Destination global tourism big data management platform based on data processing

Abstract

The present application belongs to the technical field of smart tourism, and discloses a destination global tourism big data management platform based on data processing, which comprises a sensing and clustering point module, collects global tourism data, constructs a geographic grid network, sets spatial grid reconstruction rules, and automatically refines or combines the grids; a trajectory triggering mechanism is introduced, and when the behavior frequency in the preset target area reaches the preset behavior frequency threshold, fine-grained modeling is enabled to generate a geographic sensing structure; a behavior deconstruction coding module analyzes the continuous behavior of tourists in the geographic sensing structure and constructs a multi-dimensional situation coding vector; through a behavior puzzle type deconstruction method and an implicit behavior compensation method, the complete behavior chain of tourists is reconstructed; and solid support is provided for in-depth analysis and intelligent service of global tourism data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of smart tourism, more specifically, the present application relates to a destination global tourism big data management platform based on data processing. BACKGROUND

[0002] The patent with the patent publication number CN119690943A discloses a data processing platform based on big data, the platform includes a business end and a processing end, the processing end includes an acquisition module, a storage module, a quality inspection module, a processing module, a scheduling module and a visualization module; the storage module is used for storing the structured data and the unstructured data in a preset database; the quality inspection module is used for obtaining the stored data in the preset database to determine the inspection condition; the processing module is used for determining whether the stored data needs to be processed by data conversion according to the inspection condition; the scheduling module is used for scheduling the stored data or the target data according to a preset time to generate a run batch task; the visualization module is used for visualizing the completed target run batch task to generate a target data table according to the run batch condition. The effect of reducing cost, simplifying process and meeting different data volume processing needs is achieved.

[0003] The existing destination global tourism big data management platform mainly has the following problems:

[0004] In the prior art, a fixed geometric boundary or management area is usually used as the coverage area of geographic grid division, and the geographic distribution information of global tourism data itself is not fully utilized, which may cause partial data omission or inclusion of irrelevant areas, reducing the effectiveness of the grid network. In the prior art, the length of the basic grid point is usually fixed and cannot be adjusted according to the data density of different areas. If the length is too small, the number of initial grid points is too large, and the computational complexity increases; if the length is too large, the fine differences in the area are lost, affecting the accuracy of subsequent analysis. In the prior art, the refinement or merging of grid points is usually based on static conditions, and there is a lack of mechanism for dynamic judgment according to real-time data characteristics, so that the grid network cannot efficiently adapt to data changes at different times and spatial ranges.

[0005] Currently, the resource exchange between tourists and merchants usually adopts a fixed weight-based or linear integral accumulation method, which fails to fully consider the differences in the immersion value points of tourists, leading to a lack of significant distinction between high-value tourists and low-value tourists, which may easily cause fairness disputes; at the same time, the resource request processing mechanism in the traditional platform fails to effectively combine the tourist activity and preference, which may easily cause a phenomenon of single high-activity tourist monopolizing a large amount of resources, reducing the satisfaction of other tourists; in addition, a simple first-come-first-served or round-robin strategy is usually adopted in the resource request process without introducing flow control and dynamic throttling measures, which may easily cause system congestion or resource allocation imbalance.

[0006] In view of this, the present application proposes a destination global tourism big data management platform based on data processing to solve the above problems. SUMMARY

[0007] In order to overcome the above-mentioned defects of the prior art, in order to achieve the above-mentioned purpose, the present application provides the following technical scheme: a destination global tourism big data management platform based on data processing, comprising:

[0008] The perception aggregation point module collects global tourism data and constructs a geographic grid network, sets a spatial grid reconstruction rule, and automatically refines or merges the grid points; a trajectory triggering mechanism is introduced, and when it is detected that the cumulative behavior frequency in the preset target area reaches the preset cumulative behavior frequency threshold, a fine-grained modeling is enabled to generate a geographic perception structure;

[0009] The behavior deconstruction coding module analyzes the continuous behavior of tourists in the geographic perception structure and constructs a multi-dimensional situation coding vector; through a behavior puzzle type deconstruction method and an implicit behavior compensation method, the complete behavior chain of the tourists is reconstructed;

[0010] The immersion value quantification module calculates the immersion value score of the tourists based on the multi-dimensional situation coding vector and the complete behavior chain, generates a resonance path matrix between the tourists by comparing the resonance similarity of different tourists on the complete activity path, and establishes a tourist behavior evolution graph;

[0011] The guide recommendation feedback module generates a personalized tour guide path for the tourists according to the immersion value score of the tourists and the tourist behavior evolution graph; a reverse resonance recommendation strategy is adopted, and the highest similarity but unvisited geographic grid point is pushed to the tourists in combination with the immersion value score of the tourists;

[0012] The platform response management module links the merchant resources around the geographic grid point with the highest resonance similarity; a resonance group set meal linkage mechanism is introduced to drive the publication of merchant resources in the preset destination, forming a resource sharing network;

[0013] The point exchange reward module maps the immersion value score of the tourists into consumable points to support the exchange behavior between the tourists and the merchants; a weighted fair queuing and leaky bucket flow limiting mechanism is adopted to allocate the merchant resources, dynamically balancing the tourist satisfaction and resource utilization rate.

[0014] Preferably, the method of automatically refining or merging the grid points comprises:

[0015] The global tourism data is collected, which includes tourist behavior data, environmental facility data, merchant service data, spatial geographic data and management policy data; the global tourism data is cleaned by a sliding window outlier detection technology, and is subjected to standard deviation normalization processing to obtain normalized global tourism data;

[0016] For the normalized global tourism data, a convex hull algorithm is used to extract the minimum convex boundary of all geographic coordinate points, and the required coverage area for constructing the geographic grid network is determined according to the minimum convex boundary; the required coverage area is preliminarily divided into spatial grids according to a preset basic grid side length , an adaptive quadtree division method is used to generate an initial geographic grid network, and for any point in the required coverage area, a mapping function is used to realize the encoding mapping of the geographic coordinate point to the geographic grid;

[0017] A dynamic adjustment strategy based on spatial density is introduced to dynamically adjust the basic grid side length, and the point density of any point in the required coverage area is calculated , and the basic grid side length is adjusted according to the point density by a grid side length adjustment function;

[0018] All geographic coordinate points in the global tourism data are encoded and classified into corresponding geographic grids; a spatial grid reconstruction rule is set, and for any grid in the geographic grid network, if there is a time period that satisfies the refinement trigger condition function within a preset time window, the grid is triggered to be automatically refined;

[0019] For any grid in the geographic grid network, if there is a neighboring grid that belongs to the neighboring grid set of the current arbitrary grid, and the cumulative behavior frequency in the neighboring grid is less than the cumulative behavior frequency in the grid, the grid is triggered to be automatically merged.

[0020] Preferably, the method for generating the geographic awareness structure comprises:

[0021] A trajectory trigger mechanism is introduced to collect and count the trajectory points in the global tourism data, and the cumulative behavior frequency of the trajectory points in a preset target area within a preset time period is calculated by a sliding time window;

[0022] It is judged whether the cumulative behavior frequency reaches a preset cumulative behavior frequency threshold, and if so, fine-grained modeling is triggered; the spatial division granularity is adjusted according to the trajectory point distribution density in the target area, including reducing the basic grid side length and deepening the quadtree division depth, to generate a finer-grained geographic grid network;

[0023] All trajectory points in the preset target area are remapped and encoded, and the trajectory points are classified into corresponding newly generated fine-grained grids; based on the remapped trajectory points and their corresponding fine-grained grids, a geographic awareness structure is generated, which includes grid encoding, spatial position information, trajectory point cumulative behavior frequency within a time period, and trajectory point density information.

