Service work order identification and distribution method and system based on tourist time-space behavior map

Abstract

The invention discloses a service work order identification and distribution method and system based on a tourist time-space behavior map, and solves the problem that the real intention of a user is difficult to accurately understand due to the fuzziness and diversity of languages and the existence of specific terms of a scenic spot because an existing scenic spot service work order contains various modal information such as voice, characters, pictures and the like. The method comprises the following steps: pre-constructing a tourist time-space behavior map in combination with historical associated data, performing semantic understanding on tourist demand data by the time-space behavior map, and distributing a personalized service work order to a corresponding optimal server based on an interaction degree; according to the method, the tourist time-space behavior map containing the entity extraction model is pre-constructed in combination with the historical associated data, so that the time-space behavior characteristics of tourists and servers can be captured more comprehensively, and the time-space behavior map can perform accurate semantic understanding on tourist demand data; the personalized service work order is accurately distributed to the optimal server with the highest association degree, so that the service efficiency and the tourist satisfaction degree are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and particularly relates to a method and system for identifying and distributing service work orders based on a tourist spatio-temporal behavior graph. Background Art

[0002] With the improvement of people's material living standards, more and more people choose to travel during holidays, and the tourism industry is booming. Predicting tourism demand and providing services based on demand is a key issue in the tourism industry. By accurately predicting trends and patterns, better infrastructure and resource scheduling can be provided to offer high-quality services to tourists. Knowledge graph technology has been widely applied in multiple fields. By constructing knowledge associations, it can achieve efficient management and in-depth mining of complex information. A tourist spatio-temporal behavior graph is a knowledge graph constructed based on the time and space behavior data of tourists during the travel process. It integrates information such as the location, time, stop points, and activity sequence of tourists, and displays the behavior patterns and preferences of tourists in the form of a graph structure.

[0003] A scenic area service work order is a work record form used in scenic area management to record, process, and track tourist demands, problems, or requests. It plays an important role in the daily operation of the scenic area and is an important tool for the scenic area to provide efficient services and solve problems.

[0004] At present, the traditional processing of scenic area service work orders relies on manual assignment, which is inefficient and prone to errors. The combination of tourist spatio-temporal behavior graphs and knowledge graph technology can achieve automatic classification and intelligent distribution of work orders. This intelligent processing method not only reduces manual intervention, but scenic area service work orders may contain various modal information such as voice, text, and pictures, and it may be difficult to accurately understand the true intentions of users due to the ambiguity, diversity of languages, and the existence of specific scenic area terms. How to effectively fuse these multi-modal information and accurately understand its semantics is a challenge. To address the above problems, we propose a method and system for identifying and distributing service work orders based on tourist spatio-temporal behavior graphs. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and system for identifying and distributing service work orders based on tourist spatio-temporal behavior graphs in view of the deficiencies of the prior art, and solve the problem that existing scenic area service work orders may contain various modal information such as voice, text, and pictures, and it is difficult to accurately understand the true intentions of users due to the ambiguity, diversity of languages, and the existence of specific scenic area terms.

[0006] The present invention is implemented as follows. For the method for identifying and distributing service work orders based on tourist spatio-temporal behavior graphs, the method includes:

[0007] Crawl historical correlation data associated with tourists' spatio-temporal behavior, and pre-construct a spatio-temporal behavior map of tourists in combination with the historical correlation data;

[0008] Collect real-time multi-source heterogeneous tourist demand data. The spatio-temporal behavior map performs semantic understanding on the tourist demand data based on advanced statistical functions, triggers a service work order generation instruction, and generates a personalized service work order in response to the generation instruction;

[0009] Load the personalized service work order, extract the work order priority, user tags, and work order information features of the work order based on a pre-trained entity extraction model and perform weighted fusion to obtain a fusion feature set. The entity extraction model traverses the spatio-temporal behavior map to obtain a user-service provider interaction matrix, a user-associated person interaction matrix, and a service-associated person interaction matrix based on the fusion feature set, and performs weighted combination on the interaction matrices to obtain an in-degree interactivity matrix;

[0010] Input the in-degree interactivity matrix into the entity extraction model, analyze the interaction degree between the personalized service work order and the service provider based on the ant colony optimization algorithm and fuzzy clustering, and distribute the personalized service work order to the corresponding optimal service provider based on the interaction degree.

[0011] Preferably, the method for pre-constructing a spatio-temporal behavior map in combination with historical correlation data includes:

[0012] Define the pattern layer of the map based on four dimensions: user spatio-temporal behavior, service provider spatio-temporal behavior, personnel interaction behavior, and personnel prediction behavior. Use the four dimensions as the ontology of the pattern layer to guide the construction of the map data layer;

[0013] Use the Scrapy crawler framework to crawl multi-source heterogeneous historical correlation data, extract historical correlation data entities, entity attributes, and entity relationships, and preprocess the historical correlation data. Among them, the historical correlation data includes user location data, time data, behavior data, external data, user name, gender, age, education level, home address, historical demand data, and destination information;

[0014] Pre-construct an entity extraction model, use historical correlation data to iteratively train the entity extraction model, and output a converged entity extraction model;

[0015] Load the preprocessed historical correlation data, extract the entity feature attributes of the historical correlation data based on the entity extraction model, and call the CEA integration algorithm to assign weights to the entity feature attributes from the perspectives of user spatio-temporal behavior, service provider spatio-temporal behavior, and personnel interaction behavior;

[0016] Associate the entity feature values of the historical correlation data with the attribute weight values to obtain a user spatio-temporal behavior weight matrix, a service provider spatio-temporal behavior weight matrix, and a personnel interaction behavior weight matrix of the entity feature attributes;

[0017] Calculate the predicted behavior influence weights of entity feature attributes in the user spatio-temporal behavior weight matrix, service provider spatio-temporal behavior weight matrix, and personnel interaction behavior weight matrix based on the Tanimoto coefficient;

[0018] Preset a predicted weight threshold, and extract the entity feature values and attribute weight values of the spatio-temporal behavior weight matrix, service provider spatio-temporal behavior weight matrix, and personnel interaction behavior weight matrix based on the predicted weight threshold to obtain a personnel prediction behavior matrix based on user spatio-temporal behavior, service provider spatio-temporal behavior, and personnel interaction behavior;

[0019] Load the user spatio-temporal behavior weight matrix, service provider spatio-temporal behavior weight matrix, personnel interaction behavior weight matrix, and personnel prediction behavior matrix, use the LOAD CSV command to convert the data in the matrix into the triple form of the knowledge graph data layer, store the data in the form of a graph and visualize it using Neo4j to complete the construction of the spatio-temporal behavior graph.

[0020] Preferably, the entity extraction model is based on a BP neural network model. After the input layer of the basic model, a Transformer architecture is introduced. The Transformer architecture consists of an encoder and a decoder. The FreeLB adversarial training algorithm is introduced into the encoder. The hidden layer of the basic model is replaced with a spatio-temporal federated learning framework. Advanced statistical functions are introduced into the spatio-temporal federated learning framework. A CRF annotation sequence layer embedded with a Softmax layer is added between the spatio-temporal federated learning framework and the output layer. The output of the Transformer architecture is activated using the sigmoid activation function. The entity extraction model is iteratively trained based on the FreeLB adversarial training algorithm combined with historical association data. The hyperparameters of the entity extraction model are adjusted through the Adma optimizer to minimize the cross-entropy loss function of the entity extraction model.

