Smart tourism operation media data management system

Abstract

The invention provides a smart tourism operation media data management system. The system comprises a data acquisition module, a data analysis module, a recommendation generation module and a terminal adaptation module. The data acquisition module is deployed at a plurality of physical nodes of a tourist destination, and acquires tourist track data, tourist behavior data and environment data in real time in combination with an RFID beacon, a panoramic camera device and an environment sensor. The data analysis module receives the collected data and generates a structured tourist behavior model. The recommendation generation module calculates personalized preferences of tourists based on the tourist behavior model, and generates personalized guidance data including touring routes, scenic spot recommendation and interaction content. And the terminal adaptation module adapts the personalized guidance data according to the characteristics of the terminal device used by the tourist. Through the multi-source data fusion and intelligent recommendation technology, accurate analysis of tourist behaviors and personalized touring experience optimization are realized, the intelligent level of tourism services is improved, and the operation efficiency of scenic spots is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing systems, and particularly to an intelligent tourism operation media data management system. Background Art

[0002] With the digital development of the tourism industry, the application of intelligent tourism in scenic area management and tourist services has become increasingly widespread. Existing intelligent tourism management systems usually rely on tourists to manually query scenic spot information, use fixed route navigation, or rely on simple label matching recommendation methods to push operation media content.

[0003] However, there are still many deficiencies in existing intelligent tourism management systems. First, the collection method of tourist behavior data is relatively single, and it is impossible to comprehensively obtain tourists' interest points, preference characteristics, and real-time interaction data, resulting in limited accuracy of recommendation results. Second, the data analysis method relies on traditional rule matching or static recommendation based on historical behavior, and fails to make full use of multi-modal data for comprehensive analysis, making it difficult for the system to adapt to tourists' interest changes in real time. In addition, existing systems lack in-depth optimization in personalized recommendation and cannot provide truly tourist-demand-compliant tour routes, guiding information, and interactive experiences.

[0004] In view of the above problems, there is an urgent need for an intelligent tourism operation media data management system that can integrate tourists' trajectories, behavior patterns, and environmental factors to achieve more accurate personalized guiding and more optimized scenic area operation. Summary of the Invention

[0005] This application provides an intelligent tourism operation media data management system to achieve accurate analysis of tourist behavior and improve tourists' personalized tour experience.

[0006] This application provides an intelligent tourism operation media data management system, including: A data collection module deployed on multiple physical nodes at the tourism destination, including RFID beacons, panoramic camera devices, and environmental sensors, for collecting tourists' trajectory data, tourists' behavior data, and environmental data; A data analysis module for receiving the data collected by the data collection module; according to the collected data, aligning the positions and environmental information of tourists in space and time through time synchronization, and using image recognition and natural language processing to extract tourists' interest points; integrating the data after space-time alignment and the extracted interest point information to generate a structured tourist behavior model; A recommendation generation module for generating personalized guiding data according to the tourist behavior model; wherein the personalized guiding data includes tour routes that tourists may be interested in, relevant scenic spot recommendations, and personalized interactive content; The terminal adaptation module is used to receive the personalized guidance data output by the intelligent recommendation generation module, and adjust the presentation mode of the content based on the resolution, computing power, and interaction mode of the terminal device used by the tourist.

[0007] The present application has the following beneficial technical effects: (1) Through the collaborative work of RFID beacons, panoramic camera devices, and environmental sensors, the present application can obtain the trajectory data, behavior data, and environmental data of tourists in real time, and perform spatio-temporal alignment through time synchronization technology to ensure the accuracy and consistency of the data. Compared with the existing systems that only rely on GPS or a single data source, the present application can capture the behavior characteristics of tourists more comprehensively, providing a more reliable data basis for accurately analyzing the tourist interest points. (2) The present application combines image recognition and natural language processing technologies to intelligently analyze the photos taken by tourists, comment content, and interaction records, accurately extract the tourist interest points, and perform comprehensive modeling in combination with environmental factors. Compared with the traditional recommendation systems that rely on historical records or tag matching, the present application can dynamically adapt to the real-time interest changes of tourists, improving the accuracy of recommendations and the user experience. (3) Through the recommendation generation module, the present application provides a tour route, scenic spot recommendations, and interactive content that meet the tourist's interest preferences based on the tourist's behavior model, and can dynamically adjust the recommendation strategy in combination with real-time data. Compared with the existing static or one-dimensional recommendation methods, the present application can achieve highly personalized tour optimization according to the tourist's behavior pattern, current environmental state, and interaction feedback, improving the immersive experience of tourists. (4) The terminal adaptation module of the present application can automatically adjust the presentation mode of the recommended content according to the terminal device used by the tourist (such as a smartphone, tablet, smart glasses, etc.), including resolution adjustment, interaction mode optimization, and dynamic layout adjustment. Different from the existing general display methods, the present application can provide a smoother and more efficient guided tour experience according to the performance characteristics of different terminals and the interaction habits of tourists, enhancing the applicability and scalability of the system. Description of the Drawings

[0008] Figure 1 It is a schematic diagram of a smart tourism operation media data management system provided by the first embodiment of the present application. Detailed Embodiments

[0009] Many specific details are set forth in the following description in order to provide a thorough understanding of the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the connotation of the present application. Therefore, the present application is not limited by the specific embodiments disclosed below.

[0010] The first embodiment of the present application provides a smart tourism operation media data management system. Please refer to Figure 1, this figure is a schematic diagram of the first embodiment of the present application. The following will be combined with Figure 1 to provide a detailed description of a smart tourism operation media data management system according to the first embodiment of the present application.

[0011] The smart tourism operation media data management system includes a data collection module 101, a data analysis module 102, a recommendation generation module 103, and a terminal adaptation module 104.

[0012] The data collection module 101 is deployed on multiple physical nodes at the tourist destination, including RFID beacons, panoramic camera devices, and environmental sensors, and is used to collect tourists' trajectory data, tourists' behavior data, and environmental data.

[0013] The data collection module 101 is deployed on multiple physical nodes at the tourist destination. Its main function is to collect tourists' trajectory data, behavior data, and environmental data in real time to support subsequent data analysis, recommendation generation, and terminal adaptation. This module includes RFID beacons, panoramic camera devices, and environmental sensors, and each component works together to ensure the integrity and accuracy of the data.

[0014] The RFID (Radio Frequency Identification) beacon is used to identify tourists' identities and track their trajectories. Each tourist can be equipped with an RFID tag bound to personal identity information, which can be integrated into the smart bracelet, mobile application, or other wearable devices worn by the tourist. When a tourist enters an area equipped with an RFID beacon, the RFID beacon establishes a connection with the tourist device through short-range wireless communication technology (such as ultra-high-frequency RFID or low-power Bluetooth), records the tourist's entry time, departure time, and position coordinate information, and combines multiple RFID beacons to form continuous trajectory data. This data can be used to calculate key information such as the tourist's movement path, stay duration, and visit frequency in the scenic area. The data transmission of the RFID beacon can be uploaded in real time through the wireless network or local area network in the scenic area and synchronized with the data analysis module 102.

[0015] The panoramic camera device is deployed at key locations in the scenic area, including but not limited to the entrance of scenic spots, main tourist routes, interactive areas, and the perimeters of service facilities. This camera device is used to capture the behavioral data of tourists, such as the viewing angles of tourists at specific scenic spots, photo-taking behaviors, and crowd gathering situations. The panoramic camera device can use high-definition cameras, wide-angle lenses, or fish-eye lenses to cover a large area, and extract the activity patterns of tourists through image processing technology. Based on machine vision algorithms, this device can identify the staying time, facing direction, fixation focus, and interaction situations of tourists in a certain area. For example, when a tourist stays in front of an exhibit for more than the preset time and takes photos multiple times, the system can infer that the tourist has a high interest in this exhibit, and transmit this behavioral data to the data analysis module 102 to support subsequent personalized recommendations.

[0016] The environmental sensors are used to detect the environmental parameters in the scenic area, including but not limited to temperature, humidity, light intensity, crowd density, and noise level. The temperature and humidity sensors can be installed in outdoor scenic spots, inside exhibition halls, and tourist rest areas to monitor the environmental comfort level, and analyze the influence of environmental factors on tourists' tour preferences in combination with tourists' behavioral data. The light sensors are used to detect the lighting conditions in different areas of the scenic area, and can be used to optimize night-time guided tours or adjust the content of augmented reality (AR) guided tours. The crowd density sensors can use millimeter-wave radar, infrared detection, or video analysis technology to monitor the changing trend of the number of tourists in specific areas, and analyze the mobility of tourists in combination with RFID beacon data. When the tourist density in a certain area is too high, the system can dynamically adjust the recommendation strategy in combination with the environmental sensor data to guide tourists to relatively less crowded scenic spots to optimize the overall tour experience.

[0017] The components of the data collection module 101 work together to ensure the comprehensiveness and real-time nature of the data. The RFID beacons provide the identity and trajectory information of tourists, the panoramic camera device supplements the behavioral characteristics of tourists, and the environmental sensors provide external condition data. Each data source can use timestamp synchronization technology to ensure data consistency, and perform preliminary processing through local computing or edge computing devices to reduce data transmission latency and server load. All data is transmitted to the data analysis module 102 via wireless or wired networks after collection to support tourist behavior modeling and personalized recommendations.

[0018] Through the implementation of the data collection module 101, this system can obtain the dynamic data of tourists in real time and accurately, providing high-quality data support for subsequent intelligent recommendations, tourist route optimization, and personalized interactions, thereby enhancing the tourist experience and optimizing the operation and management of the scenic area.