[0024] Preferably, the construction method of the multi-dimensional context encoding vector comprises:

[0025] The trajectory points in the tourist behavior data in the global tourism data are time-sequentially sorted, the timestamp, spatial position information, grid point encoding corresponding to the trajectory points and behavior category of each trajectory point are extracted, and a continuous trajectory behavior sequence is formed;

[0026] For each trajectory behavior sequence, the spanning path, time period, stay duration, moving speed and direction change in the geographical perception structure are analyzed, and the behavior mode features in each time period are counted; multi-dimensional trajectory features are extracted according to the spatial dimension, time dimension, behavior dimension, social dimension and semantic dimension; the multi-dimensional trajectory features extracted are combined to generate a multi-dimensional context encoding vector by using feature splicing and hash encoding processing.

[0027] Preferably, the reconstruction method of the complete behavior chain of the tourist comprises:

[0028] The multi-dimensional context encoding vector sequence obtained by analysis is divided according to the time sequence and spatial adjacency relationship, and is divided into n behavior sub-segments, the behavior sub-segments reflect the local behavior features of the tourist within a limited time period and spatial range, and are referred to as behavior puzzles;

[0029] For the incomplete behavior sub-segments caused by data loss, sensing blind area or abnormal detection, an implicit behavior compensation method is used, based on the tourist behavior data and environmental facility data, a machine learning model is used to infer the missing behavior sub-segments and their time sequence and spatial position;

[0030] The behavior sub-segments and the behavior sub-segments obtained by implicit behavior compensation are spliced based on the time sequence and spatial adjacency relationship, and the multi-dimensional trajectory features corresponding to the multi-dimensional context encoding vector are combined to construct the complete behavior chain of the tourist.

[0031] Preferably, the establishment method of the tourist behavior evolution map comprises:

[0032] According to the multi-dimensional context encoding vector and the complete behavior chain of the tourist obtained by analysis in the geographical grid point network, a weighted vector aggregation technology is used to calculate the immersion value score of the tourist, and the immersion value score of the tourist comprehensively reflects the behavior activity, behavior type, stay duration and emotional participation of the tourist in the time and space dimensions;

[0033] The behavior value is calculated comprehensively by combining the behavior category weight and the cumulative behavior frequency statistics, the tourist comment feedback data is collected, an emotion analysis model is introduced to score the emotions of the tourist comment feedback data, and the immersion value score of the tourist is generated;

[0034] A resonance path matrix is constructed, a dynamic time warping algorithm is used to match multi-dimensional situation encoding vector sequences of different tourists, and path similarity is calculated; tourist behavior data is time-aligned, and Euclidean distances between multi-dimensional feature vectors corresponding to multi-dimensional situation encoding vectors are calculated as local distances; a complete activity path is constructed based on the local distances, and a resonance similarity score of a tourist on the complete activity path is output;

[0035] A tourist behavior evolution graph is constructed according to the constructed resonance path matrix, the tourist behavior evolution graph is a weighted directed graph, each node of the tourist behavior evolution graph represents each tourist, and an edge represents a behavior relationship connection between two tourists; and a weight of the edge represents a resonance similarity score of a tourist on a complete activity path.

[0036] Preferably, the method for generating the personalized tour guide path comprises:

[0037] Using the tourist behavior evolution graph represented by the weighted directed graph, behavior similarity of any preset target tourist with other tourists in the tourist behavior evolution graph is analyzed, a neighbor tourist group with the most similar behavior mode is identified, and a tour route that the target tourist is likely to be interested in is inferred based on a historical behavior evolution path of the neighbor tourist group;

[0038] Combining environmental facility data and tourist behavior data, an A* algorithm is used to obtain an optimal tour path that meets the interest preferences and path coherence of tourists, which is defined as a personalized tour guide path; the generated personalized tour guide path is fed back to a tourist mobile terminal in the form of multi-dimensional situation encoding, and real-time navigation and behavior suggestions are provided.

[0039] Preferably, the method for pushing the geographical grid point with the highest similarity but not visited comprises:

[0040] A neighbor tourist group whose resonance similarity score on a complete activity path is greater than a resonance similarity score threshold of a preset tourist on the complete activity path is screened out; a geographical grid point that a target tourist has not visited but that is visited by the neighbor tourist group is identified as a recommended candidate grid point according to a complete behavior chain of the preset target tourist and a complete behavior chain of the neighbor tourist group;

[0041] An immersion value score of the target tourist is introduced as a weight factor, a resonance similarity score of a tourist on a complete activity path and an immersion value score of the tourist are comprehensively evaluated by a weighted vector aggregation method, a recommended priority ranking model is formed, and for all recommended candidate grid points, a geographical grid point with the highest interest matching degree and the highest immersion value score with respect to a preset target tourist is preferentially recommended;

[0042] A dynamic recommendation mechanism is designed to dynamically generate a list of recommended geographic grid points based on changes in tourist behavior data and their immersion value scores. Push notification technology is used to send the latest recommendation results to tourists' mobile terminals in real time, assisting tourists in making personalized tour plans.

[0043] Preferably, the method for forming the resource sharing network includes:

[0044] The core resonance node is determined by the geographical grid area with the highest resonance similarity score of tourists along the complete activity path. Based on spatial database technology, the core resonance node is spatially mapped with the merchant resources in the preset destination to establish the corresponding relationship between the geographical location and the merchant resources.

[0045] Through an event-driven architecture, combined with an LSTM time series prediction model, the activation status of resonance group packages is monitored in real time, resource scheduling instructions are automatically pushed to relevant merchants, and elastic expansion is achieved using a microservice architecture. After receiving the instructions, the merchant-side resource management system completes inventory adjustments, staff scheduling, and service preparation, achieving dynamic resource response.

[0046] A graph database is used to construct a resource-sharing network covering preset target geographic grid points and merchants in the surrounding areas. The nodes in the resource-sharing network represent merchant resources, and the edges represent resource-sharing cooperation relationships. A big data processing platform is used to collect tourist consumption data, merchant service feedback data, and user evaluation data in real time. The BERT sentiment classifier is used to evaluate service quality, and the resonance group package design and resource scheduling strategy are continuously optimized through multi-dimensional data fusion and causal inference methods.

