Urban document travel big operation cluster platform based on artificial intelligence

Abstract

The invention relates to the technical field of artificial intelligence, and discloses an artificial intelligence-based urban text travel big operation cluster platform, which comprises the steps of performing data capture on a government text travel database, scenic spot IOT equipment and social media UGC through distributed crawlers; identifying and rejecting data exceeding a reasonable range in the captured data based on a statistical threshold method; mapping data of different dimensions in the captured data to an interval of [0, 1] through a normalization model; carrying out characteristic value calculation on the grabbed data according to a noise reduction anomaly degree model; and an analysis engine is locally deployed in the operation cluster platform. According to the method, a government travel database, scenic spot IoT equipment and social media UGC data are integrated through distributed crawlers, and efficient cleaning, standardization and feature extraction of multi-source heterogeneous data are realized in combination with a statistical threshold method, a normalization model and a noise reduction anomaly model, so that dimensional differences are eliminated, and data quality and analysis precision are improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an urban cultural tourism operation cluster platform based on artificial intelligence. Background Art

[0002] With the acceleration of urbanization and the rapid development of the cultural and tourism industry, the efficient management of urban cultural and tourism resources and the optimization of tourist experience have become important issues.

[0003] Traditional cultural and tourism management systems often rely on single data sources (such as government databases or IoT devices in scenic spots) and lack the deep integration of heterogeneous data from multiple sources (such as user-generated content (UGC) from social media). This leads to severe information silos and hinders comprehensive understanding of cultural and tourism resource dynamics. Furthermore, existing technologies lack intuitive three-dimensional visualization tools, making it difficult for managers and visitors to understand resource status in real time, resulting in a poor interactive experience. Therefore, an intelligent platform integrating artificial intelligence, multi-source data fusion, and digital twin technologies is urgently needed to enhance the overall scheduling capabilities of cultural and tourism resources, improve emergency response efficiency, and enhance visitor service.

[0004] An existing patent discloses a self-media advertising platform and method for urban culture and tourism experiences (publication number CN119671766A). This technology, which belongs to the field of self-media advertising, uses a defined time interval to detect the corresponding transmission bandwidth of all actively transmitted map data messages. The congestion level of the mobile communication network is determined based on the transmission bandwidth and the transmission threshold. This existing technology suffers from inefficient data standardization methods, insufficient anomaly detection capabilities, and inability to meet real-time analysis requirements. In particular, during sudden congestion or security incidents, emergency response relies on manual experience, resulting in delayed decision-making and inaccurate resource allocation. Summary of the Invention

[0005] In order to solve existing technical problems, the present invention provides an urban cultural tourism operation cluster platform based on artificial intelligence, which solves the problems in the above-mentioned background technology.

[0006] To solve the above technical problems, according to one aspect of the present invention, more specifically, an artificial intelligence-based urban cultural tourism operation cluster platform includes: Use distributed crawlers to capture data from government cultural and tourism databases, scenic area IoT devices, and social media UGC; Based on the statistical threshold method, data outside the reasonable range is identified and eliminated from the captured data. The data of different dimensions in the captured data are mapped to the interval [0, 1] through the normalization model. The characteristic value of the captured data is then calculated based on the noise reduction anomaly model. An analysis engine is deployed locally within the operations cluster platform. Using a pre-trained model in one of its sub-engines, it analyzes urban cultural and tourism information in real time and identifies sudden congestion and unusual personnel incidents at urban tourist attractions. Another sub-engine then matches resource supply and demand, enabling the passive allocation of security personnel to urban tourist attractions. Build a 3D visualization map of urban cultural and tourism resources based on digital twin technology. This 3D visualization map is used for user interaction terminals. Based on the user's real-time location, preferences, and resource status, the optimal tour route is generated through graph neural networks; security personnel at urban tourist attractions are proactively allocated based on actual conditions.

[0007] Furthermore, the normalization model is used to standardize the data captured by the distributed crawler, so: ; Where, Indicates the data after normalization, which includes the crowding coefficient; Indicates real-time captured data, including the current flow of people at attractions, restaurants, or transportation hubs; Indicates the minimum amount of data; Indicates the maximum data size.

[0008] Furthermore, the pre-trained model recognizes event entities in social media UGC texts, and the steps are as follows: 1) 5,000 texts were manually annotated, including the event type and sentiment polarity of positive or negative; 2) Fine-tune the weights during pre-training, and then use the pre-trained model to assign sentiment scores to the text about the marked events that appear at the scenic spots in the text posted by social media user-generated content.