[0019] Furthermore, the RFID beacons of the data collection module are dynamically bound to the tourist identity devices, and generate trajectory data when tourists enter or leave specific areas. The trajectory data includes the current location, staying duration, and interaction behaviors of tourists; The panoramic camera device of the data acquisition module combines the photo-taking or code-scanning behavior of tourists to extract the tourists' points of interest and correlates them with the trajectory data recorded by RFID beacons for associated analysis to improve the accuracy of point-of-interest recognition; The environmental sensors of the data acquisition module are used to detect the temperature, humidity, crowd density and light intensity in the scenic area, and correlate the environmental data with the tourists' behavior data to analyze the tourists' tour preferences under different environmental conditions.

[0020] In the intelligent tourism operation media data management system, the RFID beacons of the data acquisition module can be dynamically bound to the tourists' identity devices to accurately record the tourists' trajectory data. The tourists' identity devices can be smart bracelets, applications on smartphones, electronic tags on tickets or other wearable RFID tags. When a tourist enters the scenic area or a specific area, the RFID beacon establishes a unique association with the tourist's identity device through short-range wireless communication technology and records the tourist's identity information, entry time and location coordinates. This binding process adopts a dynamic encryption and identity authentication mechanism to ensure the security of tourist information and at the same time supports real-time changes in tourist identity. For example, when a tourist changes devices, teams merge or split, the system can automatically update the binding relationship. The recognition range of the RFID beacon can be adjusted according to the needs of the scenic area. Generally, ultra-high frequency (UHF) RFID is used to ensure that the movement trajectories of tourists in different areas can be accurately captured and there will be no recognition errors due to signal interference.

[0021] When a tourist enters or leaves a specific area, the system will automatically generate trajectory data, including the tourist's current location, stay duration and interaction behavior. The generation of trajectory data depends on the continuous read and write records of RFID beacons. When a tourist enters a certain area, the system will calculate the activity time of the tourist in this area and determine the tourist's movement path in combination with the recognition of adjacent beacons. For example, in a museum exhibition hall, a tourist may read the starting position from the beacon at the entrance of the exhibition hall, and then the system infers the tourist's tour sequence and points of interest based on the beacon records of different exhibit areas passed by the tourist. The calculation of the stay duration is based on the timestamps of the entry and exit times, combined with multiple stay records of the tourist at the same location to filter out data anomalies caused by passing by briefly. The recording of interaction behavior combines operations such as code-scanning, clicking and photo-taking of tourists in this area, and the system automatically analyzes the depth of participation and degree of interest of tourists.

[0022] The panoramic camera device can capture the behavioral characteristics of tourists in the scenic area. Especially when tourists take the initiative to take pictures, scan codes or interact with specific exhibits, the system can use image analysis technology to extract the tourists' points of interest. The camera device can be deployed at key locations such as main scenic spots, exhibition halls, and interactive experience areas. The system can infer the specific objects of their attention by calculating the staying time, shooting angle, facial orientation, and gesture movements of tourists in the picture. For example, when a tourist stops in front of a certain exhibit for a long time and takes pictures of it many times, the system can determine that this exhibit is a potential point of interest for this tourist. The code-scanning behavior further enhances the recognition accuracy of the points of interest. When tourists scan QR codes, NFC tags or use the scenic area application to obtain relevant information near exhibits or scenic spots, the system can record the frequency of code scanning by tourists and the content they access, and correlate it with the trajectory data recorded by RFID beacons to improve the accuracy of point-of-interest analysis. By combining the data of the panoramic camera device with the trajectory data of RFID beacons, the system can form complete tourist behavior data and avoid misjudgments that may occur with a single data source. For example, relying solely on RFID beacons may not accurately determine whether a tourist is truly interested in a certain exhibit. After combining the records of the camera device, it is possible to further analyze the tourist's behavior pattern and thus more accurately extract their points of concern.

[0023] Environmental sensors are used to monitor the temperature, humidity, crowd density, and light intensity in the scenic area in real time, and correlate this data with tourist behavior data to analyze tourists' tour preferences under different environmental conditions. Temperature and humidity sensors are installed in different areas, such as outdoor scenic spots, inside exhibition halls, and tourist service centers, to ensure that the system can comprehensively understand the microclimate in the scenic area. When the tourist's tour trajectory passes through multiple areas with significant temperature and humidity changes, the system can analyze whether tourists will adjust their tour behavior due to temperature and humidity factors. For example, on a hot day, if tourists are more inclined to stay in shaded or air-conditioned exhibition halls, the system can infer that tourists are more sensitive to climate comfort and preferentially provide tour paths that meet their preferences in subsequent recommendations. Crowd density sensors use infrared detection, millimeter-wave radar, or video analysis technology to detect the real-time number of tourists in each area of the scenic area, and calculate the average staying time of tourists in high-density areas in combination with the data recorded by RFID beacons. If the tourist density in a certain area is high and the staying time of tourists is short, it may mean that there is congestion or a poor experience in this area. The system can dynamically adjust the recommendation strategy based on this data to guide tourists to scenic spots with fewer tourists but still meeting their interests. Light sensors are used to monitor the light changes in the scenic area at different times, especially in outdoor scenic spots or areas open at night. The system can analyze the tourists' tour experience based on the light intensity. For example, in the case of weak light, the system can infer that tourists are more likely to choose a night tour experience and add options such as night light shows and specific night tour routes in the recommended content.

[0024] Through the collaborative work of RFID beacons, panoramic cameras and environmental sensors, the data collection module can obtain more accurate tourist behavior data and combine the complementary advantages of different data sources to form a more complete tourist portrait. This data fusion method not only improves the accuracy of point of interest identification, but also enhances the understanding of tourist travel patterns, thereby providing more powerful data support for personalized recommendations, tourist guidance and scenic area management.

[0025] The data analysis module 102 is used to receive the data collected by the data collection module; based on the collected data, the location and environmental information of the tourists are spatiotemporally aligned through time synchronization, and the points of interest of the tourists are extracted by using image recognition and natural language processing; the spatiotemporally aligned data and the extracted points of interest information are integrated to generate a structured tourist behavior model.

[0026] The data analysis module 102 is used to process the tourist trajectory data, tourist behavior data and environmental data collected by the data collection module 101 to support subsequent personalized recommendations and intelligent scheduling. The core functions of this module include data reception, time synchronization, time-space alignment, point of interest extraction and behavior modeling. Each step is based on efficient data processing algorithms and optimization strategies to ensure the integrity, accuracy and real-time nature of the data.

[0027] In terms of data reception, the data analysis module 102 receives multi-source data from the data acquisition module 101 through a wired or wireless network. The trajectory data generated by the RFID beacon is transmitted in a timestamp format, including the current location of the visitor, the time of entering and leaving a specific area, the movement path and the duration of stay. The behavior data collected by the panoramic camera device extracts the activity patterns of the visitor through an image recognition algorithm, such as taking pictures, browsing directions, number of interactions, etc. The data format may include video frame sequences, behavior tags and timestamp information. The environmental data provided by the environmental sensor includes parameters such as temperature and humidity, crowd density, and light intensity. These data are updated by periodic uploading to reflect the real-time environmental conditions of the scenic area.

[0028] In the process of time synchronization and time-space alignment, the data analysis module 102 uses timestamp calibration and multi-source data fusion methods to ensure the time synchronization of different data sources. First, the data collected by RFID beacons, panoramic cameras and environmental sensors are time-normalized, and the data time is aligned based on the network time protocol (NTP) or local synchronous clock. Secondly, combined with the movement path, photo-taking behavior and environmental data of tourists, time-space matching is performed to associate the activity trajectory of tourists with environmental conditions. For example, when tourists stay in a certain exhibition hall, the corresponding environmental data can be used to analyze the tourists' tour behavior under different temperature, humidity or crowd density conditions, and combined with the data of the camera device to infer their attention time and interaction intention in front of the exhibits.

[0029] In terms of point-of-interest extraction, the data analysis module 102 combines image recognition and natural language processing technologies to deeply analyze the behavior data of tourists. For the photos taken by tourists, the system uses a convolutional neural network (CNN) model to extract features from the photo content, identify scenic spots, exhibits or landmark buildings in the photo, and correlate them with the trajectory data of tourists to judge the attention of tourists to specific scenic spots. For the comments, voice inputs or interaction content of tourists, the system uses natural language processing (NLP) technology to extract keywords and combine semantic analysis to judge the interest categories of tourists. For example, if tourists frequently mention words such as "history", "culture", "architecture" in their comments, the system can judge that they prefer historical and cultural scenic spots and give corresponding weights in subsequent recommendations.

[0030] In the process of modeling tourist behavior, the data analysis module 102 inputs the aligned data into a behavior analysis model to generate a structured tourist behavior model. This model constructs a personalized feature vector of tourists based on their historical behavior, real-time behavior and environmental conditions, including tour preferences, activity patterns, interaction habits and recommendation tendencies. For example, the system can identify whether tourists prefer free tours or guided explanations, whether they prefer taking photos as souvenirs or interactive experiences, and optimize subsequent recommendation strategies accordingly. The behavior model can be constructed using machine learning algorithms such as Bayesian networks, long short-term memory networks (LSTM) or random forests to achieve accurate prediction and dynamic update of tourist behavior.