[0021] Preferably, the predicted behavior influence weight is calculated by the following formula:

[0022]

[0023] where q d represents the predicted behavior influence weight of the entity feature attribute, and W s a , W s b , W s c are respectively the entity feature values of the entity feature attribute s in the user spatio-temporal behavior weight matrix A, service provider spatio-temporal behavior weight matrix B, and personnel interaction behavior weight matrix C. q a , q b , q c are respectively the attribute weight values of the entity feature attribute s. Sim TDenote the Tanimoto coefficients of the user spatio-temporal behavior weight matrix, the service provider spatio-temporal behavior weight matrix, and the person interaction behavior weight matrix, φ a,b,c Denote the embedding vectors of the entity feature attribute s in the user spatio-temporal behavior weight matrix A, the service provider spatio-temporal behavior weight matrix B, and the person interaction behavior weight matrix C. T represents the number of entity features, |KU i +KU j | represents the interaction degree value between the user and the service provider, |A i +B i +C i |, |A j +B j +C j | respectively represent the number of entity features of the interacting user i and service provider j in the user spatio-temporal behavior weight matrix A, the service provider spatio-temporal behavior weight matrix B, and the person interaction behavior weight matrix C. |A∩B∩C| represents the matrix intersection, and |A∪B∪C| represents the matrix union.

[0024] Preferably, the spatio-temporal behavior map performs semantic understanding on the tourist demand data based on advanced statistical functions, including:

[0025] Load the tourist demand data, preprocess the tourist demand data, and identify the preprocessed tourist demand data to determine the data type;

[0026] If the data type is non-text data, the input layer of the entity extraction model extracts the signal feature of the non-text data, and performs denoising processing on the signal feature based on the inverse fast Fourier transform;

[0027] If the data type is text data, the encoder of the Transformer architecture of the entity extraction model performs feature serialization encoding on the text data to obtain a text encoding set;

[0028] Load the signal reconstruction feature and the text encoding set. The spatio-temporal federated learning framework combines advanced statistical functions to perform fusion statistics on the signal reconstruction feature and the text encoding set to obtain a semantic output set. The CRF annotation sequence layer performs user label annotation and priority determination on the semantic output set, and outputs personalized semantic information;

[0029] Using the personalized semantic information as an index, traverse the spatio-temporal behavior map based on the retrieval enhancement generation technology. The spatio-temporal behavior map outputs the user spatio-temporal behavior weight matrix, the service provider spatio-temporal behavior weight matrix, and the person interaction behavior weight matrix corresponding to the personalized semantic information in response to the personalized semantic information;

[0030] Load the user spatio-temporal behavior weight matrix, the service provider spatio-temporal behavior weight matrix, and the person interaction behavior weight matrix corresponding to the personalized semantic information. The spatio-temporal behavior map generates a person prediction behavior matrix corresponding to the personalized semantic information;

[0031] Extract the triple entities and entity relationships corresponding to the personnel prediction behavior matrix in the knowledge graph data layer, load the triple entities and entity relationships in the knowledge graph data layer into a preset work order template, and generate personalized service work orders.

[0032] Preferably, when performing denoising processing on signal features:

[0033] Load the original signal and identify the original amplitude and phase corresponding to the original signal frequency, and construct a screening window with adjustable length and positive and negative directions;

[0034] The screening window identifies the normal amplitude, discrete amplitude, and high-energy amplitude in the original amplitude based on the clustering spectrum algorithm, and suppresses the discrete amplitude;

[0035] Use the clustering spectrum algorithm to calculate the Euclidean distance between the high-energy amplitude and the normal amplitude, and judge whether the Euclidean distance between the high-energy amplitude and the clustering center of the normal amplitude exceeds the normal amplitude threshold interval;

[0036] If the Euclidean distance between the high-energy amplitude and the clustering center of the normal amplitude exceeds the normal amplitude threshold interval, suppress the high-energy amplitude. If the Euclidean distance between the high-energy amplitude and the clustering center of the normal amplitude does not exceed the normal amplitude threshold interval, retain the current high-energy amplitude;

[0037] Load the original signal frequency after denoising processing of the signal features, and reconstruct the original signal based on the inverse fast Fourier transform to obtain the signal reconstruction features of the non-text data.

[0038] Preferably, the method for the in-degree interaction matrix corresponding to the personalized service work order and the service provider interaction degree based on the ant colony optimization algorithm and fuzzy clustering analysis includes:

[0039] Load the in-degree interaction matrix and the associated personalized service work order, identify the prediction spatio-temporal point information in the personnel prediction behavior matrix based on the personalized service work order, and use the ant colony optimization algorithm to find l paths from the current position point l1 to the prediction spatio-temporal point l n The ant colony optimization algorithm iteratively searches for the l paths to find the optimal path l best ;

[0040] Based on the optimal path l best Extract the service providers within the spatio-temporal behavior, calculate the initial interaction degree between the personalized service work order and the service provider using the Pearson correlation coefficient, and generate an initial interaction degree sequence;

[0041] The fuzzy clustering analysis algorithm characterizes the weighted interaction degree between the personalized service work order and the service provider through the Minkowski distance to obtain a weighted interaction degree sequence;

[0042] Load the initial interaction degree sequence between the personalized service work order and the service provider, and the weighted interaction degree sequence between the personalized service work order and the service provider. Compress the time series objects of the personalized service work order and the service provider locally on the time axis and space axis, optimally map the weighted interaction degree sequence to the initial interaction degree sequence to obtain the interaction degree between the personalized service work order and the service provider. Sort the service providers based on the interaction degree between the personalized service work order and the service provider to obtain the ranking position of each service provider. The ant colony optimization algorithm performs initial pheromone weighted update according to the ranking of the service providers to obtain the optimal service provider.

[0043] The interaction degree between the personalized service work order and the service provider is calculated by the following formula:

[0044]

[0045] Among them, J best , J0, respectively represent the interaction degree, the initial interaction degree, and the weighted interaction degree. δ is the spatio-temporal expected degree of the service provider in the optimal path l best Among them, G and G represent the in-degree interaction matrix and the mean of the in-degree interaction matrix respectively. k is the number of service providers within the spatio-temporal behavior. AG i , AG j respectively represent the weight matrix of the in-degree interaction matrix corresponding to user i and service provider j, respectively represent the in-degree interaction values of user i and service provider j in the in-degree interaction matrix.