[0009] Furthermore, the other sub-engine calculates the conditional coefficient for passively assigning security personnel to the area based on the current crowd conditions at the tourist attraction, restaurant, or transportation hub, and the evaluation of the current area on social media, as follows: ; Where, The conditional coefficient indicating the platform's need to passively allocate security personnel to the attraction, restaurant, or transportation hub; Indicates the crowding coefficient of the current tourist attraction, restaurant or transportation hub; Indicates the degree of negative evaluation of the current area by social media.

[0010] Furthermore, when When , it means that the platform needs to deploy additional security personnel to the inspected scenic spots, restaurants or transportation hubs; when When , it means that the platform does not need to deploy additional security personnel to the inspected scenic spots, restaurants or transportation hubs; when , it means that the platform can dispatch security personnel to the detected scenic spots, restaurants or transportation hubs.

[0011] Furthermore, the digital twin technology is used to construct a three-dimensional visualization map of urban cultural and tourism resources, including: 1) Using the Unity 3D engine to build 3D scenes based on real-time data from the government’s cultural and tourism database and scenic area IoT devices, and using Blender to repair model details; 2) Synchronize real-time passenger flow and traffic status data to the Internet access platform through the MQTT protocol.

[0012] Furthermore, the construction of the graph neural network includes: 1) Consider scenic spots, restaurants, and transportation hubs as nodes; consider the distance, walking time, and theme similarity between nodes as edges; 2) Node popularity and the text sentiment scores of users’ evaluations of each node are used as node attributes; 3) The real-time congestion level on the edge is used as the edge weight; 4) Based on the above data, a graph neural network is constructed for the scenic spot, and the tour path is optimized by combining historical tourist trajectory data.

[0013] The present invention provides an artificial intelligence-based urban cultural tourism operation cluster platform. Compared with the existing technology, this method has the following effects: 1. This invention integrates government cultural and tourism databases, scenic area IoT devices, and social media UGC data through distributed crawlers, and combines statistical threshold methods, normalization models, and noise reduction anomaly models to achieve efficient cleaning, standardization, and feature extraction of multi-source heterogeneous data, eliminate dimensional differences, and improve data quality and analysis accuracy.

[0014] 2. The present invention identifies sudden congestion and abnormal events in real time based on a pre-trained model, and quantifies the risk level through a conditional coefficient formula, thereby achieving active and passive precise allocation of security personnel, avoiding waste or shortage of resources, and significantly improving emergency response efficiency.

[0015] 3. This invention uses graph neural networks to construct a dynamic topology structure, combines node popularity, real-time congestion levels, and user preferences to generate optimal tour routes, reduce congestion, and improve tourist efficiency; the iterative optimization of historical trajectory data further enhances the adaptability of route recommendations.

[0016] 4. This invention builds a high-precision three-dimensional map based on digital twin technology, synchronizes real-time passenger flow and traffic data through the MQTT protocol, provides managers and tourists with a panoramic view, supports real-time monitoring and decision-making, and improves data transparency and interactive experience.

[0017] 5. The present invention uses a pre-trained model to accurately identify event types and sentiment polarity in social media UGC by fine-tuning weights, quantifying the degree of negative evaluation, helping managers to quickly respond to public opinion, optimize service strategies, and improve tourist satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a flow chart of the present invention; Figure 2 is a relationship diagram between the condition coefficient m and the crowding coefficient c in the present invention; Figure 3 is a relationship diagram between the condition coefficient m and the negative evaluation degree a in the present invention; Figure 4 This is a relationship diagram among the condition coefficient m, crowding coefficient c, and negative evaluation degree a in the present invention. DETAILED DESCRIPTION