[0031] The output results of the data analysis module 102 are stored in the form of structured data and transmitted to the recommendation generation module 103 to support personalized recommendations and tour optimization. This module not only realizes the standardization, cleaning and integration of data, but also improves the accuracy of tourist behavior modeling through multi-modal data fusion and in-depth behavior analysis, thus providing efficient and accurate analysis capabilities for the intelligent tourism system.

[0032] Furthermore, the data analysis module adopts a spatio-temporal alignment method based on a time window to align the trajectory data of tourists with environmental data through a time synchronization algorithm to improve the accuracy of tourist location and behavior information.

[0033] When processing the trajectory data and environmental data of tourists, the data analysis module adopts a spatio-temporal alignment method based on a time window to ensure the accuracy and consistency of the data through a time synchronization algorithm. The trajectory data of tourists is mainly generated by RFID beacons, panoramic camera devices and other sensors. These data come from different physical nodes and there may be deviations in the collection time. If directly used for behavior analysis and recommendation calculation, it may lead to inaccurate information and affect the effect of personalized recommendations. Therefore, the system processes these data through a time synchronization mechanism to make them consistent on the same time basis.

[0034] In terms of time synchronization, the data analysis module adopts a high-precision time synchronization algorithm to ensure that each data acquisition node can record and store data according to a unified time standard. This algorithm can be based on the Network Time Protocol (NTP), a local reference clock, or a distributed time synchronization method, enabling the normalization of data from different sources according to timestamps. For the trajectory data collected by RFID beacons, the system will attach timestamps each time the tourist identity device is read to accurately record the entry, exit, and stay times of tourists. For the behavior data collected by panoramic cameras, the system will match the moment when the behavior of tourists occurs in the video based on the time markers of video frames. For the data collected by environmental sensors, such as temperature and humidity, pedestrian flow density, and light intensity, a periodic recording method is adopted to ensure that the environmental conditions at different time periods can be matched with the behavior of tourists.

[0035] During the spatio-temporal alignment process, the system uses a time-window-based method to organize and align the collected data. The size of the time window is dynamically adjusted according to the needs of the application scenario to adapt to different tourist behavior patterns and data update frequencies. For example, for tourist trajectory data, the system can set a shorter time window to accurately record the stay time and movement path of tourists at each scenic spot. For environmental data, since the changes are usually relatively gentle, the time window can be set longer to reduce the computational burden of data updates. During spatio-temporal alignment, the system will select the data within the same time window for analysis and match them according to timestamps. For example, when a tourist enters a certain exhibition hall, the system will find the closest environmental data within this time window to ensure that the tourist's stay behavior matches the environmental parameters such as temperature and humidity, and pedestrian flow density at that time, so as to obtain more accurate behavior analysis results.

[0036] In addition, to improve the accuracy of spatio-temporal alignment, the system also performs abnormal data filtering and compensation processing. If there is data missing in a certain time window, for example, a certain RFID beacon fails to successfully record the trajectory of a tourist, the system can perform interpolation calculations based on the tourist's historical movement patterns, data from neighboring sensors, or the behavior patterns of other tourists to infer the possible trajectory of this tourist. Similarly, if the camera device fails to record some behavior data due to occlusion or lighting problems, the system can combine RFID data and environmental data to infer the activities of tourists in this area.

[0037] By adopting a time synchronization algorithm and a spatio-temporal alignment method based on a time window, the system can effectively reduce data errors and improve the matching accuracy between tourist trajectory data and environmental data. This processing method ensures the high accuracy of tourists' location and behavior information, enabling subsequent personalized recommendations, tour optimization, and tourist behavior analysis to be based on a more reliable data foundation, thereby enhancing the overall performance and user experience of the intelligent tourism system.

[0038] Furthermore, the data analysis module combines image recognition and natural language processing to analyze tourists' photographed pictures, interaction texts, and comment information, extract interest points, and match the preference categories of tourists based on a semantic association algorithm.

[0039] When processing tourists' behavior data, the data analysis module combines image recognition and natural language processing technologies to comprehensively extract tourists' interest points and perform preference matching based on a semantic association algorithm, thereby achieving more accurate personalized recommendations. This module can analyze the pictures taken by tourists, interaction texts, and comment information to understand tourists' interest tendencies, and fuse this data with other behavior information to build a complete tourist interest profile.

[0040] When processing the pictures taken by tourists, the system uses a deep learning model to automatically recognize the images and extract the key visual elements. This process first uses image preprocessing technologies, including denoising, color correction, and edge enhancement, to ensure that the image quality meets the analysis standard. Subsequently, the system uses a convolutional neural network (CNN) model to analyze the image content and identify the main objects, such as buildings, exhibits, natural landscapes, or specific cultural symbols. For scenarios such as museums and exhibition halls, the system can detect the exhibits or cultural relics photographed by tourists and, combined with the historical information in the database, determine whether the tourist shows a high interest in this type of exhibit. For example, if a tourist takes pictures of a certain cultural relic of a specific era multiple times in a historical exhibition, the system can classify them as users with a preference for that historical period. In addition, the system can also analyze the shooting habits of tourists through style features, such as whether they prefer specific colors, compositions, or lighting effects, to further refine their interest types.

[0041] When analyzing the interactive texts and review information of tourists, the system adopts natural language processing (NLP) technology to extract key content from the texts and deeply understand the semantics. First, the system uses techniques such as word segmentation, part-of-speech tagging, and named entity recognition (NER) to structurally process the words input by tourists and extract the scenic spots, exhibits, activities, or other relevant information involved. For example, when a tourist writes in a review that "the architectural style of this palace is very magnificent and the carving craftsmanship is exquisite", the system can identify keywords such as "palace", "architectural style", and "carving craftsmanship", and combine historical data to judge that the tourist has a high interest in ancient architecture or arts and crafts. Next, the system will use sentiment analysis algorithms to judge the tourist's attitude towards the content described to distinguish positive, negative, or neutral evaluations, so as to avoid pushing content that the tourist may not be interested in or may cause a bad experience during the recommendation. In addition, the system will also combine co-occurrence analysis and topic modeling techniques to classify the topics of the texts input by tourists. For example, if a tourist frequently mentions "natural scenery" and "hiking", then their preferences can be classified as outdoor exploration type tourists.

[0042] After completing image recognition and text analysis, the system uses semantic association algorithms to match the extracted points of interest with the tourist's preference categories. This process first establishes a mapping relationship between the points of interest and the preference categories to ensure that different types of content can be accurately classified. For example, the system has pre-constructed a knowledge graph containing multiple categories, including historical culture, modern art, natural scenery, science and technology exhibitions, etc. Each category contains corresponding keywords, image features, and text descriptions. When the system extracts points of interest from the tourist's photo and text data, it will calculate the semantic similarity between these points of interest and different preference categories and classify them based on the degree of association. For example, if the tourist's photos mainly contain ancient buildings and historical relics, and the review content involves relevant historical backgrounds, the system will classify them into the "historical culture lovers" category and give priority to providing historical-related tour routes or guided tour content in subsequent recommendations. The calculation method of semantic association can adopt semantic similarity calculation based on word vectors, such as using models like Word2Vec and BERT to ensure the accuracy of the matching results. In addition, the system will also combine collaborative filtering and tourist group behavior analysis to further optimize the matching rules, so that even when tourists do not directly indicate their interests, the system can infer their preferences based on their implicit behaviors.

[0043] Through the combined application of image recognition and natural language processing technologies by the data analysis module, the system can extract deep-level interest information from the multi-modal data of tourists and achieve preference matching based on semantic association algorithms, thereby optimizing the personalized recommendation effect of the intelligent tourism system. This method not only improves the extraction accuracy of points of interest but also effectively makes up for the deficiencies of a single data source, enabling the recommendation system to accurately adapt to the needs of tourists in different scenarios and enhancing the overall tour experience.

[0044] A recommendation generation module 103 is configured to generate personalized guiding data according to the tourist behavior model; wherein, the personalized guiding data includes tour routes that the tourist may be interested in, relevant scenic spot recommendations, and personalized interactive content.

[0045] The recommendation generation module 103 is used to generate guiding data that meets the personalized needs of tourists based on the tourist behavior model output by the data analysis module 102, so as to optimize the tour experience and improve the intelligence and interactivity of the tour process. This module comprehensively considers the tourist's historical behavior data, real-time trajectory, interest preferences, and environmental factors, and dynamically adjusts the recommended content, enabling tourists to obtain more targeted tour routes, scenic spot recommendations, and interactive information.

[0046] First, the recommendation generation module 103 receives the structured tourist behavior model provided by the data analysis module 102, which contains the tourist's tour preferences, activity patterns, interest characteristics, historical trajectories, and interaction records. Based on this model, the recommendation generation module 103 uses a multi-dimensional interest matching method to model the tourist's preferences and calculate their potential interest in different scenic spots, exhibits, or activities. Interest matching can be based on collaborative filtering algorithms, content recommendation algorithms, or hybrid recommendation methods to ensure the accuracy and timeliness of the recommendation results. For tourists who enter the scenic area for the first time, this module can provide default recommendations based on their basic information (such as age, language, travel mode, etc.); for historical tourists, their past tour data is combined to optimize the recommendation strategy to ensure the personalization degree of the recommended content.

[0047] When generating a tour route, the recommendation generation module 103 dynamically optimizes the tour path by combining the tourist's current location, scenic spot distribution, walking distance, and crowd density. The path planning not only considers the interest matching degree but also comprehensively calculates the tourist's travel time, physical route feasibility, and the current environmental state of the scenic area. For example, if a tourist has a high interest in historical and cultural scenic spots, the system can preferentially recommend routes that meet their interests, and adjust the tour order by combining their walking speed, current location, and preferred exhibition halls to enhance the overall tour experience. At the same time, this module can detect the tourist's real-time trajectory, and when it is found that they deviate from the recommended path or stay at a certain scenic spot for a long time, the recommended content is automatically adjusted to keep the tourist in the best tour state.