[0047] Preferably, the method for dynamically balancing visitor satisfaction and resource utilization includes:

[0048] The tourist immersion value points are mapped into consumable points through a nonlinear weighted mapping function, and the nonlinear weighted mapping function is defined as ;in, For tourists Spendable points; For tourists Immersion value points; An adjustment factor to control the overall expansion or reduction of consumable points; The adjustment factor for controlling the shape of the consumable integral mapping curve; The base of consumable points is used to ensure that all tourists have a minimum consumable points value;

[0049] Will tourists Spendable points The integral exchange reward module is stored, and the consumable integral balance of the visitor is updated; when the visitor initiates a resource exchange request with a merchant, the platform adopts a weighted fair queuing mechanism, sorts the visitors in descending order according to their consumable integrals, processes the visitor requests in descending order of the consumable integrals, and allocates available merchant resources;

[0050] The platform calculates the resource share that each visitor can obtain according to the current total amount of merchant resources and the consumable integrals; a leaky bucket flow control mechanism is introduced to control the resource request flow, and control parameters of the resource request flow are set, including the leaky bucket capacity and the request outflow rate, so that the visitor request rate is less than the request outflow rate, and the number of requests in the leaky bucket does not exceed the leaky bucket capacity, thereby dynamically balancing the visitor satisfaction and resource utilization.

[0051] Compared with the prior art, the present application has the following beneficial effects:

[0052] In the construction process of the initial geographic grid network, a dynamic adjustment mechanism is introduced based on the spatial distribution density of global tourism data, the density level of data points in each region is calculated, and the grid length of the corresponding region is automatically adjusted, so that smaller grids are used in data-intensive regions to enhance the precision, and larger grids are used in data-sparse regions to reduce the computational burden. The problems of low computational efficiency or loss of spatial information caused by fixed grid size in the prior art are effectively overcome, and the adaptability and processing performance of the platform are significantly improved.

[0053] A grid reconstruction rule based on spatial data change dynamic triggering is proposed, which covers two reconstruction strategies of refinement and merging. In the aspect of grid refinement, a refinement trigger judgment method that comprehensively considers the cumulative frequency and statistical variance of behavior data is adopted, so that the spatial grid can automatically refine in high-frequency areas according to actual passenger flow, activity density and other dynamic factors, thereby improving the spatial resolution accuracy. In the aspect of grid merging, the cumulative amount of behavior data of adjacent grids is compared, and low-activity adjacent grids are automatically merged to reduce unnecessary grid subdivision and computational burden. Compared with the single and static grid division method in the prior art, the adaptive dynamic reconstruction strategy of this scheme can respond to the changes of global tourism data in real time, significantly improving the real-time performance and flexibility of the platform. The adaptability, computational efficiency and spatial resolution accuracy of the spatial grid network are significantly improved, which provides a solid support for in-depth analysis and intelligent service of global tourism data.

[0054] The immersion value of the tourists is mapped into consumable points by a nonlinear weighted mapping function, the overall amplification and reduction and the curve shape of the points are controlled by an adjustment factor, the difference in points between the tourists with high immersion value and the tourists with low immersion value is effectively amplified, and the fairness and the discrimination degree of the point distribution are ensured;By setting the point base, it is ensured that all tourists obtain the basic points, and the fairness problem caused by the zero point is avoided. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 The figure is a structural schematic diagram of a destination global tourism big data management platform based on data processing according to the present application;

[0056] Figure 2 The figure is a flowchart of a method for generating a geographic perception structure according to the present application;

[0057] Figure 3 The figure is a flowchart of a method for managing destination global tourism big data based on data processing according to the present application. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0059] Embodiment one

[0060] Please refer to Figure 1 and Figure 2 Embodiment one further illustrates the destination global tourism big data management platform based on data processing according to the present application, including:

[0061] With the continuous development of global tourism concept, the acquisition, processing and management of destination tourism data have become an important means to improve the quality of tourism service and tourist experience. The existing global tourism big data management platform of the destination is usually based on fixed geometric boundary or administrative region to divide the geographical grid, which is used to cover different tourism data regions. However, this division method does not fully utilize the geographical distribution characteristics of global tourism data itself, which may cause some data to be missed or irrelevant regions to be included, thereby reducing the coverage effectiveness of the geographical grid network and the accuracy of data processing.

[0062] In the prior art, the side length setting for constructing the basic grid point is usually fixed, which cannot be dynamically adjusted according to the density of tourism data in different regions. When the side length is too small, the number of initial grid points will be too large, which increases the complexity of subsequent data processing and calculation; when the side length is too large, the fine-grained differences of the tourism data in the region will be lost, which affects the accuracy and reliability of the subsequent analysis results. In addition, the refinement or merging processing of the grid point in the prior art usually only relies on static condition judgment, lacks a mechanism for dynamic adjustment and judgment according to real-time data characteristics, and the grid network cannot flexibly adapt to data changes in different time and space ranges, which affects the adaptability and efficiency of the system in dynamic tourism scenarios.

[0063] In the existing resource exchange mechanism between tourists and merchants, fixed weight or simple linear integration accumulation method is usually used, which fails to fully reflect the difference in immersion value generated by tourists in the tourism process, resulting in a lack of significant distinction between high-immersion-value tourists and low-immersion-value tourists, which easily causes fairness problems. The mechanism for processing tourist resource requests in the traditional platform often fails to fully combine the activity and interest preferences of tourists, which easily causes some high-activity tourists to occupy a large amount of resources for a long time, reducing the participation and satisfaction of other tourists. At the same time, the processing method of resource request usually adopts first-come-first-served or simple polling strategy, which lacks effective flow control and dynamic throttling measures, and easily causes problems such as system congestion, request delay and resource allocation imbalance, which limits the stability and response efficiency of the global tourism data platform under high load conditions.

[0064] In order to effectively solve the above problems, the present application proposes a global tourism big data management platform of destination based on data processing, which comprises:

[0065] The perception aggregation grid point module collects global tourism data and constructs a geographical grid network, sets a spatial grid point reconstruction rule, and automatically refines or merges the grid points; a trajectory triggering mechanism is introduced, when the cumulative behavior frequency in the preset target region reaches the preset cumulative behavior frequency threshold, the fine-grained modeling is enabled to generate a geographical perception structure;

[0066] The behavior deconstruction coding module analyzes the continuous behavior of the tourists in the geographical perception structure, and constructs a multi-dimensional situation coding vector; the behavior deconstruction coding module reconstructs the complete behavior chain of the tourists through a behavior puzzle deconstruction method and an implicit behavior compensation method;

[0067] The immersion value quantification module calculates the immersion value score of the tourists based on the multi-dimensional situation coding vector and the complete behavior chain, generates a resonance path matrix between the tourists by comparing the resonance similarity of different tourists on the complete activity path, and establishes a tourist behavior evolution graph;

[0068] The guide recommendation feedback module generates a personalized tour guide path for the tourists according to the immersion value score of the tourists and the tourist behavior evolution graph, and pushes the highest similarity geographical grid that has not been visited to the tourists by using a reverse resonance recommendation strategy combined with the immersion value score of the tourists;

[0069] The platform response management module links the merchant resources around the geographical grid with the highest resonance similarity, and introduces a resonance group package linkage mechanism to drive the publication of the merchant resources in the preset destination, thereby forming a resource sharing network;

[0070] The integral exchange reward module maps the immersion value score of the tourists into consumable points to support the exchange behavior between the tourists and the merchants, and allocates the merchant resources by using a weighted fair queuing and leaky bucket flow limiting mechanism to dynamically balance the tourist satisfaction and the resource utilization rate.