[0019] In order to make the technical solution of the present invention clearer, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] Example 1 like Figure 1As shown, according to one aspect of the present invention, an artificial intelligence-based urban cultural and tourism operation cluster platform is provided, including: data capture of government cultural and tourism databases, scenic area IoT devices, and social media UGC through distributed crawlers; identification and elimination of data beyond a reasonable range in the captured data based on a statistical threshold method; and mapping data of different dimensions in the captured data to the [0, 1] interval through a normalization model; and then calculating the characteristic values of the captured data based on a noise reduction anomaly model; an analysis engine is locally deployed in the operation cluster platform, and the pre-trained model in one of the sub-engines is used to analyze urban cultural and tourism information in real time, and identify sudden congestion and abnormal personnel events at urban tourist attractions; and then the supply and demand of resources are matched through another sub-engine, and security personnel are passively allocated to urban tourist attractions; a three-dimensional visualization map of urban cultural and tourism resources is constructed based on digital twin technology, and the three-dimensional visualization map is used for user interaction terminals; the optimal tour path is generated through a graph neural network based on the user's real-time location, preferences and resource status; and security personnel are actively allocated to urban tourist attractions according to actual conditions. This implementation builds the overall platform architecture. Using distributed crawlers, it integrates government cultural and tourism databases, scenic area IoT devices, and user-generated content (UGC) data from social media. Statistical thresholding and normalization models are used to clean and standardize the data. The pre-trained model in the analytics engine then identifies emergencies (such as congestion and abnormal events) in real time. Digital twin technology is then used to generate a 3D visualization map. A graph neural network dynamically generates optimal tour routes based on user location, preferences, and resource status, while proactively allocating security resources.

[0021] Realize real-time fusion and intelligent analysis of multi-source data to improve the global dispatch capability of urban cultural and tourism resources; enhance emergency response efficiency through a security allocation mechanism that combines active and passive security; and significantly enhance the visitor experience through 3D visualization and path optimization functions.

[0022] Example 2 like Figure 1 As shown, the normalization model is used to standardize the data captured by the distributed crawler, so: ; Where, Indicates the data after normalization, which includes the crowding coefficient; Indicates real-time captured data, including the current flow of people at attractions, restaurants, or transportation hubs; Indicates the minimum amount of data; Indicates the maximum data size.

[0023] Among them, the crowd congestion coefficient of a certain scenic spot is calculated. The real-time captured traffic data of the scenic spot is obtained. (people / 10 minutes). The minimum traffic volume of the attraction in that month is (people / 10 minutes). The maximum number of visitors to the scenic spot in the month is taken (person / ten minutes), then: ; From the above calculations, we can know that after the normalization of the captured data, the crowding coefficient of the scenic spot in the current state is Among them, the larger the crowd congestion coefficient is, the more people there are at the attraction, and the more security personnel need to be dispatched to the attraction to maintain order.

[0024] This embodiment is used to eliminate data dimension differences, ensure the comparability of indicators such as the crowding coefficient, provide consistent input for subsequent analysis, and enhance the stability and prediction accuracy of the model.

[0025] Example 3 like Figure 1 As shown in the figure, the pre-trained model recognizes event entities in social media UGC texts. The steps are: 1) 5,000 texts were manually annotated, including event type and positive / negative sentiment. Based on these 5,000 manually annotated social media texts, a pre-trained model was fine-tuned to identify event types (e.g., congestion, safety incidents) and sentiment polarity (positive / negative), and sentiment scores were used to quantify users' evaluations of attractions.

[0026] 2) Fine-tune the weights during pre-training, and then use the pre-trained model to assign sentiment scores to text posted on social media about events related to tourist attractions. This accurately captures real-time user sentiment and event dynamics, helping managers quickly identify potential issues (such as negative public opinion) and adjust operational strategies in a timely manner, improving service quality and visitor satisfaction.

[0027] The pre-trained model calculates the negative rating of the attraction based on the negative vocabulary of the attraction in social media UGC, which is: ; Where, Indicates the degree of negative evaluation of tourist attractions in social media UGC by the pre-trained model; Indicates the total effective vocabulary size collected by the pre-training model (all words that fall into the 5000 manually annotated texts are effective words); Words that express negative emotions, such as "not worth it," "sad," and "boring"; Words that express positive emotions, such as "beautiful scenery", "fresh", and "passionate". represents the weight used for fine-tuning on pre-training, which is used to control the distribution of neutral words, where .

[0028] For example, the total effective vocabulary size collected by the pre-training model is Among them, the vocabulary with obvious negative emotions is (), the vocabulary with obvious positive emotions is taken (pieces), and , , then we have: ; From the above conclusions, we can know that the negative evaluation degree obtained by the pre-training model after collecting social media UGC data from a certain scenic spot is .

[0029] Example 4 like Figure 1 As shown, another sub-engine in the operation cluster platform calculates the conditional coefficients for passively assigning security personnel to an area based on the current crowd conditions at a tourist attraction, restaurant, or transportation hub, and social media reviews of the area. The coefficients are: ; Where, The conditional coefficient indicating the platform's need to passively allocate security personnel to the attraction, restaurant, or transportation hub; Indicates the crowding coefficient of the current tourist attraction, restaurant or transportation hub; Indicates the degree of negative evaluation of the current area by social media.