[0046] On the other hand, the present invention also provides a service work order identification and distribution system based on the spatio-temporal behavior map of tourists. The service work order identification and distribution system based on the spatio-temporal behavior map of tourists includes:

[0047] A knowledge graph module for crawling historical associated data related to the spatio-temporal behavior of tourists, and pre-constructing a spatio-temporal behavior map of tourists in combination with the historical associated data;

[0048] A demand identification module for real-time collecting multi-source heterogeneous tourist demand data. The spatio-temporal behavior map performs semantic understanding on the tourist demand data based on advanced statistical functions, triggers a service work order generation instruction, and generates a personalized service work order in response to the generation instruction;

[0049] A work order index module for loading personalized service work orders, extracting work order priorities, user tags, and work order information features based on a pre-trained entity extraction model and performing weighted fusion to obtain a fusion feature set. The entity extraction model traverses the spatio-temporal behavior map to obtain a user-service provider interaction matrix, a user-associated person interaction matrix, and a service-associated person interaction matrix based on the fusion feature set, and performs weighted combination on the interaction matrices to obtain an in-degree interaction matrix;

[0050] The work order distribution module is used to input the in-degree interaction matrix into the entity extraction model, analyze the interaction degree between the personalized service work order and the service provider corresponding to the in-degree interaction matrix based on the ant colony optimization algorithm and fuzzy clustering analysis, and distribute the personalized service work order to the corresponding optimal service provider based on the interaction degree.

[0051] Compared with the prior art, the embodiments of the present application mainly have the following beneficial effects:

[0052] In the embodiments of the present invention, a tourist spatio-temporal behavior map including an entity extraction model is pre-constructed in combination with historical association data. The entity extraction model introduces an entity extraction model based on the BP neural network and the Transformer architecture, and combines the FreeLB adversarial training algorithm, which significantly improves the model's processing ability and robustness for complex data. Through the spatio-temporal federated learning framework and advanced statistical functions, the model's ability to model spatio-temporal data is further enhanced, and it can more comprehensively capture the spatio-temporal behavior characteristics of tourists and service providers, ensuring the accuracy and efficiency of feature extraction. The spatio-temporal behavior map can accurately understand the semantic meaning of tourist demand data, generate personalized semantic information and a personnel prediction behavior matrix based on the tourist demand data, dynamically evaluate the matching degree between the service provider and the work order, ensure that the personalized service work order is accurately distributed to the optimal service provider with the highest correlation, thereby improving service efficiency and tourist satisfaction, and overcoming the problem that existing scenic area service work orders may contain various modal information such as voice, text, and pictures, and it is difficult to accurately understand the true intention of users due to the ambiguity and diversity of language and the existence of specific scenic area terms.

[0053] In the embodiments of the present invention, the entity extraction model significantly improves the accuracy, robustness, and generalization ability of entity extraction by combining the non-linear modeling ability of the BP neural network and the long-distance dependence capture ability of the Transformer architecture. At the same time, the Transformer architecture can effectively capture long-distance dependence relationships through the self-attention mechanism, especially suitable for processing long sequence data. Its parallel computing ability significantly improves the training and inference efficiency. Introducing the FreeLB adversarial training algorithm can enhance the robustness of the model, enabling it to more accurately identify semantic entities when facing modal information with strong ambiguity and diversity. Adversarial training helps the model learn more stable feature representations by simulating adversarial samples, thereby improving the generalization ability. Replacing the hidden layer of the BP neural network with the spatio-temporal federated learning framework can better handle the complexity of spatio-temporal data. This framework further improves the model's ability to model spatio-temporal features by introducing advanced statistical functions. Adding a CRF annotation sequence layer embedded with a Softmax layer between the spatio-temporal federated learning framework and the output layer can effectively handle the annotation dependence problem in sequence annotation tasks. The Softmax layer is used to activate the output of the Transformer architecture to further optimize the output results of the model.

[0054] An embodiment of the present invention provides a method for semantic understanding of tourist demand data by a spatio-temporal behavior map based on advanced statistical functions. By preprocessing tourist demand data in multiple ways, combining the inverse fast Fourier transform to denoise signal features, fusing statistics through a spatio-temporal federated learning framework, and generating a personnel prediction behavior matrix corresponding to personalized semantic information by the spatio-temporal behavior map, accurate semantic understanding of tourist demand data is achieved. This method can not only process multi-modal data, but also dynamically adjust service content, generate personalized service work orders, and significantly improve the quality and efficiency of scenic area services.

[0055] In the embodiment of the present invention, the processed signal features are reconstructed by the inverse fast Fourier transform (iFFT), and the time-domain characteristics of the signal can be completely restored. This method not only removes noise, but also retains the original structure and key features of the signal, ensuring the usability of the signal after reconstruction. The clustering spectrum algorithm and the dynamic threshold adjustment mechanism make the entire signal reconstruction process highly flexible and scalable. It can be adjusted according to different application scenarios and signal characteristics to adapt to a variety of complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic diagram of the implementation process of the service work order identification and distribution method based on the tourist spatio-temporal behavior map provided by the present invention.

[0057] Figure 2 It shows a schematic diagram of the implementation process of the method for pre-constructing a spatio-temporal behavior map by combining historical association data.

[0058] Figure 3 It shows a schematic diagram of the implementation process of the method for semantic understanding of tourist demand data by the spatio-temporal behavior map based on advanced statistical functions.

[0059] Figure 4 It shows a schematic diagram of the implementation process of the method for the interaction degree between personalized service work orders and service providers corresponding to the in-degree interaction matrix based on the ant colony optimization algorithm and fuzzy clustering analysis.

[0060] Figure 5 It shows a schematic diagram of the structure of the service work order identification and distribution system based on the tourist spatio-temporal behavior map. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0061] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which this application belongs; the terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above drawings are intended to cover non-exclusive inclusion. The terms "first", "second", etc. in the specification and claims of this application or the above drawings are used to distinguish different objects and are not used to describe a specific order.

[0062] Existing scenic area service work orders may contain various modal information such as voice, text, and pictures. Due to the ambiguity and diversity of languages and the existence of specific scenic area terms, it is difficult to accurately understand the true intentions of users. To address the above problems, we propose a service work order recognition and distribution method and system based on the tourist spatio-temporal behavior map. Briefly, when implementing the method, first, a tourist spatio-temporal behavior map is pre-constructed in combination with historical association data. The spatio-temporal behavior map performs semantic understanding on tourist demand data based on advanced statistical functions, generates personalized service work orders, extracts work order priorities, user tags, and work order information features based on a pre-trained entity extraction model and performs weighted fusion to obtain a fusion feature set. Based on the fusion feature set, a user-service provider interaction matrix, a user-associated person interaction matrix, and a service-associated person interaction matrix are obtained. The interaction matrices are weighted and combined to obtain an in-degree interactivity matrix. Based on the ant colony optimization algorithm and fuzzy clustering analysis, the interactivity between the personalized service work order and the service provider corresponding to the in-degree interactivity matrix is analyzed, and the personalized service work order is distributed to the corresponding optimal service provider based on the interactivity. In the embodiments of the present invention, a tourist spatio-temporal behavior map including an entity extraction model is pre-constructed in combination with historical association data. The entity extraction model introduces an entity extraction model based on a BP neural network and a Transformer architecture, and combines the FreeLB adversarial training algorithm, significantly improving the model's processing ability and robustness for complex data. Through the spatio-temporal federated learning framework and advanced statistical functions, the model's ability to model spatio-temporal data is further enhanced, enabling it to more comprehensively capture the spatio-temporal behavior characteristics of tourists and service providers, ensuring the accuracy and efficiency of feature extraction, enabling the spatio-temporal behavior map to perform precise semantic understanding on tourist demand data, generating personalized semantic information and a personnel prediction behavior matrix based on tourist demand data, dynamically evaluating the matching degree between the service provider and the work order, ensuring that the personalized service work order is accurately distributed to the optimal service provider with the highest degree of association, thereby improving service efficiency and tourist satisfaction, and overcoming the problem that existing scenic area service work orders may contain various modal information such as voice, text, and pictures, and it is difficult to accurately understand the true intentions of users due to the ambiguity and diversity of languages and the existence of specific scenic area terms.