[0048] The recommendation generation module 103 is also responsible for pushing detailed information about relevant scenic spots and personalized interactive content. When a tourist approaches a certain scenic spot, this module can provide the historical background, cultural interpretation, introduction of key exhibits, and interactive experience options of this scenic spot according to the tourist's interest weights. For example, for tourists who like in-depth exploration, the system can push more detailed historical documents, expert explanations, or extended reading materials; for tourists who prefer interaction, AR guides, game interactions, or multimedia display content can be recommended. In addition, this module supports dynamic recommendations based on tourists' behaviors. When a tourist stays in front of an exhibit for a long time or interacts frequently with the exhibit, the system can automatically push detailed information about the relevant exhibit or recommend similar exhibits to further enhance the tourist's sense of immersion.

[0049] In addition, this module can optimize the recommendation strategy by combining the real-time feedback of tourists. When a tourist clicks, favorites, or has a voice interaction with the recommended content, the system will adjust the recommendation algorithm according to the feedback information to provide more accurate personalized content. For example, if a tourist skips a specific type of recommended content multiple times, the system can reduce the recommendation weight of this type of content and instead provide content that better matches the tourist's current interests. For group tours, the system can integrate the interest characteristics of group members to generate a tour route that takes into account the needs of all members to ensure that each member can obtain the best tour experience.

[0050] The recommendation generation module 103 can also work in coordination with the terminal adaptation module 104 to ensure that the recommended content can be optimally displayed according to the terminal device used by the tourist. When a tourist uses a smartphone, the recommended content can be presented in the form of text, images, and brief voice introductions; for tourists using smart glasses, an augmented reality guide mode is provided so that the recommended information can be combined with the tourist's actual perspective to provide a more intuitive guiding experience. If a tourist is in a multimedia interaction area, this module can push video explanations, 360° panoramic guides, or immersive experience content to enhance the tourist's interactive feeling.

[0051] Through the implementation of the recommendation generation module 103, this system can analyze tourists' interests in real time, combine personalized tour preferences with scenic area environmental data, dynamically optimize the recommended content, make the tourists' tour process more efficient and intelligent, and provide a highly customized interactive experience, thereby improving the overall tourism experience quality and scenic area management efficiency.

[0052] Furthermore, the recommendation generation module combines the tourist's historical behavior data, real-time trajectory, and point-of-interest information, adopts behavior pattern analysis and multi-dimensional interest matching methods to predict the tourist's potential interests, and dynamically adjusts the personalized guiding data; and when it detects that a tourist stays at a designated scenic spot for a long time or returns multiple times, it automatically pushes the background introduction, historical story, or interactive experience content of this scenic spot.

[0053] When calculating personalized guiding data, the recommendation generation module fully combines the historical behavior data, real-time trajectory, and point of interest information of tourists to construct a complete user profile, and uses behavior pattern analysis and multi-dimensional interest matching methods to predict the potential interests of tourists. This module can, through in-depth analysis of tourists' past visit records, identify their preference categories, and dynamically adjust the recommendation strategy based on the current real-time behavior to ensure that the provided guiding content highly matches the interests and needs of tourists.

[0054] When processing historical behavior data, the system summarizes and organizes the past visit trajectories, stay durations, interaction behaviors, and feedback data of tourists. For example, if a tourist stays at multiple historical attractions for a long time, or queries the detailed information of the same type of exhibits multiple times, the system will record these behaviors and assign corresponding interest weights. Through machine learning algorithms such as clustering analysis or Bayesian classification, the system can identify whether a tourist prefers a certain specific type of attraction, such as historical relics, modern art galleries, or natural scenery areas. At the same time, the system will also analyze the feedback of tourists on the recommended content, such as the frequency of collection, like, or viewing the tour details, so as to further optimize its interest model.

[0055] When integrating real-time trajectory data, the system uses the information such as the current location, walking path, and stay time of tourists provided by the data collection module to judge the current tour state of tourists. For example, when a tourist enters a specific attraction area, the system will detect whether they are following the recommended route or whether they are interested in the recommended attractions. In addition, by analyzing the stay time of tourists at different locations, their interest intensity can be inferred. If a tourist stops in front of a certain exhibit for a long time, or makes multiple round trips within the same area, the system can infer that the exhibit or attraction has high attractiveness to them, and thus give it a higher priority in subsequent recommendations.

[0056] When performing multi-dimensional interest matching, the system comprehensively analyzes the historical behavior data, real-time trajectory, and point of interest information, and calculates the potential interest degree of tourists in various types of attractions. For example, the system not only considers the types of attractions that tourists have visited, but also combines their real-time behavior to judge whether there are new interest biases. If a tourist mainly visited historical attractions in the past, but shows a high degree of attention to a science and technology museum during the current visit, the system will dynamically adjust the recommendation strategy, increase the guiding information of science and technology exhibits, and provide corresponding extended reading or interactive experience content. In addition, the system can also combine the social data of tourists, such as their interaction with other tourists, to further optimize the recommendation results. For example, if multiple tourists with similar interests show a high degree of participation in a certain attraction, the system can judge that the attraction has high potential attractiveness and recommend it to tourists with the same type of interests.

[0057] When tourists show high attention to a designated scenic spot, the system will automatically push corresponding guided tour information. When it detects that tourists stay in front of a certain scenic spot or exhibit for a long time, or return to the same place multiple times, the system will trigger an intelligent push mechanism to provide more in-depth background introductions, historical stories, or interactive experience content. For example, if tourists stay in front of a historical building for a long time, the system will push detailed information about the building's construction history, design concept, and related historical events, and can provide a virtual restoration model using augmented reality (AR) technology to enhance the tourists' sense of immersion. For scenic spots with interactive exhibits, the system can recommend experiential projects for tourists to participate in, such as virtual puzzles, voice Q&A, or social interactions, to enhance the fun of the tour. In addition, if tourists return to the same scenic spot multiple times within a short period, the system will recognize their high interest in this scenic spot and further push advanced guided tour content, such as expert explanations, behind-the-scenes stories, or special event information during specific time periods.

[0058] Through the implementation of the recommendation generation module, the system can combine tourists' historical behaviors and real-time status, predict their potential interests, and provide dynamically adjusted personalized guiding data based on behavior patterns and multi-dimensional interest matching. By automatically pushing content highly relevant to tourists' interests, the system can not only enhance the guided tour experience but also enable tourists to obtain a deeper cultural and knowledge enjoyment, thereby enhancing the interactivity and intelligence level of the smart tourism system.

[0059] Furthermore, the recommendation generation module further combines the analysis of tourists' group behaviors to optimize the recommended tour routes and tourist diversion scheduling. The group behavior analysis includes: Based on the tourist trajectory data obtained by the data acquisition module, RFID beacon data, and the tourist group behavior data collected by the panoramic camera device, the data analysis module identifies and classifies the group types of tourists. The group types include self-guided tour groups, parent-child tour groups, group tour groups, and in-depth exploration groups; Based on the characteristic information of the group types, the recommendation generation module dynamically optimizes the recommended content, where: For self-guided tour groups, personalized tour routes are generated based on individual historical behavior data and interest characteristics, and optional tour path branches are provided to enhance the tourists' freedom of choice; For parent-child tour groups, scenic spots that meet children's interests and have educational value are preferentially recommended, while considering the interests of parents to balance the tour needs of children and parents; For group tour groups, recommended routes that adapt to the team's schedule are provided to ensure the interest integration of team members and optimize the order of scenic spot visits for the team to reduce tour time conflicts; For the group of tourists with a strong interest in in-depth exploration, scenic spots with longer explanatory content, detailed historical background introductions, and interactive experiences are recommended first to meet their needs for in-depth exploration of culture and knowledge. The recommendation generation module combines the real-time crowd density data collected by environmental sensors. When the tourist density at a specified scenic spot or area exceeds the preset threshold, the tour recommendation strategy is adjusted, specifically as follows: Before tourists reach crowded areas, alternative scenic spots that match their preferences and have a lower crowd density are recommended based on their preferences. After tourists enter high-density areas, diversion suggestions within the scenic spot are provided. The diversion suggestions include recommending tourists to enter specific exhibition areas or participate in guided tours with a limited time to relieve local congestion. The recommendation generation module continuously monitors changes in the behavior patterns of tourist groups. When it detects that group members are dispersed, merged, or there are significant changes in their tour interests, the recommendation strategy is dynamically adjusted, specifically as follows: When a group that was originally touring together splits into individual tourists, the system resumes individualized recommendations and provides personalized tour routes. When multiple individuals form a new tour group, the system recalculates the combined interests of the group and generates a new tour route based on the interest integration strategy.

[0060] The recommendation generation module combines the analysis of tourist group behavior to optimize tour route recommendations and tourist diversion scheduling, enabling the system to provide more accurate personalized guided tour suggestions based on tourists' tour styles, group characteristics, and real-time environmental conditions. This module uses the tourist trajectory data obtained by the data analysis module, RFID beacon data, and group behavior data collected by panoramic camera devices to identify tourists' tour patterns and classify group types. By categorizing the group attributes of tourists, the system can accurately judge their tour needs and optimize the recommended content accordingly.