[0030] The above formula is an empirical analysis conducted to solve practical problems in enterprises. The data obtained in turn and the characteristic relationships between the data are the external manifestations of the empirical formula. The reasoning process is as follows: 1) The conditional coefficient for the platform to passively allocate security personnel to the scenic spot, restaurant or transportation hub The formula is fitted.

[0031] Among them, the platform needs to perform the conditional coefficient for passive allocation of security personnel to the scenic spot, restaurant or transportation hub , which can be represented by the number of devices with abnormal communication status in the sample energy storage cabinet.

[0032] For example, if data from 100 energy storage cabinet samples is collected and manual evaluation shows that the degree of communication anomaly in a certain energy storage cabinet exceeds the data in the other 50 samples, then the condition coefficient g for the energy storage cabinet to call the backup power module is 50%.

[0033] Other data are calculated and fitted using data collected by the device itself.

[0034] 2) Establish a mathematical model for the relationship between the condition coefficient m and the crowding coefficient c (such as Figure 2 As shown in the figure, the red dots are the distribution of the 100 samples collected), then: (Formula 1) In the above formula 1, k represents an empirical constant for adjusting the sensitivity of the above model.

[0035] 3) Establish a mathematical model for the relationship between the condition coefficient m and the negative evaluation degree a (such as Figure 3 As shown in the figure, the red dots are the distribution of the 100 samples collected), then: (Formula 2) In the above formula 2, k represents an empirical constant for adjusting the sensitivity of the above model.

[0036] 4) Establish a mathematical model for the relationship between the condition coefficient m, crowding coefficient c, and negative evaluation degree a (e.g. Figure 4 As shown, the model can be used to know that the crowd flow coefficient c and the negative evaluation degree a are inversely correlated). Combined with the characteristic relationship of the above formula 1 and formula 2, we have: g=(Formula 1)×(Formula 2) According to the above derivation, the condition coefficient g for the energy storage cabinet to call the backup power module is: ;

[0037] Then, when calculating the conditional coefficient of passive allocation of security personnel for a certain scenic spot, the crowd congestion coefficient of the current state of the scenic spot is ; The negative evaluation degree obtained after collecting the social media UGC data of this scenic spot is , then we have: ; From the above calculations, we can know that the conditional coefficient for the platform to passively allocate security personnel to the scenic spot is , then it means that the platform needs to deploy additional security personnel to the scenic spots, restaurants or transportation hubs being tested. And there are: Table 1 Monitoring data for some areas and the relationship between security personnel dispatch

[0038] Based on the data in Table 1 above, we can know that when the sample data tends to infinity, there will be a boundary based on the condition coefficient m to determine whether to dispatch or increase security personnel to the scenic spot. That is, when When , it means that the platform needs to send additional security personnel to the detected scenic spots, restaurants or transportation hubs; when When , it means that the platform does not need to send additional security personnel to the inspected scenic spots, restaurants or transportation hubs; when , it means that the platform can dispatch security personnel to the detected scenic spots, restaurants or transportation hubs.

[0039] Based on this, the risk level can be dynamically quantified, security resources can be accurately allocated, resource waste or shortage can be avoided, and the safety management efficiency of scenic spots can be improved.

[0040] Example 5 like Figure 1 As shown in the figure, digital twin technology is used to build a three-dimensional visualization map of urban cultural and tourism resources, including: 1) Based on real-time data from the government's cultural and tourism database and IoT devices at scenic spots, we used the Unity 3D engine to build a 3D scene and used Blender to repair model details. Based on the government database and IoT device data, we used Unity 3D and Blender to build a high-precision 3D scene, and synchronized passenger flow and traffic data to a visual map in real time via the MQTT protocol.

[0041] 2) Real-time passenger flow and traffic status data are synchronized with the internet access platform via the MQTT protocol. This provides managers and tourists with an intuitive panoramic view of the city's cultural and tourism resources, supporting real-time monitoring and decision-making, and enhancing data transparency and interactive experience.

[0042] Example 6 like Figure 1 As shown in the figure, the construction of the graph neural network includes: 1) Consider attractions, restaurants, and transportation hubs as nodes; the distance, walking time, and theme similarity between nodes as edges. Using attractions and restaurants as nodes, and distance, walking time, and theme similarity as edges, we build a graph neural network based on real-time congestion levels and user sentiment scores to optimize tour routes.