[0063] The embodiments of the present invention provide a service work order recognition and distribution method based on a tourist spatio-temporal behavior map.Figure 1 The figure shows a schematic implementation flow diagram of a service work order identification and distribution method based on a tourist spatio-temporal behavior map. The service work order identification and distribution method based on a tourist spatio-temporal behavior map specifically includes:

[0064] Step S10: Crawl historical associated data related to tourist spatio-temporal behavior, and pre-construct a tourist spatio-temporal behavior map in combination with the historical associated data;

[0065] In the embodiment of the present invention, by crawling multi-source heterogeneous historical associated data, rich information related to tourist spatio-temporal behavior can be comprehensively collected, providing a solid data basis for map construction. Compared with the traditional method that relies on limited data sources or manual data collation, this method can more comprehensively reflect tourist behavior patterns and preferences.

[0066] Step S20: Real-time collect multi-source heterogeneous tourist demand data. The spatio-temporal behavior map performs semantic understanding on the tourist demand data based on advanced statistical functions, triggers a service work order generation instruction, and in response to the generation instruction, generates a personalized service work order;

[0067] It should be noted that tourist demand data includes, but is not limited to, tourism decision-making behavior data, tourist activity trajectories within scenic spots, tourist tour plans, tour times, willingness to visit scenic spots, whether to participate in scenic spot activities, whether to communicate with staff, tourist consumption behavior data during the tour process, online reservation data, tourist names, contact information, admission times, locations, specific descriptions of problems or needs encountered by tourists, work orders submitted when tourists need additional services, work orders submitted when tourists are dissatisfied with the services, facilities or management of the scenic spot, and work orders submitted when tourists encounter emergencies (such as injuries, getting lost, bad weather, etc.).

[0068] By real-time collecting multi-source heterogeneous tourist demand data, including various modal information such as voice, text, and pictures, and performing fusion processing on them. The spatio-temporal behavior map can make full use of the advantages of multi-modal data. Compared with the single-modal data processing method, it can more comprehensively capture tourist demand information, improve the accuracy of work order generation. Using the spatio-temporal behavior map to perform semantic understanding on tourist demand data in combination with advanced statistical functions can more accurately understand the intentions of tourists, and can accurately identify even in the face of the ambiguity and diversity of language. This solves the problem of incorrect or incomplete work order generation caused by inaccurate semantic understanding in traditional methods, and improves the quality and pertinence of work orders.

[0069] Step S30: Load the personalized service work order, extract the work order priority, user tags, and work order information features based on the pre-trained entity extraction model, and perform weighted fusion to obtain a fused feature set. Compared with simple feature extraction methods, this deep feature extraction and fusion method can more comprehensively characterize the features of the work order, providing a more accurate basis for subsequent work order distribution. The entity extraction model traverses the spatio-temporal behavior graph to obtain the user-service provider interaction matrix, user-associated person interaction matrix, and service-associated person interaction matrix based on the fused feature set, and combines the interaction matrices with weights to obtain the in-degree interactivity matrix. The entity extraction model traverses the spatio-temporal behavior graph to obtain various interaction matrices, which reflect the complex relationships between users and service providers, users and associated persons, and services and associated persons. By combining the interaction matrices with weights to obtain the in-degree interactivity matrix, these relationships are further quantified, providing an important basis for subsequent service work order distribution. For example, in a travel service platform, the interaction matrix between a user and a hotel service staff can reflect the user's satisfaction with hotel services and the frequency of demand; the interaction matrix between a user and a travel agency staff can reflect the user's preferences and feedback on travel products. The in-degree interactivity matrix synthesizes this information to help the system more accurately distribute service work orders to the most suitable service providers.

[0070] It should be noted that the user-service provider interaction matrix is an interaction matrix representing the association information between the current user and the service provider. The rows of this matrix represent users, the columns represent service providers, and each element in the matrix represents the interaction intensity or association degree between the user and the service provider. For example, the values in the matrix can be the user's rating of the service provider, the number of service times, or other quantitative indicators. The user-associated person interaction matrix represents the matrix of the association information between the current user and other users in the same spatio-temporal context, and the service-associated person interaction matrix represents the matrix of the association information between other users and service providers in the same spatio-temporal context. Through weighted combination, the association information between users and service providers, users and other users, and service providers and other users can be comprehensively considered, thus more comprehensively reflecting the interaction relationships between different entities.

[0071] Step S40: Input the in-degree interactivity matrix into the entity extraction model, analyze the interaction degree between the personalized service work order and the service provider based on the ant colony optimization algorithm and fuzzy clustering, and distribute the personalized service work order to the corresponding optimal service provider based on the interaction degree.

[0072] In this embodiment, analyzing the in-degree interactivity matrix based on the ant colony optimization algorithm and fuzzy clustering can intelligently evaluate the interaction degree between the personalized service work order and the service provider, and distribute the work order to the optimal service provider accordingly. Compared with traditional manual distribution or simple rule engine distribution methods, this intelligent optimization method can allocate resources more efficiently, improving service efficiency and quality.

[0073] In the embodiments of the present invention, a spatio-temporal behavior map of tourists including an entity extraction model is pre-constructed in combination with historical association data. The entity extraction model introduces an entity extraction model based on a BP neural network and a Transformer architecture, and combines with the FreeLB adversarial training algorithm, significantly improving the model's processing ability and robustness for complex data. Through the spatio-temporal federated learning framework and advanced statistical functions, the model's ability to model spatio-temporal data is further enhanced, enabling it to more comprehensively capture the spatio-temporal behavior characteristics of tourists and service providers, ensuring the accuracy and efficiency of feature extraction, enabling the spatio-temporal behavior map to accurately semantically understand tourist demand data, and generating personalized semantic information and a personnel prediction behavior matrix based on tourist demand data, capable of dynamically evaluating the matching degree between service providers and work orders, ensuring that personalized service work orders are accurately distributed to the optimal service provider with the highest correlation, thereby improving service efficiency and tourist satisfaction, and overcoming the problem that existing scenic area service work orders may contain various modal information such as voice, text, and pictures, and it is difficult to accurately understand the true intentions of users due to the ambiguity and diversity of languages and the existence of specific scenic area terms.