[0071] The method for automatically refining or merging the grids comprises the following steps:

[0072] Collecting global tourism data, wherein the global tourism data comprises tourist behavior data, environmental facility data, merchant service data, spatial geographical data and management policy data; cleaning the global tourism data by using a sliding window outlier detection technology, and performing standard deviation normalization processing to obtain normalized global tourism data;

[0073] The tourist behavior data comprises the moving track of the tourists in the destination, the access record of the gate of the scenic spot, the mobile communication base station signaling data (position and stay time) and the electronic ticket usage data; the environmental facility data comprises traffic condition data, environmental monitoring data and scenic spot facility state data; the traffic condition data comprises public transportation operation data, road traffic data, parking lot data and tourist transfer point data;

[0074] The environmental monitoring data comprises meteorological data, air quality data, water quality data and noise monitoring data; the scenic spot facility state data comprises tourist service facilities, scenic spot opening information, scenic spot equipment operating conditions and intelligent navigation facilities; the merchant service data comprises merchant business data and tourist consumption record data; the spatial geographical data comprises scenic spot topographic and geomorphic data and road network data; the management policy data comprises tourism policy announcements and flow limiting measures data;

[0075] For the normalized global tourism data, a convex hull algorithm is used to extract the minimum convex boundary of all geographic coordinate points, and the required coverage area for constructing the geographic grid network is determined according to the minimum convex boundary; the required coverage area is preliminarily divided into spatial grids according to a preset basic grid side length ; an adaptive quadtree division method is used to generate an initial geographic grid network, and for any point in the required coverage area, an encoding mapping of the geographic coordinate point to the geographic grid is realized through a mapping function; the mapping function is ; wherein, represents the mapping function of the arbitrary point to the corresponding geographic grid code; represents a grid code connection operation;

[0076] However, the prior art does not consider the reasonable setting of the basic grid side length, and unreasonable setting of the basic grid side length may result in excessive number of initial grids (increasing the calculation overhead) or insufficient number of initial grids (losing fine information). A dynamic adjustment strategy based on spatial density is introduced to dynamically adjust the basic grid side length, and the point density of any point in the required coverage area is calculated ; wherein, represents the total number of points within a radius around the arbitrary point ; according to the point density , the basic grid side length is adjusted through a grid side length adjustment function;

[0077] The grid side length adjustment function is: ; wherein, represents the grid side length corresponding to the arbitrary point after adjustment; represents a preset allowed minimum grid side length; represents a preset allowed maximum grid side length; represents a decay factor, which controls the influence degree of the point density on the grid side length adjustment; according to the expert experience method, the value range of is between 0 and 1;

[0078] By encoding all geographic coordinate points in the global tourism data, they are classified into corresponding geographic grids; a spatial grid reconstruction rule is set, and for any grid in the geographic grid network, if there is a time period that satisfies the refinement trigger condition function within a preset time window, the grid is triggered to be automatically refined;

[0079] The refinement trigger condition function is ; wherein, denotes the refinement trigger decision result for the i-th grid point in the geographical grid network; denotes the cumulative behavior frequency of the grid point within the time period denotes the preset maximum behavior frequency of the grid point; denotes the statistical variance of the behavior frequency; denotes the maximum behavior frequency factor, used to adjust the influence degree of the maximum behavior frequency on the trigger condition; denotes the variance factor, used to adjust the influence degree of the variance term on the trigger condition; denotes all time periods within the time window; according to the expert experience method, and the value range of is between 0 and 1; denotes the index of the grid point;

[0080] For any grid point in the geographical grid network, if there is an adjacent grid point belonging to the adjacent grid point set of the current any grid point, and the cumulative behavior frequency in the adjacent grid point is less than the cumulative behavior frequency in the grid point, then the automatic merging of the grid point is triggered.

[0081] For example, taking a certain coastal global tourism demonstration area (covering an area of about 1500 square kilometers) as an example, the area includes a main urban area, multiple key tourism scenic spots (such as ancient towns, coastal resorts, and mountain parks), rural tourism sites, and nature reserves. The area receives more than 30 million tourists per year, and the distribution and flow of tourists show obvious spatiotemporal dynamic characteristics.

[0082] In order to realize the accurate dynamic management of tourist flow, environmental pressure and service supply in the area, the tourism management platform accesses the following multi-source global tourism data:

[0083] Tourist behavior data: including real-time positioning trajectories of about 500,000 mobile devices, consumption payment data, and scenic spot access records; environmental facility data: covering 300 public service facilities (including parking lots, toilets, and scenic service stations), about 60 environmental monitoring points (air quality and noise monitoring); merchant service data: about 3,500 tourism-related merchants' business status, service capacity, and consumer frequency data; spatial geographic data: high-precision topographic maps, street maps, and road network information; management policy data: including time-limited flow, temporary traffic control during holidays, and emergency evacuation plans.

[0084] ​​​​The convex hull algorithm is used to automatically extract the minimum boundary range (an area of about 1200 square kilometers) of the global tourism area, and an adaptive quadtree division method is used to construct an initial geographic grid network according to a preset basic grid length (initially set to 300 meters). Further, the basic grid length is automatically adjusted based on the spatial density (the point density is about 50-4000 points / square kilometer).

[0085] For example, in the ancient town scenic area (an area of about 15 square kilometers), the point density of the high-density tourist gathering area reaches about 3500 points / square kilometer, and the platform automatically reduces the grid length to 50 meters to achieve fine monitoring of tourist flow and service status.

[0086] In the nature reserve (an area of about 200 square kilometers), the point density is only about 100 points / square kilometer, and the platform dynamically expands the grid length to 500 meters to reduce the platform computing overhead while maintaining the necessary monitoring accuracy.

[0087] In actual operation, when a holiday (such as the "National Day Golden Week") arrives, the platform detects that the cumulative behavior frequency of some grids in the core area of the ancient town scenic area within a 30-minute window exceeds 15,000 times, and the behavior frequency variance is high, which automatically triggers grid refinement to generate smaller granularity (such as 25 meters) sub-grids to facilitate targeted guidance of tourist flow and adjustment of temporary service measures (such as adding temporary passages, opening emergency parking points, etc.).

[0088] At the same time, in the nature reserve and some unpopular scenic areas, the platform finds that the cumulative behavior frequency of multiple adjacent grids in the same time period is lower than that of adjacent grids (such as the cumulative behavior frequency is only 30% of the main grid), and the platform automatically triggers merging, and the merged grid length is expanded to 700 meters, reducing the platform burden and avoiding invalid data occupying processing resources.

[0089] Through the method of the application, the platform realizes real-time perception, dynamic response and spatial management optimization of tourist flow, providing a wide coverage, adjustable granularity, timely response and efficient computing solution for management, especially suitable for global tourism management scenarios with extensive coverage and complex tourist distribution.

[0090] The problems existing in the prior art are solved: in the prior art, fixed geometric boundaries or management areas are usually used as the coverage area of geographic grid division, the geographic distribution information of the global tourism data itself is not fully utilized, which may cause part of the data to be missed or irrelevant areas to be included, and the effectiveness of the grid network is reduced. In the prior art, the length of the basic grid point is usually fixed and cannot be adjusted according to the data density of different regions. If the length is too small, the number of initial grid points is too large, and the computational complexity is increased; if the length is too large, the fine differences in the region are lost, and the accuracy of subsequent analysis is affected. In the prior art, the refinement or merging of the grid points is usually based on static conditions, and there is a lack of a mechanism for dynamic determination according to real-time data characteristics, so that the grid network cannot efficiently adapt to data changes in different time and space ranges.