[0061] The identification of tourist group types mainly relies on the analysis of trajectory data, behavior patterns, and interaction methods. Independent travel groups usually exhibit the characteristics of walking freely, randomly shuttling between multiple scenic spots, and having a flexible itinerary arrangement. For such tourists, the system generates personalized tour routes based on their personal historical behavior data and interest preferences, and provides multiple optional paths, enabling tourists to freely switch between recommended scenic spots to enhance the flexibility of the tour experience. The characteristics of family travel groups include frequent stops at child-friendly facilities within a short period of time, such as science popularization exhibition halls, interactive experience areas, or amusement facilities. At the same time, parents may query educational content. When recommending tour routes, the system preferentially selects scenic spots that combine children's entertainment and educational value, and also takes into account the interests of parents. For example, it recommends exhibitions or activities that can meet the needs of both children and adults to achieve a balance of family travel. Group tour groups usually have a fixed itinerary arranged by a tour guide or travel agency, and the actions of tourists are relatively consistent. The tour time is restricted by the team plan. In response to this characteristic, the system provides routes that conform to the team's time arrangement, optimizes the visiting order of each scenic spot, and ensures that team members can obtain a guided tour experience that combines commonality and personalization during the tour, reducing time conflicts caused by improper itinerary arrangements. The characteristics of in-depth exploration groups are that they show extremely high attention to a certain type of scenic spot. For example, they stay in a historical museum for a long time or conduct in-depth inquiries about exhibits. When recommending scenic spots for this group, the system preferentially selects exhibitions with detailed explanatory content, rich background information, and interactive experiences, and provides in-depth learning paths, such as extended reading, virtual restoration displays, or expert explanations, to meet the tourists' in-depth exploration needs for culture and knowledge.

[0062] The recommendation generation module not only optimizes the recommended content according to the types of visitors, but also combines the real-time crowd density data obtained by environmental sensors to dynamically adjust the tour route, ensuring that visitors can experience the scenic spots under the best tour conditions. The system continuously monitors the visitor density in various areas of the scenic area. When the number of visitors in a certain scenic spot or exhibition hall exceeds the set threshold, the system will adjust the tour recommendation strategy to prevent visitors from entering overly crowded areas. Before visitors enter high-density areas, the system recommends corresponding alternative scenic spots based on the interest matching degree of visitors. For example, when a certain museum exhibition hall is crowded with people and visitors stay for a long time, the system will recommend adjacent unsaturated exhibition halls to visitors with historical and cultural preferences, or recommend art exhibitions to visitors with stronger art interests, so as to ensure that visitors can complete the tour experience in a more comfortable environment. If visitors have already entered high-density areas, the system provides more refined crowd diversion suggestions, including recommending visitors to enter specific exhibition areas, adjusting the tour order, or guiding visitors to participate in guided tours with a limited time, so as to optimize the fluidity of the scenic area and the experience of visitors. For example, in a popular exhibition area, when the system detects a high crowd density and a long stay time of visitors, the system will push the exhibit explanation activities at specific time periods to visitors or guide them to the same type of exhibition area at an appropriate time to disperse the crowd and improve the tour comfort.

[0063] The system also continuously monitors changes in the behavior patterns of visitor groups and dynamically adjusts the recommendation strategy when group members disperse, merge, or change their interests. When the members of a group that originally toured together split into independent individuals during the itinerary, such as family members going their separate ways or group tourists temporarily leaving the main group, the system will automatically resume individualized recommendations and provide each visitor with a tour route that matches their personal interests. For newly formed tour groups due to temporary situations, such as independent visitors forming a temporary group after social interaction, the system will recalculate the combined interests of the group and generate a new tour route based on the interest integration strategy to ensure that the recommended content can meet the overall needs of the newly formed group. For example, if multiple visitors stay in the same exhibition area for a long time and start discussing the exhibits together, the system can infer that they have formed a temporary interest group and recommend subsequent related exhibitions or interactive experiences that they can participate in together to enhance the coherence and interactivity of the group tour.

[0064] By combining the recommendation generation module with group behavior analysis, the system can more accurately meet the needs of different types of visitors, intelligently divert crowds in high-crowd situations, and improve the comfort and efficiency of the tour. At the same time, when detecting changes in the structure of visitor groups, the module can flexibly adjust the recommendation strategy, enabling the smart tourism system to have stronger adaptability and dynamic optimization capabilities, thereby enhancing the overall tour experience and the operating efficiency of the scenic area.

[0065] Furthermore, the recommendation generation module adopts a dynamic personalized tour recommendation algorithm to calculate the personalized tour score of tourists for each scenic spot based on the tourists' historical behavior data, real-time trajectory data, environmental factors and points of interest information, and optimizes the recommended tour route according to the personalized tour score. The dynamic personalized tour recommendation algorithm uses the following formula 1 to calculate the personalized tour score of tourists: in, Recommended attractions The higher the score, the more attractive the attraction is to tourists, and the system will give priority to this attraction when recommending.

[0066] The first influencing factor is the tourists’ historical stay time, which reflects the tourists’ past interest in the same category of attractions. is the influence weight of historical stay time. The recommended value range of this coefficient is 0.3 to 0.5. A higher value means that the system is more inclined to recommend attractions where tourists have stayed for a long time in the past. It is a historical and recommended attraction for tourists. The average length of stay at attractions of the same category; It is the maximum length of stay of tourists in all scenic spots in history; The second factor is the number of historical interactions of tourists at attractions in the same category. is the influence weight of the number of historical interactions, and its recommended value is between 0.2 and 0.4, which is suitable for adjusting the system's emphasis on interactive behaviors. If it is too high, the system may overly favor attractions that have frequent interactions but are not necessarily of real interest; if it is too low, it may ignore tourists' actual interactive interest in certain attractions. It is a historical and recommended attraction for tourists. The average number of interactions with attractions of the same category, such as clicking to view details, taking photos, scanning codes, or participating in activities. This indicator is used to measure the degree of active participation of tourists in this type of attraction. is the maximum number of interactions among all attractions in the history of tourists; The third factor is the tourists’ interest matching, which is measured by calculating the similarity between tourists’ interest characteristics and the theme of the attractions. is the influence weight of interest matching, and the recommended value range is 0.4∼0.6. A higher value can enhance the consideration of interest matching and make the recommendation more personalized. It is a feature of interest The importance weight in the overall interest feature can be calculated based on the visitor's behavioral data, such as the number of past visits to the interest category, browsing time, and interaction frequency. is the set of tourists' interest characteristics, defined as , where the set each element in represents an interest characteristic of a tourist, is the total number of tourists' interest characteristics; is the similarity between the tourists' interests and the theme of the scenic spot , which can be calculated by collaborative filtering, semantic analysis or deep learning methods, such as vector distance calculation based on Word2Vec, BERT or other embedding representation models.

[0067] The similarity between the tourists' interests and the theme of the scenic spot can be calculated by various methods. The system needs to comprehensively analyze the tourists' past behavior data, text descriptions, interaction methods and the preferences of other tourists to accurately judge the potential interest degree of tourists in a certain scenic spot. This similarity calculation not only depends on direct behavior matching, but can also be completed by semantic analysis, collaborative filtering or deep learning methods.

[0068] In semantic association calculation, the system will match based on the text description of the scenic spot and the tourists' interest tags. For example, if a tourist has visited scenic spots related to "ancient architecture" and "cultural heritage" many times, the system will assign these theme tags to their interest set. When recommending a new scenic spot, if the introduction of the scenic spot contains similar keywords, such as "Qing Dynasty architecture", "history museum", "world cultural heritage", etc., the system will calculate the semantic similarity between these words. If the tourists' interest keywords and the theme of the scenic spot are semantically close, it can be considered that the tourist has a high interest in the scenic spot. This method can use Word2Vec, BERT or other natural language processing models to convert text into vectors and calculate their distances in the semantic space. If the distances between two vectors are close, it means they are semantically similar, and the system will increase the recommendation weight of the scenic spot.

[0069] The collaborative filtering method uses the behavior of other similar tourists to calculate the similarity. For example, if multiple tourists stay in the history museum for a long time and then all choose to visit the ancient architecture park, the system can infer that there is an association between these two types of scenic spots in tourists' preferences. When a new tourist shows a high interest in a certain type of exhibition but has not visited the relevant ancient architecture park, the system can, based on the behavior patterns of other tourists, infer that the tourist may also be interested in ancient architecture and give priority to recommending similar scenic spots to them. Collaborative filtering is not only based on the individual historical behavior of tourists, but also combines the overall tour data of all tourists to discover hidden interest associations and improve the accuracy of recommendations.

[0070] In deep learning methods, the system can use a neural network model to learn the complex relationships between tourists' interests and scenic spot themes. For example, BERT can be used to process tourists' text reviews, visit records, and scenic spot introduction texts to extract key features. Then, using vectorization methods, tourists' interests and scenic spot content are mapped into the same semantic space, and their cosine similarity or Euclidean distance is calculated. If the interest point vector of a certain tourist and the theme vector of a certain scenic spot have a high similarity, the system will consider that this scenic spot meets the tourist's preferences and increase its ranking in the recommendation list.

[0071] For example, if a tourist has frequently visited "technology"-related exhibition halls in multiple scenic areas in the past, such as the Artificial Intelligence Expo, the Space Science and Technology Museum, and the Robot Interaction Area, and this tourist has never visited the "Future Technology Experience Center" in the current scenic area, then the system can discover through semantic analysis that the "Future Technology Experience Center" is similar in theme to the exhibition halls the tourist has visited before, and based on collaborative filtering, it can also find that other tourists interested in the AI exhibition also tend to visit this technology center. Therefore, even if the tourist has not actively searched for or clicked on this scenic spot, the system can still infer their potential interest and give priority to the recommendation.