[0074] The embodiments of the present invention provide a method for pre-constructing a spatio-temporal behavior map in combination with historical association data. Briefly, the construction of the spatio-temporal behavior map is carried out through multi-dimensional data collection and preprocessing, entity extraction and feature weighting, weight matrix construction and prediction behavior analysis, combined with graph database storage and visualization, to construct a spatio-temporal behavior map of tourists. Figure 2 The figure shows a schematic implementation flow diagram of the method for pre-constructing a spatio-temporal behavior map in combination with historical association data. The method for pre-constructing a spatio-temporal behavior map in combination with historical association data specifically includes:

[0075] Step S101, define the schema layer of the map based on four dimensions: user spatio-temporal behavior, service provider spatio-temporal behavior, personnel interaction behavior, and personnel prediction behavior, and use the four dimensions as the schema layer ontology to guide the construction of the map data layer;

[0076] In the embodiments of the present invention, the schema layer ontology of the map is defined from multiple key dimensions, covering the behaviors of users and service providers as well as the interactions and prediction behaviors between personnel, providing a comprehensive framework for the construction of the entire spatio-temporal behavior map. This multi-dimensional consideration can more systematically reflect the mutual relationships between various factors in the real scenario, making the map more comprehensive and practical.

[0077] Step S102, use the Scrapy crawler framework to crawl multi-source heterogeneous historical association data, extract historical association data entities, entity attributes, and entity relationships, and preprocess the historical association data. Among them, the historical association data includes user location data, time data, behavior data, external data, user name, gender, age, education level, home address, historical demand data, and destination information;

[0078] It should be noted that the Scrapy crawler framework is used to collect historical associated data of multiple sources and heterogeneous types, covering various aspects of information such as user location, time, behavior, and external data, providing a rich data foundation for graph construction. The preprocessing steps ensure data quality, reduce noise and redundancy, and improve the efficiency of subsequent processing.

[0079] Step S103: Pre-build an entity extraction model, iteratively train the entity extraction model using historical associated data, and output a converged entity extraction model;

[0080] In the embodiment of the present invention, the entity extraction model can not only perform entity extraction on historical associated data, but also perform entity recognition on tourist demand data. It can solve the problem that the existing graph is difficult to accurately understand semantic entities due to the existence of various modal information with ambiguity and specific terms of scenic spots, and the existing graph recognition has problems of fuzzy entity boundaries and weak generalization ability. The entity extraction model is based on the BP neural network model. After the input layer of the basic model, a Transformer architecture is introduced. The Transformer architecture consists of an encoder and a decoder. The FreeLB adversarial training algorithm is introduced into the encoder. The hidden layer of the basic model is replaced by a spatio-temporal federated learning framework. Advanced statistical functions are introduced into the spatio-temporal federated learning framework. A CRF annotation sequence layer embedded with a Softmax layer is added between the spatio-temporal federated learning framework and the output layer. The sigmoid activation function is used to activate the output of the Transformer architecture. The entity extraction model is iteratively trained based on the FreeLB adversarial training algorithm combined with historical associated data. The hyperparameters of the entity extraction model are adjusted by the Adma optimizer to minimize the cross-entropy loss function of the entity extraction model. The performance of the model is regularly evaluated during the training process, and the validation set is used to check the accuracy and generalization ability of the model. The training strategy and hyperparameters are adjusted according to the evaluation results. And after the training is completed, an independent test set is used to finally test the model to ensure its effectiveness and reliability in actual application scenarios.

[0081] In the embodiments of the present invention, the entity extraction model combines the non - linear modeling ability of the BP neural network and the long - distance dependence capture ability of the Transformer architecture, significantly improving the accuracy, robustness, and generalization ability of entity extraction. At the same time, the Transformer architecture can effectively capture long - distance dependence relationships through the self - attention mechanism, especially suitable for processing long - sequence data. Its parallel computing ability significantly improves the training and inference efficiency. Introducing the FreeLB adversarial training algorithm can enhance the robustness of the model, enabling it to more accurately identify semantic entities when facing highly ambiguous and diverse modal information. Adversarial training helps the model learn more stable feature representations by simulating adversarial samples, thus improving the generalization ability. Replacing the hidden layer of the BP neural network with a spatio - temporal federated learning framework can better handle the complexity of spatio - temporal data. This framework further enhances the model's ability to model spatio - temporal features by introducing advanced statistical functions. Adding a CRF annotation sequence layer embedded with a Softmax layer between the spatio - temporal federated learning framework and the output layer can effectively handle the annotation dependence problem in sequence annotation tasks. The Softmax layer is used to activate the output of the Transformer architecture, further optimizing the output results of the model.

[0082] Step S104: Load the pre - processed historical association data, extract the entity feature attributes of the historical association data based on the entity extraction model, and call the CEA integration algorithm to assign weights to the entity feature attributes from the perspectives of user spatio - temporal behavior, service provider spatio - temporal behavior, and personnel interaction behavior.

[0083] In the embodiments of the present invention, extracting entity feature attributes based on the entity extraction model can quickly locate and extract key information related to spatio - temporal behavior from massive data. The CEA integration algorithm assigns weights to entity feature attributes from different perspectives, further clarifying the importance degree of each feature, helping to highlight key factors, and providing a more targeted basis for subsequent analysis and decision - making.

[0084] Step S105: Associate the entity feature values of the historical association data with the attribute weight values to obtain the user spatio - temporal behavior weight matrix, service provider spatio - temporal behavior weight matrix, and personnel interaction behavior weight matrix of the entity feature attributes.

[0085] Associating the entity feature values with the attribute weight values forms an intuitive weight matrix, realizing a quantitative representation of the complex relationships among user spatio - temporal behavior, service provider spatio - temporal behavior, and personnel interaction behavior. This quantitative method is convenient for computer processing and analysis, providing convenience for subsequent calculations and predictions. For example, in constructing the spatio - temporal behavior map of a tourism service platform, by establishing the user spatio - temporal behavior weight matrix, it is possible to clearly see the tourism behavior preference degrees of different user groups at different times and in different regions, providing data support for the design and promotion of tourism products.

[0086] Step S106, calculate the prediction behavior influence weights of the entity feature attributes in the user spatio-temporal behavior weight matrix, the service provider spatio-temporal behavior weight matrix, and the personnel interaction behavior weight matrix based on the Tanimoto coefficient. Using the Tanimoto coefficient to calculate the prediction behavior influence weights can scientifically evaluate the influence degree of each entity feature attribute on the prediction behavior. This method considers the similarity and difference between features, making the calculated weights more reasonable and accurate, and helping to improve the reliability and effectiveness of the prediction;

[0087] In the embodiment of the present invention, the prediction behavior influence weights are calculated by the following formula:

[0088]

[0089] where q d represents the prediction behavior influence weight of the entity feature attribute, W s a , W s b , W s c are respectively the entity feature values of the entity feature attribute s in the user spatio-temporal behavior weight matrix A, the service provider spatio-temporal behavior weight matrix B, and the personnel interaction behavior weight matrix C, q a , q b , q c are respectively the attribute weight values of the entity feature attribute s, Sim T represents the Tanimoto coefficient of the user spatio-temporal behavior weight matrix, the service provider spatio-temporal behavior weight matrix, and the personnel interaction behavior weight matrix, φ a,b,c represents the embedding vector of the entity feature attribute s in the user spatio-temporal behavior weight matrix A, the service provider spatio-temporal behavior weight matrix B, and the personnel interaction behavior weight matrix C, T represents the number of entity features, |KU i +KU j | represents the user-service provider interaction degree value, |A i +B i +C i |, |A j +B j +C j | respectively represent the number of entity features of the interacting user i and the service provider j in the user spatio-temporal behavior weight matrix A, the service provider spatio-temporal behavior weight matrix B, and the personnel interaction behavior weight matrix C, |A∩B∩C| represents the matrix intersection, and |A∪B∪C| represents the matrix union.