[0091] The beneficial effects of the prior art are: in the construction process of the initial geographic grid network, a dynamic adjustment mechanism is introduced based on the spatial distribution density of the global tourism data, the density level of the data points in each region is calculated, and the grid length of the corresponding region is automatically adjusted, so that smaller grids are used in data-intensive regions to enhance the fineness, and larger grids are used in data-sparse regions to reduce the computational burden. The problems of low computational efficiency or loss of spatial information caused by fixed grid size in the prior art are effectively overcome, and the adaptability and processing performance of the platform are significantly improved.

[0092] A grid reconstruction rule based on spatial data changes is proposed, which covers two reconstruction strategies of refinement and merging. In the aspect of grid refinement, a refinement trigger judgment method that comprehensively considers the cumulative frequency and statistical variance of behavior data is adopted, so that the spatial grid can automatically refine in high-frequency areas according to actual passenger flow, activity density and other dynamic factors, thereby improving the spatial resolution accuracy. In the aspect of grid merging, the cumulative amount of behavior data of adjacent grid points is compared, and low-activity adjacent grid points are automatically merged to reduce unnecessary grid subdivision and computational burden. Compared with the single and static grid division method in the prior art, the adaptive dynamic reconstruction strategy of this scheme can respond to changes in global tourism data in real time, significantly improving the real-time performance and flexibility of the platform. The adaptability, computational efficiency and spatial resolution accuracy of the spatial grid network are significantly improved, providing a solid foundation for in-depth analysis and intelligent services of global tourism data.

[0093] The method for generating a geographic perception structure comprises:

[0094] A trajectory trigger mechanism is introduced to collect and count the trajectory points of the tourist behavior data in the global tourism data, and the cumulative behavior frequency of the trajectory points in a preset target region within a preset time period is calculated through a sliding time window;

[0095] determining whether the accumulated behavior frequency reaches a preset accumulated behavior frequency threshold, and if so, triggering fine-grained modeling; adjusting the spatial division granularity according to the trajectory point distribution density in the target area, including reducing the basic grid side length and deepening the quadtree division depth, to generate a finer-grained geographic grid network;

[0096] re-mapping and encoding all trajectory points in the preset target area, classifying the trajectory points into corresponding newly generated fine-grained grids; based on the re-mapped trajectory points and their corresponding fine-grained grids, generating a geographic perception structure containing grid encoding, spatial location information, trajectory point accumulated behavior frequency in a time period, and trajectory point density information.

[0097] The construction method of the multi-dimensional context encoding vector includes:

[0098] sequentially ordering the trajectory points in the tourist behavior data in the global tourism data, extracting the timestamp, spatial location information, grid encoding corresponding to the trajectory point, and behavior category of each trajectory point to form a continuous trajectory behavior sequence;

[0099] For each trajectory behavior sequence, analyze the crossing path, time period, stay duration, moving speed, and direction change in the geographic perception structure, and count the behavior pattern characteristics in each time period; extract multi-dimensional trajectory features according to spatial dimensions (including geographic grid encoding, jump sequence, relative position information), time dimensions (including timestamp, time interval, stay / moving duration), behavior dimensions (including behavior category, frequency, intensity), social dimensions, and semantic dimensions; use feature splicing and hash encoding processing to combine the extracted multi-dimensional trajectory features to generate a multi-dimensional context encoding vector.

[0100] The reconstruction method of the complete behavior chain of the tourist includes:

[0101] divide the analyzed multi-dimensional context encoding vector sequence into n behavior sub-segments according to the chronological order and spatial adjacency relationship, the behavior sub-segments reflecting the local behavior characteristics of the tourist within a limited time period and spatial range, referred to as behavior puzzles;

[0102] For incomplete behavior sub-segments caused by data loss, sensing blind area, or abnormal detection reasons, use an implicit behavior compensation method to infer the missing behavior sub-segments and their time sequence and spatial location based on the tourist behavior data and environmental facility data using a machine learning model;

[0103] splice the behavior sub-segments and the behavior sub-segments obtained through implicit behavior compensation based on the chronological order and spatial adjacency relationship, combine the multi-dimensional trajectory features corresponding to the multi-dimensional context encoding vector, and construct the complete behavior chain of the tourist; the complete behavior chain includes the moving trajectory of the tourist from entering the preset target area to leaving.

[0104] The method for establishing the evolution map of tourist behavior comprises the following steps:

[0105] According to the multi-dimensional situation coding vector and the complete behavior chain obtained by analyzing the tourist in the geographical grid network, the immersion value score of the tourist is calculated by using the weighted vector aggregation technology, and the immersion value score of the tourist comprehensively reflects the behavior activity, behavior type, stay duration and emotional participation of the tourist in the time and space dimensions;

[0106] The behavior value is calculated by combining the behavior category weight and the cumulative behavior frequency statistics, the tourist comment feedback data is collected, the sentiment analysis model is introduced to score the emotions of the tourist comment feedback data, and the immersion value score of the tourist is generated;

[0107] A resonance path matrix is constructed, the multi-dimensional situation coding vector sequence of different tourists is matched by using the dynamic time warping algorithm, the path similarity is calculated, the behavior data of the tourist is time-aligned, the Euclidean distance between the multi-dimensional feature vectors corresponding to the multi-dimensional situation coding vector is calculated as a local distance, a complete activity path is constructed based on the local distance, and a resonance similarity score of the tourist on the complete activity path is output;

[0108] According to the constructed resonance path matrix, an evolution map of tourist behavior is constructed, the evolution map of tourist behavior is a weighted directed graph, each node of the evolution map of tourist behavior represents each tourist, and an edge represents a behavior relationship connection between two tourists; the weight of the edge represents the resonance similarity score of the tourist on the complete activity path.

[0109] The method for generating the personalized tour guide path comprises the following steps:

[0110] The evolution map of tourist behavior represented by the weighted directed graph is used to analyze the behavior similarity of any preset target tourist with other tourists in the evolution map of tourist behavior, identify the most similar adjacent tourist group in behavior mode, and infer the tour route that the target tourist may be interested in based on the historical behavior evolution path of the adjacent tourist group;

[0111] The A* algorithm is used to obtain the optimal tour path that meets the interest preferences and path continuity of the tourist by combining the environmental facility data and the behavior data of the tourist, the optimal tour path is defined as the personalized tour guide path, the generated personalized tour guide path is fed back to the mobile terminal of the tourist in the form of multi-dimensional situation coding, and real-time navigation and behavior suggestions are provided.

[0112] The method for pushing the geographical grid point with the highest similarity but not visited comprises the following steps:

[0113] Screening out the adjacent tourist group whose resonance similarity score on the complete activity path is greater than the preset resonance similarity score threshold of the complete activity path of the tourists; according to the complete behavior chain of the preset target tourist, marking the geographical grid points visited by the target tourist, combining the complete behavior chain of the adjacent tourist group, identifying the geographical grid points not visited by the target tourist but visited by the adjacent tourists as recommended candidate grid points;

[0114] Introducing the immersion value score of the target tourist as a weight factor, and through a weighted vector aggregation method, the resonance similarity score of the tourists on the complete activity path and the immersion value score of the tourists are comprehensively evaluated to form a recommendation priority ranking model. For all recommended candidate grid points, the geographical grid point with the highest interest matching degree and the highest immersion value score of the preset target tourist is preferentially recommended;

[0115] Designing a dynamic recommendation mechanism, dynamically generating a recommended geographical grid point list according to the changes of the tourist behavior data and the immersion value score of the tourists, and using a push notification technology to send the latest recommended results to the mobile terminal of the tourists in real time to assist the tourists in personalized tour planning.