[0043] 2) Node popularity and the text sentiment scores of users’ evaluations of each node are used as node attributes.

[0044] 3) The real-time congestion level on the edge is used as the edge weight.

[0045] 4) Build a graph neural network for the attraction based on the aforementioned data and combine it with historical visitor trajectory data to optimize tour routes. Dynamically generate personalized tour routes to reduce congestion and improve visitor efficiency. Iterate and optimize the model using historical data to improve the accuracy and adaptability of route recommendations.

[0046] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. An artificial intelligence-based urban cultural tourism operation cluster platform, characterized by: include: Use distributed crawlers to capture data from government cultural and tourism databases, scenic area IoT devices, and social media UGC; Based on the statistical threshold method, data outside the reasonable range is identified and eliminated from the captured data. The data of different dimensions in the captured data are mapped to the interval [0, 1] through the normalization model. The characteristic value of the captured data is then calculated based on the noise reduction anomaly model. An analysis engine is deployed locally within the operations cluster platform. Using a pre-trained model in one of its sub-engines, it analyzes urban cultural and tourism information in real time and identifies sudden congestion and unusual personnel incidents at urban tourist attractions. Another sub-engine then matches resource supply and demand, enabling the passive allocation of security personnel to urban tourist attractions. Build a 3D visualization map of urban cultural and tourism resources based on digital twin technology. This 3D visualization map is used for user interaction terminals. Based on the user's real-time location, preferences, and resource status, the optimal tour route is generated through graph neural networks; security personnel at urban tourist attractions are proactively allocated based on actual conditions.

2. The artificial intelligence-based urban cultural tourism operation cluster platform according to claim 1 is characterized by: The normalization model is used to standardize the data captured by the distributed crawler, so: ； Where, Indicates the data after normalization, which includes the crowding coefficient; Indicates real-time captured data, including the current flow of people at attractions, restaurants, or transportation hubs; Indicates the minimum amount of data; Indicates the maximum data size.

3. The artificial intelligence-based urban cultural tourism operation cluster platform according to claim 1 is characterized by: The pre-trained model recognizes event entities in social media UGC texts in the following steps: 1) 5,000 texts were manually annotated, including the event type and sentiment polarity of positive or negative; 2) Fine-tune the weights during pre-training, and then use the pre-trained model to assign sentiment scores to the text about the marked events that appear at the scenic spots in the text posted by social media user-generated content.

4. The artificial intelligence-based urban cultural tourism operation cluster platform according to claim 2 is characterized by: The other sub-engine calculates the conditional coefficient for passively assigning security personnel to the area based on the current crowd conditions at the scenic spot, restaurant, or transportation hub and the evaluation of the current area on social media. The conditional coefficients are: ； Where, The conditional coefficient indicating the platform's need to passively allocate security personnel to the attraction, restaurant, or transportation hub; Indicates the crowding coefficient of the current tourist attraction, restaurant or transportation hub; Indicates the degree of negative evaluation of the current area by social media.

5. The artificial intelligence-based urban cultural tourism operation cluster platform according to claim 4 is characterized by: when When , it means that the platform needs to deploy additional security personnel to the inspected scenic spots, restaurants or transportation hubs; when When , it means that the platform does not need to deploy additional security personnel to the inspected scenic spots, restaurants or transportation hubs; when , it means that the platform can dispatch security personnel to the detected scenic spots, restaurants or transportation hubs.

6. The artificial intelligence-based urban cultural tourism operation cluster platform according to claim 1 is characterized by: The digital twin technology is used to construct a three-dimensional visualization map of urban cultural and tourism resources, including: 1) Using the Unity 3D engine to build 3D scenes based on real-time data from the government’s cultural and tourism database and scenic area IoT devices, and using Blender to repair model details; 2) Synchronize real-time passenger flow and traffic status data to the Internet access platform through the MQTT protocol.

7. The artificial intelligence-based urban cultural tourism operation cluster platform according to claim 1 is characterized by: The construction of the graph neural network includes: 1) Consider scenic spots, restaurants, and transportation hubs as nodes; consider the distance, walking time, and theme similarity between nodes as edges; 2) Node popularity and the text sentiment scores of users’ evaluations of each node are used as node attributes; 3) The real-time congestion level on the edge is used as the edge weight; 4) Based on the above data, a graph neural network is constructed for the scenic spot, and the tour path is optimized by combining historical tourist trajectory data.

Images