[0090] In the embodiments of the present invention, a calculation method and a calculation formula for predicting the influence weight of behaviors are defined. By combining the weight matrices of user spatio-temporal behaviors, service provider spatio-temporal behaviors, and personnel interaction behaviors, this formula can comprehensively consider information from multiple dimensions. This multi-dimensional integration makes the prediction more comprehensive and accurate. Using embedding vectors and the Tanimoto coefficient to calculate weights, the weights can be dynamically adjusted according to the importance of different entity feature attributes. This improves the adaptability and flexibility of the model to different situations.

[0091] Step S107: Preset a prediction weight threshold, which can be set to 0.15 - 0.3. Based on the prediction weight threshold, extract the entity feature values and attribute weight values of the spatio-temporal behavior weight matrix, service provider spatio-temporal behavior weight matrix, and personnel interaction behavior weight matrix, and obtain a personnel prediction behavior matrix based on user spatio-temporal behaviors, service provider spatio-temporal behaviors, and personnel interaction behaviors.

[0092] Step S108: Load the user spatio-temporal behavior weight matrix, service provider spatio-temporal behavior weight matrix, personnel interaction behavior weight matrix, and personnel prediction behavior matrix. Use the LOAD CSV command to convert the data in the matrix into the triple form of the knowledge graph data layer, store the data in the form of a graph, and visually present it using Neo4j to complete the construction of the spatio-temporal behavior graph. For example, in the tourism planning of a city, by constructing and visually displaying the spatio-temporal behavior graph, urban planners and tourism enterprises can intuitively understand the behavior patterns of tourists, the distribution of tourism resources, and the trends of market demands, so as to formulate more scientific and reasonable tourism development plans and marketing strategies.

[0093] In the embodiments of the present invention, by presetting the prediction weight threshold, features and weights that have a significant impact on the prediction behavior can be screened out, noise and redundant information can be removed, and the accuracy of the prediction can be improved. The obtained personnel prediction behavior matrix directly reflects the comprehensive influence of different behavior factors on personnel's future behaviors, providing strong support for personalized recommendation and service optimization.

[0094] In the embodiments of the present invention, by integrating the four dimensions of user spatio-temporal behaviors, service provider spatio-temporal behaviors, personnel interaction behaviors, and personnel prediction behaviors, the behavior patterns and preferences of tourists can be comprehensively described. The spatio-temporal behavior graph provides rich context information, helps the entity extraction model better understand ambiguous and diverse modal information, and improves the accuracy of entity recognition. Through the spatio-temporal correlation information in the graph, the model can more clearly define entity boundaries, reduce misidentifications. Combining the graph information, the entity extraction model can more accurately identify tourists' needs and generate service work orders that better meet the actual needs.

[0095] The embodiment of the present invention provides a method for semantically understanding tourist demand data based on advanced statistical functions using a spatiotemporal behavior graph. This method achieves accurate semantic understanding of tourist demand data by pre-processing tourist demand data, denoising signal features with inverse fast Fourier transform, fusion statistics in a spatiotemporal federated learning framework, and generating a personnel prediction behavior matrix corresponding to personalized semantic information using a spatiotemporal behavior graph. This method can not only process multimodal data, but also dynamically adjust service content and generate personalized service work orders, significantly improving the quality and efficiency of scenic area services. Figure 3 The schematic diagram of the implementation process of the method for semantically understanding tourist demand data based on the spatiotemporal behavior graph based on advanced statistical functions is shown. The spatiotemporal behavior graph semantically understands tourist demand data based on advanced statistical functions, specifically including:

[0096] Step S201, loading tourist demand data, preprocessing the tourist demand data, and identifying the preprocessed tourist demand data;

[0097] In the embodiment of the present invention, the method of preprocessing tourist demand data includes but is not limited to outlier cleaning and deletion processing, removing noise and redundant information in the data, and improving data quality. By identifying the preprocessed data, a clear and clean data foundation is provided for subsequent semantic understanding, ensuring the accuracy of subsequent processing.

[0098] Step S202, determine the data type, where the data type is determined based on whether it is text data, so that the appropriate processing method can be selected in a targeted manner to avoid a one-size-fits-all processing method. This classification processing can better adapt to the characteristics of different modal data and improve the efficiency and accuracy of semantic understanding;

[0099] Step S203: If the data type is non-text data, the entity extraction model input layer extracts the signal features of the non-text data, and performs denoising on the signal features based on the inverse fast Fourier transform. For non-text data (such as speech, images, etc.), by extracting signal features and using the inverse fast Fourier transform (iFFT) for denoising, noise interference can be effectively removed and key information can be retained. This step significantly improves the availability of non-text data and provides high-quality input for subsequent semantic understanding;

[0100] Step S204, if the data type is text data, the encoder of the Transformer architecture of the entity extraction model serializes and encodes the features of the text data to obtain a text encoding set;

[0101] Step S205, loading signal reconstruction features and text encoding sets, the spatiotemporal federated learning framework combines advanced statistical functions to perform fusion statistics on the signal reconstruction features and text encoding sets to obtain a semantic output set, and the CRF annotation sequence layer annotates the semantic output set with user labels and determines the priority to output personalized semantic information;

[0102] In the embodiments of the present invention, the spatio-temporal federated learning framework combines advanced statistical functions to perform fusion statistics on signal reconstruction features and text encoding sets, can comprehensively process multimodal data, and obtain a more comprehensive semantic output set. The CRF annotation sequence layer further performs user label annotation and priority determination on the semantic output set, providing clear semantic information for personalized services.

[0103] Step S206: Using the personalized semantic information as an index, traverse the spatio-temporal behavior graph based on the retrieval enhancement generation technology. The spatio-temporal behavior graph outputs a user spatio-temporal behavior weight matrix, a service provider spatio-temporal behavior weight matrix, and a personnel interaction behavior weight matrix corresponding to the personalized semantic information in response to the personalized semantic information.

[0104] Step S207: Load the user spatio-temporal behavior weight matrix, the service provider spatio-temporal behavior weight matrix, and the personnel interaction behavior weight matrix corresponding to the personalized semantic information. The spatio-temporal behavior graph generates a personnel prediction behavior matrix corresponding to the personalized semantic information.