[0116] The method for forming a resource sharing network comprises:

[0117] The geographical grid point area with the highest resonance similarity score of the tourists on the complete activity path is determined as a core resonance node; based on a spatial database technology, the core resonance node is spatially mapped with the merchant resources in the preset destination to establish a corresponding relationship between the geographical location and the merchant resources;

[0118] Through an event-driven architecture, combining an LSTM time series prediction model to monitor the activation state of the resonance group package in real time, automatically pushing resource scheduling instructions to related merchants, and using a micro-service architecture for elastic expansion; after receiving the instructions, the resource management system on the merchant side completes inventory adjustment, personnel scheduling and service preparation, and realizes dynamic resource response;

[0119] A graph database is used to construct a resource sharing network covering the preset target geographical grid points and the surrounding area merchants, the nodes in the resource sharing network represent the merchant resources, and the edges represent the resource sharing cooperation relationship; a big data processing platform is used to collect tourist consumption data, merchant service feedback data and user evaluation data in real time; a BERT sentiment classifier is used to evaluate the service quality, and a multi-dimensional data fusion and causal inference method is used to continuously optimize the resonance group package design and resource scheduling strategy.

[0120] The method for dynamically balancing the tourist satisfaction and the resource utilization rate comprises:

[0121] The immersion value score of the tourists is mapped to consumable points through a nonlinear weighted mapping function, and the nonlinear weighted mapping function is defined as ; wherein, is the immersion value score of the tourists consumable points; immersion value points of the visitors a regulation factor for controlling the overall amplification or reduction of the consumable points, according to the expert experience method, the value range of which is 0 to 1; a regulation factor for controlling the mapping curve shape of the consumable points, to realize increasing or decreasing nonlinear mapping, so as to reflect the greater difference in the points of the visitors with high immersion value points; a consumable point base, to ensure that all visitors have the minimum consumable point value;

[0122] the consumable points of the visitors are stored in the point exchange reward module, and the consumable point balance of the visitors is updated; when the visitors initiate a resource exchange request with a merchant resource, the platform adopts a weighted fair queuing mechanism to sort the visitors in descending order according to their consumable points, and processes the visitor requests and allocates available merchant resources in descending order of the consumable points;

[0123] The platform calculates the resource share that each visitor can obtain according to the current total amount of merchant resources and consumable points; a leaky bucket flow control mechanism is introduced to control the resource request flow, and control parameters of the resource request flow are set, including the leaky bucket capacity and the request outflow rate, to ensure that the visitor request rate is less than the request outflow rate, and the number of requests in the leaky bucket does not exceed the leaky bucket capacity, so as to dynamically balance the visitor satisfaction and resource utilization.

[0124] The following technical problems existing in the prior art are solved: At present, the resource exchange between visitors and merchants usually adopts a fixed weight-based or linear point accumulation-based mode, which fails to fully consider the difference in the immersion value points of the visitors, resulting in a lack of significant distinction between high-value visitors and low-value visitors, which easily causes fairness disputes; at the same time, the resource request processing mechanism in the traditional platform fails to effectively combine the visitor activity and preference, which easily causes a phenomenon that a single high-activity visitor occupies a large amount of resources, thereby reducing the satisfaction of other visitors; in addition, a simple first-come-first-served or round-robin strategy is usually adopted in the resource request process, without introducing flow control and dynamic flow control measures, which easily causes system congestion or resource allocation imbalance.

[0125] The beneficial effects of the present application relative to the prior art are as follows: the immersion value points of the visitors are mapped to consumable points through a nonlinear weighted mapping function, the overall amplification or reduction and the curve shape of the points are controlled by using a regulation factor, the difference in the points between high-immersion-value visitors and low-immersion-value visitors is effectively amplified, and the fairness and distinction degree of point allocation are ensured;

[0126] ​​By setting the integral base, it ensures that all tourists obtain basic points, avoiding the fairness problem caused by zero points. By using the weighted fair queuing mechanism, the tourists' consumption points are sorted to process resource requests, taking into account the tourists' activity and immersion experience, and realizing the rationality and fairness of the priority of resource request processing. By introducing the leaky bucket flow control mechanism, by setting the leaky bucket capacity and request flow rate, it ensures that the tourist request flow is controlled and does not exceed the system carrying capacity, realizes the system load balancing, and avoids the response delay caused by resource overload.

[0127] The preset cumulative behavior frequency threshold is set by the staff, and the average value of multiple cumulative behavior frequencies is taken as the preset cumulative behavior frequency threshold by collecting different cumulative behavior frequencies; similarly, the preset resonance similarity score threshold of the tourists on the complete activity path is set.

[0128] In the construction process of the initial geographic grid network, a dynamic adjustment mechanism is introduced based on the spatial distribution density of global tourism data. By calculating the density level of data points in each region, the grid length of the corresponding region is automatically adjusted, so that smaller grids are used in data-intensive areas to enhance the precision, and larger grids are used in data-sparse areas to reduce the computational burden. Effectively overcome the problem of low computational efficiency or loss of spatial information caused by fixed grid size in the prior art, significantly improve the adaptability and processing performance of the platform.

[0129] A grid reconstruction rule based on spatial data change dynamic triggering is proposed, which covers two reconstruction strategies of refinement and merging. In terms of grid refinement, a refinement trigger judgment method that comprehensively considers the cumulative frequency and statistical variance of behavior data is adopted, so that the spatial grid can automatically refine in high-frequency areas according to actual passenger flow, activity density and other dynamic factors, thereby improving the spatial resolution accuracy. In terms of grid merging, the behavior data accumulation of adjacent grids is compared, and low-activity adjacent grids are automatically merged to reduce unnecessary grid subdivision and computational burden. Compared with the single and static grid division method in the prior art, the adaptive dynamic reconstruction strategy of this scheme can respond to the changes of global tourism data in real time, significantly improving the real-time performance and flexibility of the platform. Significantly improve the adaptability, computational efficiency and spatial resolution accuracy of the spatial grid network, and provide a solid support for in-depth analysis and intelligent services of global tourism data.

[0130] The immersion value points of the tourists are mapped into consumable points by a nonlinear weighted mapping function, the overall amplification and reduction and the curve shape of the points are controlled by an adjustment factor, the difference in points between tourists with high immersion value and tourists with low immersion value is effectively amplified, and the fairness and discrimination degree of the distribution of the points are ensured; by setting a point base, the basic points are ensured to be obtained by all tourists, and the fairness problem caused by the zeroing of the points is avoided. A weighted fair queuing mechanism is adopted, the resource requests are sorted and processed according to the consumable points of the tourists, the activity and immersion experience of the tourists are taken into account, and the rationality and fairness of the priority of the processing of the resource requests are realized. A leaky bucket flow control mechanism is introduced, the leaky bucket capacity and the request flow rate are set, the tourist request flow is ensured to be controlled and not to exceed the system carrying capacity, the system load balancing is realized, and the response delay caused by resource overload is avoided.