[0072] Through these methods, the system can dynamically analyze tourists' interest preferences, combine the theme information of scenic spots, calculate recommended content that better meets tourists' needs, and thus improve the personalization and intelligence level of the tour experience.

[0073] Tourists' interest characteristics mainly include their tour preferences, interaction methods, content choices, and social behaviors. These characteristics can help the system more accurately recommend scenic spots and guided tour content that match tourists' interests.

[0074] Tour preferences can be analyzed from the types of scenic spots tourists have visited in the past. For example, some tourists prefer to visit historical sites, museums, or cultural exhibition halls, while others tend to modern art, technology interaction halls, or outdoor natural scenic areas. If a certain tourist selects scenic spots of similar types in multiple scenic areas, the system can infer their preferences and give priority to providing similar content in future recommendations.

[0075] Interaction methods are also an important basis for judging interest characteristics. Some tourists take photos in the exhibition area, scan codes to obtain additional information, or watch multimedia introductions, which indicates that they have a high degree of attention to such exhibits. If a tourist stays in front of a certain type of exhibit for a long time or frequently clicks on relevant information, the system can determine that they are interested in this theme and increase the guided tour content of similar exhibits in subsequent recommendations.

[0076] Content selection reflects the information acquisition preferences of tourists. Some tourists prefer to learn about scenic spots through brief audio guides, while others are more inclined to read detailed background introductions or even consult expert reviews. If the system discovers that tourists often choose the in-depth reading mode, it will provide more detailed materials in subsequent recommendations. For tourists who like to take a quick tour, the system will optimize the recommendations to make the content more concise and easy to read.

[0077] Social behavior can also help the system judge interest characteristics. If tourists are accustomed to sharing certain types of scenic spots or activities with friends or on social platforms, such as internet-famous check-in spots, food streets, or parent-child experience halls, the system can infer that they prefer a more social tour experience. Additionally, if tourists often browse the same types of scenic spots with some users who have similar interests, the system can optimize the recommendation strategy based on the behavior of similar tourists to make the recommended content more in line with their preferences.

[0078] Taking these factors into comprehensive consideration, the system can accurately identify the interest characteristics of tourists and dynamically adjust the content during personalized recommendations to make the tour experience more in line with the real needs of tourists.

[0079] The last factor is the crowding degree of scenic spots, which is modeled through an exponential decay function to reflect tourists' tendency to avoid high-density areas. is the crowding degree influence weight, and the recommended value range is 0.2 - 0.4, which is used to adjust the degree to which the system considers the crowding degree. is the adjustment coefficient, and its recommended range is 0.5 - 1.5. A larger value indicates that tourists are more sensitive to the crowding degree, while a smaller value indicates a smaller impact of the crowding factor. represents the recommended scenic spots is the current crowd density, which can be calculated through sensors in the scenic area, real-time camera analysis, or tourist location data, with the unit of the number of tourists per square meter.

[0080] After calculating the personalized tour scores of each scenic spot, the system needs to optimize the tour route so that the recommended order of scenic spots not only conforms to tourists' interests but also ensures tour efficiency. The route optimization problem is modeled as a weighted traveling salesman problem, and the objective function is as follows: Among them, is the number of recommended scenic spots; is the path length penalty factor, and its recommended range is 0.1 - 0.5, which is used to control the degree to which the system attaches importance to path optimization. A larger value will tend to recommend shorter paths, while a smaller value pays more attention to the interest scores, making the recommendation results more flexible. represents the scenic spot to the scenic spot the walking distance.

[0081] Furthermore, in the dynamic personalized tour recommendation algorithm, the interest feature weight is calculated according to the following formula 3: First, represents the total number of the tourist's historical visit records. Each visit record contains the tourist's behavior data of visiting a certain interest feature at a specific time, including the number of interactions and the stay time . These behavior data are used to measure the tourist's attention degree to the interest feature at different time points.

[0082] The number of interactions represents the interaction frequency of the tourist with the content related to the interest feature during the th visit, including interactive behaviors such as clicks, likes, and comments. These interactions reflect the tourist's active interest in this type of content. If the tourist clicks on the detailed information of a certain type of scenic spot multiple times, comments on it, or collects relevant content, it indicates that the tourist has a higher degree of attention to this interest feature. Therefore, the number of interactions is used to calculate the interest feature weight, so that the features with more interactions are given higher priorities in subsequent recommendations. To avoid the influence of the absolute number of interactions of different tourists on the calculation results, a normalization method needs to be used, so the denominator uses , that is, the maximum number of interactions of the tourist with the content related to the interest feature in all historical visit records.

[0083] The stay time represents the stay duration of the tourist at the scenic spots related to the interest feature during the th visit record, usually in seconds or minutes. A longer stay time generally means that the tourist has a more in-depth attention to this interest category. For example, if the tourist watches the robot interaction demonstration for a long time in the science and technology museum and only stays briefly in front of other exhibits, it can be inferred that the tourist has a high interest in artificial intelligence or robotics technology. Therefore, this factor is also an important variable affecting the interest feature weight. Similarly, to ensure the stability of the calculation, normalization is required, and the denominator uses , that is, the maximum stay time of the tourist at the scenic spots related to the interest feature in all historical visit records.

[0084] The parameters and They are the influence weights of the interaction times and the stay time respectively, used to balance the contributions of the two to the interest feature weight. Generally, if the system hopes to emphasize the active interaction behavior of tourists more, the value of can be appropriately increased. For example, take ; if it hopes to pay more attention to the stay time of tourists, the weight of can be increased, for example, set to . If the behavior pattern of getting off the vehicle tends to be quick browsing rather than in-depth interaction, the value of can be relatively high to ensure that the influence of the stay time is not ignored.

[0085] Exponential decay factor controls the degree of influence of historical behaviors, so that the weights of earlier behavior data gradually decrease during calculation. Its role is to simulate the natural law of interest change, that is, recent behaviors have a greater impact on the current interest, while the contribution of earlier behaviors to interest judgment gradually decays. The recommended value range of the exponential decay factor is usually between 0.01 and 0.10. A smaller value means that the influence of historical behaviors does not decay for a long time, which is suitable for users with relatively stable interests. A larger value means that the interest changes quickly, and the system assigns a lower weight to earlier behavior data. For example, if a tourist has visited an art exhibition hall many times within a year but has not visited this type of exhibition hall in the past three months, a larger value will reduce the influence of past behaviors on the current interest, while a smaller value will maintain the preference for art exhibition halls.

[0086] Timestamp represents the time when the tourist visits the relevant interest feature for the th time, and is the current time. The exponential decay term makes the newer visit records have higher weights during calculation, while the weights of earlier visit records will decrease over time. For example, if a tourist has recently visited a science fiction exhibition hall frequently, while mainly visiting a history museum a year ago, the exponential decay term will reduce the influence of old visit records on the calculation of the current interest feature, making the recommendation more in line with the tourist's current interest state.

[0087] The denominator part is a normalization factor, , used to ensure the stability of the sum of interest feature weights and avoid the calculation results being affected by the number of historical visits. If a tourist has more visit records at a certain type of interest point, the normalization operation of the denominator can balance the weights of each visit and ensure that the calculated truly reflects the tourist's interest intensity without being distorted due to too many or too few visit times.

[0088] Overall, by combining the interactive behaviors, stay times, and time decay factors of tourists, this formula enables the system to dynamically adjust the weights of interest features, ensuring that the recommended content can accurately reflect the latest preferences of tourists. In this way, in the intelligent tourism operation media data management system, the system can not only identify the long-term interest tendencies of tourists but also flexibly adapt to changes in interests, making personalized recommendations more accurate and efficient.

[0089] The terminal adaptation module 104 is used to receive the personalized guidance data output by the intelligent recommendation generation module and adjust the presentation mode of the content based on the resolution, computing power, and interaction method of the terminal device used by the tourist.

[0090] The terminal adaptation module 104 is used to receive the personalized guidance data output by the recommendation generation module 103 and perform adaptation according to the characteristics of the terminal device used by the tourist to ensure that the recommended content can be presented in the best way on different devices. The core functions of this module include device feature recognition, content format adjustment, interaction method optimization, and real-time performance adjustment to improve the experience of tourists when using different terminal devices.

[0091] In terms of device feature recognition, the terminal adaptation module 104 can detect the type of device currently used by the tourist and automatically adjust the presentation mode of the recommended content. This module can identify smartphones, tablets, smart glasses, and other terminal devices with display or interaction functions, and obtain the hardware parameters of the device, such as screen resolution, refresh rate, computing power, storage space, and network connection status. In addition, this module can also identify the operating system type, browser kernel, or application running environment of the device by parsing the User-Agent information or operating system interface (API) to optimize the compatibility of the content. For example, when it is detected that the tourist is using smart glasses, the system will preferentially provide information in the augmented reality (AR) mode, while for smartphones, an interface adapted to touch operations will be used for display.