[0105] Step S208: Extract the triple entities and entity relationships corresponding to the personnel prediction behavior matrix in the knowledge graph data layer, and load the triple entities and entity relationships in the knowledge graph data layer into a preset work order template to generate a personalized service work order.

[0106] In the embodiments of the present invention, triple entities and relationships in the knowledge graph data layer are extracted and loaded into a preset work order template to generate a personalized service work order. This method not only improves the automation degree of work order generation, but also ensures the accuracy and personalization of the work order content, improving the satisfaction of tourists and the management efficiency of scenic spots.

[0107] In this embodiment, when performing denoising processing on signal features: load the original signal and construct an adjustable screening window, use the clustering spectrum algorithm to identify and suppress discrete amplitudes, then calculate the Euclidean distance between the high-energy amplitude and the conventional amplitude, and determine whether to suppress the high-energy amplitude according to the threshold, and perform an inverse fast Fourier transform (iFFT) on the processed signal features to reconstruct the signal. Specifically, it includes:

[0108] Step S2031: Load the original signal and identify the original amplitude and phase corresponding to the original signal frequency, and construct a screening window with adjustable length and positive and negative directions.

[0109] Step S2032: The screening window identifies the conventional amplitude, discrete amplitude, and high-energy amplitude in the original amplitude based on the clustering spectrum algorithm, and suppresses the discrete amplitude.

[0110] By using the clustering spectrum algorithm to identify discrete amplitudes and high-energy amplitudes, the noise components and useful information in the signal can be accurately distinguished. Compared with the traditional fixed threshold method, this method can more flexibly adapt to the characteristics of different signals and avoid mis-suppressing useful information.

[0111] Step S2033: Calculate the Euclidean distance between the high-energy amplitude and the normal amplitude using the clustering spectrum algorithm, and determine whether the Euclidean distance between the high-energy amplitude and the clustering center of the normal amplitude exceeds the normal amplitude threshold interval.

[0112] In the embodiment of the present invention, by calculating the Euclidean distance between the high-energy amplitude and the normal amplitude and comparing it with the dynamic threshold interval, the suppression strategy can be dynamically adjusted according to the actual characteristics of the signal. This method avoids the problems of over-suppression or under-suppression caused by fixed thresholds.

[0113] Step S2034: If the Euclidean distance between the high-energy amplitude and the clustering center of the normal amplitude exceeds the normal amplitude threshold interval, suppress the high-energy amplitude; if the Euclidean distance between the high-energy amplitude and the clustering center of the normal amplitude does not exceed the normal amplitude threshold interval, retain the current high-energy amplitude.

[0114] Step S2035: Load the original signal frequency after denoising the signal characteristics, and reconstruct the original signal based on the inverse fast Fourier transform to obtain the signal reconstruction characteristics of non-text data.

[0115] In the embodiment of the present invention, by reconstructing the processed signal characteristics through the inverse fast Fourier transform (iFFT), the time-domain characteristics of the signal can be completely restored. This method not only removes noise but also retains the original structure and key features of the signal, ensuring the usability of the signal after reconstruction. The clustering spectrum algorithm and the dynamic threshold adjustment mechanism make the entire signal reconstruction process highly flexible and scalable. It can be adjusted according to different application scenarios and signal characteristics to adapt to various complex environments.

[0116] The embodiment of the present invention provides a method for the interaction degree between personalized service work orders and service providers based on the ant colony optimization algorithm and fuzzy clustering analysis of the in-degree interaction matrix. Briefly, the implementation method is to find the optimal path through the ant colony optimization algorithm, extract the service provider information on the path, calculate the initial interaction degree between the personalized service work order and the service provider, calculate the weighted interaction degree using fuzzy clustering analysis, map the weighted interaction degree sequence to the initial interaction degree sequence to obtain the final interaction degree, and rank the service providers, and update the pheromone weight according to the ranking to determine the optimal service provider. Figure 4 The schematic flow chart of the implementation of the method for the interaction degree between personalized service work orders and service providers based on the ant colony optimization algorithm and fuzzy clustering analysis of the in-degree interaction matrix is shown. The method for the interaction degree between personalized service work orders and service providers based on the ant colony optimization algorithm and fuzzy clustering analysis of the in-degree interaction matrix specifically includes:

[0117] Step S301, load the in-degree interactivity matrix and the associated personalized service work orders, identify the predicted spatio-temporal point information in the personnel prediction behavior matrix based on the personalized service work orders, and use the ant colony optimization algorithm to find l paths from the current location point l1 to the predicted spatio-temporal point l n The ant colony optimization algorithm iteratively searches for these l paths to find the optimal path l best ;

[0118] In the embodiment of the present invention, the ant colony optimization algorithm can efficiently find the optimal path in a complex environment by simulating the behavior of ants looking for food. In this embodiment, the algorithm is used for path search from the current location point to the predicted spatio-temporal point, which can dynamically adapt to environmental changes and update the path in real time. This method not only improves the efficiency of path search but also ensures the optimality of the path, providing a scientific basis for the subsequent selection of service providers.

[0119] Step S302, extract the service providers within the spatio-temporal behavior based on the optimal path l best Calculate the initial interactivity between the personalized service work order and the service provider using the Pearson correlation coefficient, and generate an initial interactivity sequence;

[0120] Calculating the initial interactivity between the personalized service work order and the service provider using the Pearson correlation coefficient can quantify the correlation between the two. This method provides basic data for the subsequent optimization of interactivity, ensuring the scientificity and accuracy of interactivity calculation.

[0121] Step S303, the fuzzy clustering analysis algorithm characterizes the weighted interactivity between the personalized service work order and the service provider through the Minkowski distance to obtain a weighted interactivity sequence. In this embodiment, calculating the weighted interactivity through the Minkowski distance can more accurately reflect the interaction intensity between the personalized service work order and the service provider. This method not only improves the accuracy of interactivity calculation but also enhances the robustness of the system;

[0122] Step S304, load the initial interactivity sequence between the personalized service work order and the service provider and the weighted interactivity sequence between the personalized service work order and the service provider, perform local compression on the time series objects of the personalized service work order and the service provider on the time axis and the space axis, optimally map the weighted interactivity sequence to the initial interactivity sequence to obtain the interactivity between the personalized service work order and the service provider, rank the service providers based on the interactivity between the personalized service work order and the service provider to obtain the ranking position of each service provider, and the ant colony optimization algorithm performs initial pheromone weighted update according to the ranking of the service providers to obtain the optimal service provider.

[0123] In this embodiment, the interactivity between the personalized service work order and the service provider is calculated by the following formula:

[0124]

[0125] Among them, J best , J0, respectively represent the interaction degree, the initial interaction degree, and the weighted interaction degree. δ is the spatio-temporal expectation degree of the service provider in the optimal path l best In this embodiment, it can be 0.2 - 0.8. G, represents the in-degree interaction matrix and the mean value of the in-degree interaction matrix. k is the number of service providers within the spatio-temporal behavior. AG i , AG j respectively represent the weight matrix of the in-degree interaction matrix corresponding to user i and service provider j, respectively represent the in-degree interaction values between user i and service provider j in the in-degree interaction matrix, and can be 0.2 - 5.