[0131] Embodiment two

[0132] Please refer to Figure 3 The embodiment does not describe some parts in detail, and the description of embodiment one is provided. A destination global tourism big data management method based on data processing is provided, which includes the following steps:

[0133] S1, collect global tourism data and construct a geographic grid network, set a spatial grid reconstruction rule, and automatically refine or merge the grids; a trajectory triggering mechanism is introduced, when it is detected that the cumulative behavior frequency in the preset target area reaches the preset cumulative behavior frequency threshold, a fine-grained modeling is enabled to generate a geographic perception structure;

[0134] S2, analyze the continuous behavior of tourists in the geographic perception structure, and construct a multi-dimensional situation coding vector; through a behavior puzzle type deconstruction method and an implicit behavior compensation method, the complete behavior chain of the tourists is reconstructed;

[0135] S3, based on the multi-dimensional situation coding vector and the complete behavior chain, the immersion value points of the tourists are calculated, a resonance path matrix generation mechanism is used to compare the resonance similarity of different tourists on the complete activity path, a resonance path matrix between the tourists is generated, and a tourist behavior evolution map is established;

[0136] S4, according to the immersion value points of the tourists and the tourist behavior evolution map, a personalized tour guide path is generated for the tourists; a reverse resonance recommendation strategy is adopted, and the highest similarity but unvisited geographic grid is pushed to the tourists in combination with the immersion value points of the tourists;

[0137] S5, the platform is around the geographic grid with the highest resonance similarity, and the business resources are linked; a resonance group set meal linkage mechanism is introduced, the publication of the business resources in the preset destination is driven, and a resource sharing network is formed;

[0138] S6, map the immersion value of the tourists into consumable points, support the exchange behavior between the tourists and the merchants; adopt the weighted fair queuing and the leaky bucket flow limiting mechanism to distribute the merchant resources, and dynamically balance the tourist satisfaction and the resource utilization rate.

[0139] The above formulas are all dimensionless values calculated, the formulas are obtained by collecting a large amount of data to simulate a formula of the latest real situation, and preset parameters and threshold values in the formulas are set by a person skilled in the art according to actual conditions.

[0140] The above only describes the preferred embodiments of the present application, and the protection scope of the present application is not limited to the above-mentioned embodiments, and any technical scheme falling within the concept of the present application belongs to the protection scope of the present application. It should be noted that, for ordinary technical users in the technical field, some improvements and refinements without departing from the principles of the present application are also considered as the protection scope of the present application.