[0092] In terms of content format adjustment, the terminal adaptation module 104 can convert and optimize the format of the personalized guidance data provided by the recommendation generation module 103 to ensure that the content maintains good readability and interactive experience on different terminals. For example, for smartphones with smaller screen sizes, the system will display information in the form of concise text, pictures and voice broadcasts to avoid information overload, and automatically adjust the layout according to the screen width to keep the content clear and easy to read. For tablet devices, due to the larger screen, the system can provide richer multimedia content, such as embedded videos, detailed historical introductions or interactive maps to enhance the immersion of tourists. In the smart glasses mode, the system will use a projection information display method to superimpose the recommended content in the form of semi-transparent floating text or images in the actual field of vision of tourists, so that tourists can directly obtain the required information without looking down at the device during the tour. In addition, for devices that support virtual reality (VR) or augmented reality (AR), the system can also provide special formats of content such as panoramic images, 3D models or real-life guides to enhance interactivity.

[0093] In terms of interaction mode optimization, the terminal adaptation module 104 can automatically adjust the user operation mode according to the interactive characteristics of the device to ensure that tourists can obtain information in the most natural way on different terminals. For example, on a smartphone or tablet, the system uses a touch operation mode, allowing tourists to browse recommended content through gestures such as sliding, clicking, or pinching. On smart glasses or AR devices, the system provides voice control or eye tracking interaction, allowing tourists to navigate through voice commands or gaze retention options, thereby reducing the need for manual operation. On some devices that support gesture recognition, such as AR glasses or smart watches with depth cameras, the system can perform corresponding interface switching or information query operations based on the gestures of tourists, such as waving, tapping, or grabbing gestures. In addition, for tourists with limited mobility, the system also provides an interaction mode based on head movement or voice input to ensure barrier-free access to recommended content.

[0094] In terms of real-time performance adjustment, the terminal adaptation module 104 can dynamically adjust the loading method of recommended content according to the computing power and network conditions of the device to ensure a smooth user experience. For example, when it is detected that the tourist is using a low-power or storage-constrained device, the system will automatically reduce the resolution of the multimedia content, reduce dynamic effects, and use a lightweight data format to reduce the processing burden. When the network connection is unstable or in a low-bandwidth environment, the system will switch to offline mode, give priority to providing locally cached recommended content, and automatically update the guide information after the network is restored. For terminals that support edge computing, the system can assign some computing tasks to local devices for processing, such as voice recognition, gesture analysis, or path optimization, to reduce request delays from cloud servers and improve response speed.

[0095] The terminal adaptation module 104 works in cooperation with the recommendation generation module 103 to ensure that tourists can smoothly access personalized guided tour content on different devices. Through device feature recognition, content format optimization, interaction mode adjustment, and real-time performance management, this module can flexibly adjust the information display mode according to the terminal environment used by tourists, thereby enhancing the accessibility of information during the tour, improving the immersive experience, and ensuring the stability and efficiency of the system.

[0096] Furthermore, the terminal adaptation module automatically adjusts the presentation mode of personalized guidance data according to the type of terminal device used by tourists, where the type of terminal device includes smartphones, tablets, and smart glasses; the presentation modes include AR augmented reality display, 3D navigation mode, or standard list mode.

[0097] The terminal adaptation module is used to ensure that the intelligent tourism system can present personalized guidance data in the most suitable way according to the type of terminal device used by tourists, thereby enhancing the interaction experience and information acquisition efficiency of tourists. This module can identify the type of tourists' devices and dynamically adjust the presentation mode of guided tour information according to the screen size, computing power, input method, and interaction characteristics of the devices, enabling tourists to obtain the best usage experience on different devices.

[0098] When tourists use a smartphone as the terminal device, the system will preferentially use the standard list mode or the interactive map mode to display personalized guidance data. Due to the portability and high-frequency use characteristics of smartphones, the system will adaptively adjust the information layout according to the screen size to ensure the readability and operability of text, images, and interactive buttons. The system will use methods such as swiping to browse, pinching and zooming gestures, and hierarchical menus to enable tourists to quickly view the recommended scenic spots, navigation paths, and relevant guided tour information. For smartphones with touchscreens, the system can optimize the button size, text layout, and interactive area to reduce accidental touches and improve operation efficiency. In addition, when tourists are on an outdoor tour, the system will automatically adjust the interface color contrast according to the environmental brightness to enhance readability and ensure that tourists can clearly view the information in strong light or low light environments.

[0099] When tourists use tablet devices, the system will take advantage of the large screen of the tablet and adopt a more intuitive 3D navigation mode or split-screen display mode to provide a more diverse way of presenting information. In the 3D navigation mode, through virtual scenarios, tourists can view the scenic area layout, attraction distribution, and walking routes from a more immersive perspective. The system will use 3D modeling technology based on WebGL or a local rendering engine to generate a three-dimensional view of the scenic area in real time and overlay personalized recommendation information on the map, enabling tourists to intuitively understand the relative positions of attractions, recommended routes, and their current locations. When tourists click on an attraction, the system will display detailed introductions, including the history of the attraction, recommended tour routes, interactive content, etc., in the sidebar or a pop-up window. For tablet devices that support multi-window operations, the system can also display detailed recommendation information synchronously in the secondary screen or a floating window while showing the map navigation on the main interface, allowing tourists to view the required tour guide content at any time during the tour.

[0100] When tourists use smart glasses, the system will preferentially adopt the AR augmented reality display mode to provide a more immersive and intuitive tour guide experience. Smart glasses have the feature of overlaying virtual information on the real environment. The system will directly overlay personalized guiding data on the real scene based on the tourist's current perspective and location information. For example, when a tourist looks at a historical building, the system can display the name of the building, the construction year, historical background, and relevant interactive content within the field of view of the smart glasses. The system can dynamically adjust the presentation of information in combination with the tourist's head movement, gaze tracking, or voice commands to avoid information occlusion or overload. When tourists move to different attractions, the system will update the display content in real time to ensure that the recommended information matches the current tour scene. In addition, the gesture control and voice recognition functions of smart glasses can be used for interaction. For example, tourists can switch the recommended content through simple gesture operations or request a detailed explanation through voice commands, making the entire tour guide process more natural and smooth.

[0101] In practical applications, the system not only adapts based on the device type but also makes dynamic adjustments in combination with the device's computing power, storage space, and network connection status. For example, when the system detects that a tourist is using a low-power smartphone or a device with a small running memory, it will automatically reduce the picture resolution, decrease dynamic effects, and optimize the data loading method to ensure smooth operation. When the system detects that a tourist is in a low-network-speed environment, it will preferentially load the locally cached tour guide data or adopt a hierarchical loading strategy to display the key information first and supplement the detailed content after the network is restored. For tourists using 5G or high-performance computing devices, the system can provide higher-quality 3D rendering effects, high-definition video explanations, and richer interactive content.

[0102] The terminal adaptation module ensures that the smart tourism system can provide the best tour guide experience on different devices. Whether it is a smartphone, tablet or smart glasses, it can present personalized guiding data in the most suitable way, enabling tourists to obtain the required information more conveniently and enhancing the interactivity and immersion of the tour.

[0103] Furthermore, the terminal adaptation module dynamically adjusts the recommended content based on the interactive feedback of tourists to optimize the personalized experience, where the interactive feedback includes clicks, voice inputs or gesture operations.

[0104] After receiving and parsing the interactive feedback of tourists, the terminal adaptation module can dynamically adjust the recommended content to make the personalized guiding data more in line with the real-time needs and interest preferences of tourists. During the tour, tourists will interact with the system in various ways, including clicks, voice inputs or gesture operations. The system will capture these interactive behaviors in real time and adjust the recommendation strategy based on the feedback of tourists to provide more accurate and personalized tour guide information.

[0105] When using a smartphone, tablet or other touch devices, tourists usually interact with the system by clicking. When a tourist clicks on the detailed introduction of a scenic spot in the recommended list, the system will record this operation and analyze the tourist's interest tendency. If a tourist clicks on scenic spots of a certain category multiple times, such as historical sites or art galleries, the system will automatically increase the recommendation weight of scenic spots in this category, making the subsequent recommended tour routes more in line with the tourist's preferences. In addition, if a tourist repeatedly clicks on or stays on certain recommended content for a long time, the system will judge that this content may be a high-interest point for the tourist and actively provide more detailed tour guide information, such as the historical background of the exhibits, expert interpretations or recommended related scenic spots. On the contrary, if a tourist skips a certain category of scenic spots multiple times or quickly returns to the previous page, the system will reduce the recommendation weight of this category of scenic spots to avoid providing content that does not match the interests and optimize the accuracy of subsequent recommendations.

[0106] When using smart glasses or voice interaction devices, tourists can obtain the required information by voice input. The system can recognize the voice commands of tourists in real time, such as "Recommend nearby historical buildings", "Play the detailed explanation of this exhibit" or "Show the tour route of the next stop", and adjust the recommended content according to these commands. If a tourist requests the same category of tour guide information multiple times, such as continuously querying the background introductions of multiple cultural sites, the system will automatically adjust the recommendation strategy, give priority to pushing scenic spots or activities related to cultural sites, and provide in-depth interpretation content. In addition, the system can analyze the frequency and pattern of tourists' voice interactions, judge their preferred interaction methods, and dynamically optimize the content of voice responses. For example, for tourists who like concise introductions, the system will provide concise voice explanations, while for tourists who like detailed content, it will push richer interpretation information.

