanism to analyze the interaction effect between parameters and output the degree of influence of user behavior. The attribution analysis results are integrated with the raw data to generate a structured evaluation record that includes basic behavioral indicators, parameter influence distribution, anomaly pattern identification, and optimization suggestions.

9. The method for automatically generating personalized promotional copy for tourist attractions based on AIGC according to claim 1, characterized in that: Step S7 specifically includes: A reinforcement learning environment is constructed using the generated attribution analysis results. The initialized Actor-Critic network is used as the agent to be optimized. By calculating the causal relationship strength between user interaction data and each generation parameter, a multi-level reward function is established, including basic conversion rate reward, content quality score reward and dissemination innovation reward. The proximal policy optimization algorithm is used to iteratively update the policy network parameters to achieve online continuous optimization of the generation control parameter mapping policy. High-quality new samples are selected from the evaluation records, including content data with a conversion rate higher than the threshold and its corresponding tourist characteristics and generation parameters. A course learning strategy is used to incrementally train the multi-task generation model. During the training process, the updated content of the knowledge graph is integrated synchronously. New cultural elements and historical allusions are embedded into the parameter space of the generation model through the knowledge adaptation layer. Establish a system performance monitoring index system, including generation quality stability index, personalized matching accuracy index and computing resource efficiency index, and design an anomaly detection mechanism. When a performance anomaly is detected, the diagnostic process is automatically triggered, and model rollback or retraining operations are implemented.

10. An AIGC-based automatic generation system for personalized promotional copy for tourist attractions, applied to the AIGC-based automatic generation method for personalized promotional copy for tourist attractions as described in any one of claims 1-9, characterized in that: include: The tourist feature perception module collects tourist location, environment, interaction and social data in real time through multi-source IoT sensors, and uses multimodal deep neural networks to align and fuse the data, outputting a quantitative feature vector containing spatiotemporal preferences, interest intensity and emotional state. The contextual knowledge retrieval module, based on a pre-built fine-grained scenic area knowledge graph, combines tourists' real-time location, time, and interest tags, and uses a graph neural network semantic retrieval algorithm to output a personalized knowledge set that is highly matched with the current tourist context. The style parameter generation module uses an Actor-Critic reinforcement learning network to map tourist feature vectors into style control parameters such as cultural concentration coefficient, language style code, emotional tone vector, and information density threshold, and outputs the optimal parameter combination to guide copywriting generation. The content generation optimization module integrates tourist characteristics, contextual knowledge, and style parameters. It implements three-layer progressive decoding through a Transformer-based multi-task generation model and outputs a set of personalized candidate texts with multiple versions that have undergone quality filtering. The multi-dimensional assessment and adaptation module uses a three-dimensional assessment model of cultural fit, personalization matching and dissemination potential to score candidate copy in parallel. It also intelligently reconstructs the format in conjunction with the target channel specifications and outputs the final content that is ready for the channel and the corresponding release tracking identifier. The effect monitoring and attribution module collects user exposure, interaction and conversion behavior data in real time through the tracking system, establishes causal relationships between behavior and generated parameters using a deep attribution analysis model, and outputs a structured evaluation record that includes parameter impact analysis and optimization suggestions. The closed-loop optimization and evolution module constructs a reinforcement learning environment based on the effect evaluation data, optimizes the style parameter generation model online, and performs incremental training on the content generation model, outputting continuously iterative model parameters and performance indicators.

Images