Claims

1. A destination global tourism big data management platform based on data processing, characterized in that, The method comprises the following steps: A perception aggregation point module collects global tourism data and constructs a geographic grid network, sets spatial grid reconstruction rules, and automatically refines or combines the grids; a trajectory triggering mechanism is introduced, and when the cumulative behavior frequency in the preset target area reaches the preset cumulative behavior frequency threshold, a fine-grained modeling is enabled to generate a geographic perception structure; A behavior deconstruction coding module analyzes the continuous behavior of tourists in the geographic perception structure and constructs a multi-dimensional situation coding vector; through a behavior puzzle deconstruction method and an implicit behavior compensation method, the complete behavior chain of the tourists is reconstructed; An immersion value quantification module calculates the immersion value score of the tourists based on the multi-dimensional situation coding vector and the complete behavior chain, generates a resonance path matrix of the tourists by comparing the resonance similarity of different tourists on the complete activity path, and establishes a tourist behavior evolution graph; The tourist behavior evolution graph is a weighted directed graph, each node of the tourist behavior evolution graph represents each tourist, and the edge represents the behavior relationship connection between two tourists; the weight of the edge represents the resonance similarity score of the tourists on the complete activity path; A guide recommendation feedback module generates a personalized tour guide path for the tourists according to the immersion value score of the tourists and the tourist behavior evolution graph; a reverse resonance recommendation strategy is adopted, and the highest similarity but unvisited geographic grid point is pushed to the tourists in combination with the immersion value score of the tourists; The method for pushing the highest similarity but unvisited geographic grid point comprises the following steps: Filtering out a group of adjacent tourists whose resonance similarity scores on the complete activity path are greater than a preset resonance similarity score threshold of the complete activity path; according to the complete behavior chain of the preset target tourist, marking the visited geographic grid points of the target tourist, and combining the complete behavior chains of the adjacent tourist group, identifying the geographic grid points that are not visited by the target tourist but visited by the adjacent tourists as recommended candidate grid points; Introducing the immersion value score of the tourists as a weight factor, and through a weighted vector aggregation method, the resonance similarity score of the tourists on the complete activity path and the immersion value score of the tourists are comprehensively evaluated to form a recommendation priority ranking model, and for all the recommended candidate grid points, the geographic grid point with the highest resonance similarity score and the highest immersion value score is preferentially recommended to the preset target tourist; A dynamic recommendation mechanism is designed, the recommended geographic grid point list is dynamically generated according to the changes of the tourist behavior data and the immersion value score of the tourists, the push notification technology is adopted, the latest recommended result is sent to the mobile terminal of the tourists in real time, and the personalized tour planning of the tourists is assisted; A platform response management module links the merchant resources around the geographic grid point with the highest resonance similarity; a resonance group set meal linkage mechanism is introduced to drive the publication of the merchant resources in the preset destination, and a resource sharing network is formed; An integral exchange reward module maps the immersion value score of the tourists into consumable points to support the exchange behavior between the tourists and the merchants; a weighted fair queuing and leaky bucket flow limiting mechanism is adopted to allocate the merchant resources, and the dynamic balance between the tourist satisfaction and the resource utilization rate is achieved; The method for dynamically balancing the tourist satisfaction and the resource utilization rate comprises the following steps: The tourist immersion value points are mapped into consumable points through a nonlinear weighted mapping function, and the nonlinear weighted mapping function is defined as ;in, For tourists Spendable points; For tourists Immersion value points; An adjustment factor to control the overall expansion or reduction of consumable points; The adjustment factor for controlling the shape of the consumable integral mapping curve; The base of consumable points is used to ensure that all tourists have a minimum consumable points value; Will tourists Spendable points The points are stored in the points redemption reward module and the visitor's available points balance is updated. When a visitor initiates a request to redeem merchant resources, the platform uses a weighted fair queuing mechanism to sort visitors in descending order based on their available points, and processes the visitor's request in descending order of available points and allocates available merchant resources. The platform calculates a resource share available for each tourist according to a total amount of current merchant resources and consumable points; a leaky bucket flow control mechanism is introduced to control resource request flow, and control parameters of the resource request flow are set, including a leaky bucket capacity and a request outflow rate, so that a tourist request rate is less than the request outflow rate, and the number of requests in the leaky bucket is not more than the leaky bucket capacity, thereby dynamically balancing tourist satisfaction and resource utilization. 2.The destination global tourism big data management platform based on data processing according to claim 1, characterized in that, The method for automatically refining or merging the grid points comprises: Collect global tourism data, including tourist behavior data, environmental facility data, merchant service data, spatial geographic data, and management policy data; clean the global tourism data through a sliding window outlier detection technique, and perform standard deviation normalization processing to obtain normalized global tourism data; For the normalized global tourism data of geographical coordinate points, a convex hull algorithm is used to extract the minimum convex boundary of all geographical coordinate points, and the required coverage area for constructing a geographical grid network is determined according to the minimum convex boundary; the required coverage area is preliminarily divided into spatial grids according to a preset basic grid side length An adaptive quadtree division method is used to generate an initial geographical grid network, and for any point in the required coverage area , a mapping function is used to realize the encoding mapping of geographical coordinate points to geographical grids. A dynamic adjustment strategy based on spatial density is introduced, the basic grid side length is dynamically adjusted, and the point density of any point in the required coverage area is calculated According to the point density The basic grid side length is adjusted by a grid side length adjustment function.​ Encode all geographic coordinate points in the global tourism data, classify them into corresponding geographic grid points, and set a spatial grid point reconstruction rule; for any grid point in the geographic grid point network, if there is a time period that satisfies the refinement trigger condition function within a preset time window, the grid point is triggered to be automatically refined; For any grid point in the geographic grid point network, if there is an adjacent grid point belonging to the adjacent grid point set of the current arbitrary grid point, and the cumulative behavior frequency in the adjacent grid point is less than the cumulative behavior frequency in the grid point, the grid point is triggered to be automatically merged. 3.The destination global tourism big data management platform based on data processing according to claim 2, characterized in that, The method for generating the geographic perception structure comprises: Introduce a trajectory trigger mechanism to collect and count trajectory points in the tourist behavior data contained in the global tourism data, and calculate the cumulative behavior frequency of the trajectory points in a preset target area within a preset time period through a sliding time window; Determine whether the cumulative behavior frequency reaches a preset cumulative behavior frequency threshold, and if so, trigger fine-grained modeling; adjust the spatial division granularity in the target area according to the trajectory point distribution density, including reducing the base grid length and deepening the quadtree division depth, to generate a finer-grained geographic grid point network; Remap and encode all trajectory points in the preset target area, classify the trajectory points into corresponding newly generated fine-grained grid points; and generate a geographic perception structure containing grid point encoding, spatial location information, trajectory point cumulative behavior frequency in a time period, and trajectory point density information based on the remapped trajectory points and their corresponding fine-grained grid points. 4.The destination global tourism big data management platform based on data processing according to claim 3, characterized in that, The method for constructing the multi-dimensional context encoding vector comprises: Time-sequentially sort the trajectory points in the tourist behavior data in the global tourism data, extract the timestamp, spatial location information, grid point encoding corresponding to the trajectory point, and behavior category of each trajectory point, and form a continuous trajectory behavior sequence; For each trajectory behavior sequence, analyze the crossing path, time period, stay duration, moving speed, and direction change in the geographic perception structure, and count the behavior pattern features in each time period; extract multi-dimensional trajectory features according to the spatial dimension, time dimension, behavior dimension, social dimension, and semantic dimension; and use feature splicing and hash encoding processing to combine the extracted multi-dimensional trajectory features to generate a multi-dimensional context encoding vector. 5.The destination global tourism big data management platform based on data processing according to claim 4, characterized in that, The method for reconstructing the complete behavior chain of the tourist comprises: The parsed tourist multi-dimensional situation coding vector sequence is divided according to the time sequence and the space adjacency relationship, segmented into n behavior sub-segments, and the behavior sub-segments reflect the local behavior characteristics of the tourist within a limited time period and space range, which are called behavior puzzles; For the incomplete behavior sub-segments caused by data missing, sensing blind area or abnormal detection, an implicit behavior compensation method is adopted, based on the tourist behavior data and the environmental facility data, and the missing behavior sub-segments and their time sequence and space position are inferred by using a machine learning model; The behavior sub-segments and the behavior sub-segments obtained through the implicit behavior compensation are spliced based on the time sequence and the space adjacency relationship, and the multi-dimensional trajectory features corresponding to the multi-dimensional situation coding vectors are combined to construct a complete behavior chain of the tourist. 6.The destination global tourism big data management platform based on data processing according to claim 5, characterized in that, The method for establishing the tourist behavior evolution graph comprises: According to the multi-dimensional situation coding vectors and the complete behavior chain of the tourist obtained by parsing the tourist in the geographical grid network, a weighted vector aggregation technology is used to calculate the immersion value score of the tourist, and the immersion value score of the tourist comprehensively reflects the behavior activity, behavior type, stay time and emotional participation of the tourist in the time and space dimensions; The behavior value is calculated by combining the behavior category weight and the cumulative behavior frequency statistics, and the immersion value score of the tourist is generated by collecting the tourist comment feedback data, introducing a sentiment analysis model to score the tourist comment feedback data, and generating the immersion value score of the tourist; A resonance path matrix is constructed, a dynamic time warping algorithm is used to match the multi-dimensional situation coding vector sequences of different tourists, the path similarity is calculated, the tourist behavior data is time-aligned, the Euclidean distance between the multi-dimensional feature vectors corresponding to the multi-dimensional situation coding vectors is calculated as a local distance, a complete activity path is constructed based on the local distance, and a resonance similarity score of the tourist on the complete activity path is outputted; A tourist behavior evolution graph is constructed according to the constructed resonance path matrix. 7.The destination global tourism big data management platform based on data processing according to claim 6, characterized in that, The method for generating the individualized tour guide path comprises: The tourist behavior evolution graph represented by a weighted directed graph is used to analyze the behavior similarity of any one preset target tourist with other tourists in the tourist behavior evolution graph, identify the nearest neighbor tourist group with the most similar behavior mode, and infer the tour route that the target tourist may be interested in based on the historical behavior evolution path of the nearest neighbor tourist group; The A* algorithm is used to obtain an optimal tour path that meets the interest preferences and path continuity of the tourist, which is defined as an individualized tour guide path; and the generated individualized tour guide path is fed back to the mobile terminal of the tourist in the form of multi-dimensional situation coding, and real-time navigation and behavior suggestions are provided. 8.The destination global tourism big data management platform based on data processing according to claim 7, characterized in that, The method for forming the resource sharing network comprises: The geographical grid area with the highest resonance similarity score of the tourist on the complete activity path is determined as a core resonance node; and the core resonance node and the merchant resources in the preset destination are spatially mapped based on a spatial database technology to establish a correspondence between the geographical position and the merchant resources; Through the event-driven architecture, combined with the LSTM time series prediction model, the activation state of the resonance group package is monitored in real time, and resource scheduling instructions are automatically pushed to related merchants. The micro-service architecture is used for elastic expansion. After receiving the instructions, the resource management system on the merchant side completes inventory adjustment, personnel scheduling and service preparation, and realizes dynamic resource response; A resource sharing network covering the preset target geographic grid and surrounding area merchants is constructed using a graph database. The nodes in the resource sharing network represent merchant resources, and the edges represent resource sharing and cooperation relationships. Tourist consumption data, merchant service feedback data and user evaluation data are collected in real time using a big data processing platform. The service quality is evaluated using a BERT sentiment classifier, and the resonance group package design and resource scheduling strategy are continuously optimized using multi-dimensional data fusion and causal inference methods.

Images