[0107] On devices that support gesture interaction, visitors can interact with the system through gesture operations. For example, they can wave their hands to switch recommended content, raise their hands to select scenic spot details, or request more information through specific gestures. The system will monitor these interaction behaviors in real time and, combined with the visitors' browsing trajectories and interest preferences, dynamically optimize the recommended content. If a visitor stays in front of a certain exhibit and makes the zoom-in gesture multiple times, the system can infer that they hope to obtain more detailed information about the exhibit, and thus automatically provide high-definition pictures, 3D models, or expert interpretations. If a visitor frequently swipes left to skip a certain type of recommended content during the tour, the system will adjust the recommendation algorithm, reduce the recommendation frequency of this type, and give priority to providing content that the visitor is more likely to be interested in, in order to enhance the personalization of the tour experience.

[0108] In the process of comprehensively processing these interaction feedbacks, the system will not only immediately adjust the visitors' current preferences, but also combine historical interaction data to form a long-term user interest model. In this way, when visitors use the system in different scenic spots or at different times, the recommended content can still maintain a high level of personalized accuracy. For example, if a certain visitor tends to view recommended content about natural landscapes during the tour of multiple scenic spots, the system will give priority to providing information about natural scenic spots such as forest parks, lakes, and waterfalls in subsequent tour suggestions, rather than recommending too many urban buildings or historical sites. In addition, the system can also further optimize the recommended content in combination with environmental factors such as crowd density and weather conditions, to ensure that visitors can not only obtain recommendations that match their interests, but also take the tour under the best environmental conditions.

[0109] Through the interaction feedback mechanism of the terminal adaptation module, the system can continuously optimize personalized recommendations, making the guided tour experience more intelligent, accurate, and efficient. Whether it is a click, voice input, or gesture operation, the system can analyze the visitors' preferences in real time and, by dynamically adjusting the recommended content, make the smart tourism system truly have a high degree of self-adaptability and personalized optimization ability.

[0110] Although this application is disclosed above in preferred embodiments, it is not intended to limit this application. Any person skilled in the art can make possible changes and modifications without departing from the spirit and scope of this application. Therefore, the protection scope of this application should be subject to the scope defined by the claims of this application.

Claims

1. A smart tourism operation media data management system, characterized in that: include: A data collection module is deployed at multiple physical nodes of a tourist destination, including RFID beacons, panoramic camera devices and environmental sensors, and is used to collect tourists' trajectory data, tourists' behavior data and environmental data; A data analysis module, used for receiving the data collected by the data collection module; Based on the collected data, the location and environmental information of tourists are aligned in time and space through time synchronization, and the points of interest of tourists are extracted using image recognition and natural language processing; Integrate the spatiotemporally aligned data and the extracted POI information to generate a structured tourist behavior model; A recommendation generation module, used to generate personalized guidance data according to the tourist behavior model; wherein the personalized guidance data includes tour routes that tourists may be interested in, related scenic spot recommendations and personalized interactive content; The terminal adaptation module is used to receive the personalized guidance data output by the intelligent recommendation generation module, and adjust the content presentation method based on the resolution, computing power and interaction method of the terminal device used by the tourists.

2. The smart tourism operation media data management system according to claim 1, characterized in that: The RFID beacon of the data collection module is dynamically bound to the visitor's identity device and generates trajectory data when the visitor enters or leaves a specific area. The trajectory data includes the visitor's current location, length of stay, and interactive behavior; The panoramic camera device of the data acquisition module extracts the tourists' points of interest in combination with the tourists' photo taking or code scanning behavior, and associates and analyzes the points of interest with the trajectory data recorded by the RFID beacon to improve the accuracy of the identification of points of interest; The environmental sensor of the data acquisition module is used to detect the temperature and humidity, crowd density and light intensity in the scenic area, and associate the environmental data with the tourist behavior data to analyze the tourists' sightseeing preferences under different environmental conditions.

3. The smart tourism operation media data management system according to claim 1, characterized in that: The data analysis module adopts a time-space alignment method based on a time window, and aligns the tourists' trajectory data with the environmental data through a time synchronization algorithm to improve the accuracy of the tourists' location and behavior information.

4. The smart tourism operation media data management system according to claim 1, characterized in that: The data analysis module combines image recognition and natural language processing to analyze tourists' photos, interactive texts and comment information, extract points of interest, and match tourists' preference categories based on a semantic association algorithm.

5. The smart tourism operation media data management system according to claim 1, characterized in that: The recommendation generation module combines tourists' historical behavior data, real-time trajectory and points of interest information, adopts behavior pattern analysis and multi-dimensional interest matching methods to predict tourists' potential interests and dynamically adjust personalized guidance data; and when it is detected that tourists stay at a designated attraction for a long time or return multiple times, it automatically pushes the background introduction, historical stories or interactive experience content of the attraction.

6. The smart tourism operation media data management system according to claim 1, characterized in that: The terminal adaptation module automatically adjusts the presentation method of personalized guidance data according to the type of terminal device used by the tourist, wherein the terminal device types include smart phones, tablets, and smart glasses; the presentation methods include AR augmented reality display, 3D navigation mode or standard list mode.

7. The smart tourism operation media data management system according to claim 1, characterized in that: The terminal adaptation module dynamically adjusts the recommended content based on the interactive feedback of the tourists to optimize the personalized experience, wherein the interactive feedback includes clicks, voice input or gesture operations.

8. The smart tourism operation media data management system according to claim 1, characterized in that: The recommendation generation module further combines tourist group behavior analysis to optimize tour route recommendations and tourist diversion scheduling. The group behavior analysis includes: The data analysis module identifies and classifies the group types of tourists based on the tourist trajectory data acquired by the data acquisition module, the RFID beacon data, and the tourist group behavior data collected by the panoramic camera device. The group types include independent tour groups, parent-child tour groups, group tour groups, and in-depth exploration groups. The recommendation generation module dynamically optimizes the recommended content based on the characteristic information of the group type, wherein: For autonomous tour groups, personalized tour routes are generated based on individual historical behavior data and interest characteristics, and optional tour path branches are provided to enhance tourists' freedom of choice; For parent-child tour groups, priority will be given to recommending attractions that are in line with children's interests and have educational value, while also considering the interests of parents to balance the tour needs of children and parents; For group tours, we provide recommended routes that fit the group's schedule, ensure the integration of group members' interests, and optimize the order of the group's scenic spots to visit in order to reduce conflicts in tour time; For in-depth exploration groups, priority is given to recommending attractions with longer explanations, detailed historical background introductions, and interactive experiences to meet tourists' needs for in-depth exploration of culture and knowledge; The recommendation generation module combines the real-time crowd density data collected by the environmental sensor, and when the density of tourists in the designated scenic spot or area exceeds the preset threshold, adjusts the tour recommendation strategy, wherein: Before tourists reach crowded areas, they are recommended alternative attractions that match their interests and have lower crowd density based on their preferences; After tourists have entered a high-density area, provide diversion suggestions within the attraction, including recommending tourists to enter a specific exhibition area or participate in a limited-time guided tour to alleviate local congestion; The recommendation generation module continuously monitors the behavior pattern changes of tourist groups, and dynamically adjusts the recommendation strategy when it detects that group members are dispersed, merged, or their sightseeing interests have changed significantly, wherein: When the group that originally toured together is split into independent individuals, the system resumes individualized recommendations and provides personalized tour routes; When multiple individuals form a new tour group, the system recalculates the comprehensive interests of the group and generates a new tour route based on the interest fusion strategy.

9. The smart tourism operation media data management system according to claim 1, characterized in that: The recommendation generation module adopts a dynamic personalized tour recommendation algorithm to calculate the personalized tour score of tourists for each scenic spot based on the tourists' historical behavior data, real-time trajectory data, environmental factors and points of interest information, and optimizes the recommended tour route according to the personalized tour score. The dynamic personalized tour recommendation algorithm uses the following formula 1 to calculate the personalized tour score of tourists: in, Recommended attractions interest score; is the impact weight of historical stay time; It is a historical and recommended attraction for tourists. The average length of stay at attractions of the same category; It is the maximum length of stay of tourists in all scenic spots in history; is the influence weight of the number of historical interactions; It is a historical and recommended attraction for tourists. The average number of interactions for attractions with the same category; is the maximum number of interactions among all attractions in the history of tourists; is the influence weight of interest matching; It is a feature of interest The importance weight of the feature in the overall interest; is a set of tourist interest features, defined as , where the set Each element in Indicates a tourist's interest feature. is the total number of tourists’ interest characteristics; It is based on semantic association or collaborative filtering method to calculate tourist interests Attraction theme The similarity between is the weight of crowding influence; is the adjustment factor; Recommended attractions Current crowd density; Based on the calculated interest score and real-time trajectory data of tourists , using the weighted traveling salesman problem solving method, using the objective function provided by the following formula 2, to obtain the tour route; in, The number of recommended attractions; is the path length penalty factor; Indicate attractions To the attractions walking distance.

10. The smart tourism operation media data management system according to claim 9, characterized in that: In the dynamic personalized tour recommendation algorithm, the interest feature weight Calculate according to the following formula 3: in, The total number of visitors' historical visit records; Historical visit records for visitors Interest characteristics The number of interactions with relevant content, including clicks, likes, and comments; The interest characteristics of tourists in all historical visit records The maximum number of interactions with the relevant content, used for normalization; Historical visit records for visitors Interest characteristics Duration of stay at relevant attractions; The interest characteristics of tourists in all historical visit records The maximum stay time of the relevant attractions, used for normalization; and is the impact weight of the number of interactions and the dwell time, which is used to measure their contribution to the intensity of interest; is the current time; For tourists The timestamp of the first visit to the relevant point of interest; It is the time decay factor, which controls the influence of historical behaviors and ensures that the contribution of earlier historical behaviors gradually decreases.

Images