[0126] In the embodiment of the present invention, by optimally mapping the weighted interaction degree sequence to the initial interaction degree sequence, it is possible to comprehensively consider the initial and weighted interaction degrees and obtain a more reasonable personalized service order and service provider interaction degree. Sorting the service providers based on this interaction degree and updating the pheromone weights through the ant colony optimization algorithm can dynamically adjust the priorities of the service providers to ensure the selection of the optimal service provider.

[0127] The embodiment of the present invention provides a service order recognition and distribution system based on the spatio-temporal behavior map of tourists, Figure 5 which shows a schematic structural diagram of the service order recognition and distribution system based on the spatio-temporal behavior map of tourists. The service order recognition and distribution system based on the spatio-temporal behavior map of tourists specifically includes:

[0128] A knowledge graph module 100, which is used to crawl historical associated data related to the spatio-temporal behavior of tourists and pre-construct a spatio-temporal behavior map of tourists in combination with the historical associated data;

[0129] A demand recognition module 200, which is used to collect multi-source heterogeneous tourist demand data in real time. The spatio-temporal behavior map performs semantic understanding on the tourist demand data based on advanced statistical functions, triggers a service order generation instruction, and generates a personalized service order in response to the generation instruction;

[0130] A work order indexing module 300, which loads the personalized service order, extracts the work order priority, user tags, and work order information features based on a pre-trained entity extraction model and performs weighted fusion to obtain a fusion feature set. The entity extraction model traverses the spatio-temporal behavior map to obtain a user-service provider interaction matrix, a user-associated person interaction matrix, and a service-associated person interaction matrix based on the fusion feature set, and performs weighted combination on the interaction matrices to obtain an in-degree interaction matrix;

[0131] The work order distribution module 400 is used to input the in-degree interaction matrix into the entity extraction model, analyze the interaction degree between the personalized service work order and the service provider corresponding to the in-degree interaction matrix based on the ant colony optimization algorithm and fuzzy clustering analysis, and distribute the personalized service work order to the corresponding optimal service provider based on the interaction degree.

[0132] In this embodiment, the knowledge graph module 100, the demand recognition module 200, the work order indexing module 300, and the work order distribution module 400, as well as the service work order recognition and distribution system based on the tourist spatio-temporal behavior graph they form, correspond to the above-mentioned service work order recognition and distribution method based on the tourist spatio-temporal behavior graph. For the explanations, examples, beneficial effects, etc. of the relevant content, reference can be made to the corresponding content in the service work order recognition and distribution method based on the tourist spatio-temporal behavior graph, and will not be elaborated here.

[0133] In summary, the present invention provides a service work order recognition and distribution method and system based on the tourist spatio-temporal behavior graph. In the embodiments of the present invention, a tourist spatio-temporal behavior graph including an entity extraction model is pre-constructed in combination with historical association data. The entity extraction model introduces an entity extraction model based on the BP neural network and the Transformer architecture, and combines the FreeLB adversarial training algorithm, which significantly improves the model's processing ability and robustness for complex data. Through the spatio-temporal federated learning framework and advanced statistical functions, the model's ability to model spatio-temporal data is further enhanced, enabling it to more comprehensively capture the spatio-temporal behavior characteristics of tourists and service providers, ensuring the accuracy and efficiency of feature extraction, enabling the spatio-temporal behavior graph to accurately semantically understand tourist demand data, generating personalized semantic information and a personnel prediction behavior matrix based on tourist demand data, dynamically evaluating the matching degree between service providers and work orders, ensuring that personalized service work orders are accurately distributed to the optimal service provider with the highest correlation, thereby improving service efficiency and tourist satisfaction, and overcoming the problem that existing scenic area service work orders may contain various modal information such as voice, text, and pictures, and it is difficult to accurately understand the true intention of users due to the ambiguity, diversity of languages, and the existence of specific scenic area terms.

[0134] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the protection scope of the invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all embodiments. Based on these embodiments, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art can still, without conflict, make combinations, additions, deletions, or other adjustments to the features in the embodiments of the present invention according to the situation without creative efforts, so as to obtain different technical solutions that are essentially not divorced from the concept of the present invention, and these technical solutions also belong to the scope of protection of the present invention.

Claims

1. A method for identifying and distributing service work orders based on a spatio-temporal behavior map of tourists, characterized in that Including: Crawl historical associated data related to tourists' spatio-temporal behavior, and pre-construct a spatio-temporal behavior map of tourists in combination with the historical associated data; Collect multi-source heterogeneous tourists' demand data in real time. The spatio-temporal behavior map conducts semantic understanding on the tourists' demand data based on advanced statistical functions, triggers a service work order generation instruction, and in response to the generation instruction, generates a personalized service work order; Load the personalized service work order, extract the work order priority, user tags, and work order information features of the work order based on a pre-trained entity extraction model and perform weighted fusion to obtain a fused feature set. The entity extraction model traverses the spatio-temporal behavior map to obtain a user-service provider interaction matrix, a user-associated person interaction matrix, and a service-associated person interaction matrix based on the fused feature set, and performs weighted combination on the interaction matrices to obtain an in-degree interactivity matrix; Input the in-degree interactivity matrix into the entity extraction model, analyze the interaction degree between the personalized service work order and the service provider based on the ant colony optimization algorithm and fuzzy clustering for the in-degree interactivity matrix, and distribute the personalized service work order to the corresponding optimal service provider based on the interaction degree.

2. The service order identification and distribution method based on the tourist spatio-temporal behavior map according to claim 1, characterized in that: The method for pre-constructing a spatio-temporal behavior map in combination with historical associated data includes: Define the pattern layer of the map based on four dimensions: user spatio-temporal behavior, service provider spatio-temporal behavior, personnel interaction behavior, and personnel prediction behavior. Use the four dimensions as the ontology of the pattern layer to guide the construction of the data layer of the map; Use the Scrapy crawler framework to crawl multi-source heterogeneous historical associated data, extract historical associated data entities, entity attributes, and entity relationships, and preprocess the historical associated data. Among them, the historical associated data includes user location data, time data, behavior data, external data, user name, gender, age, education level, home address, historical demand data, and destination information; Pre-construct an entity extraction model, and use the historical associated data to iteratively train the entity extraction model to output a converged entity extraction model; Load the preprocessed historical associated data, extract the entity feature attributes of the historical associated data based on the entity extraction model, and call the CEA integration algorithm to assign weights to the entity feature attributes from the perspectives of user spatio-temporal behavior, service provider spatio-temporal behavior, and personnel interaction behavior; Associate the entity feature values of the historical associated data with the attribute weight values to obtain a user spatio-temporal behavior weight matrix, a service provider spatio-temporal behavior weight matrix, and a personnel interaction behavior weight matrix of the entity feature attributes; Calculate the prediction behavior influence weights of the entity feature attributes in the user spatio-temporal behavior weight matrix, the service provider spatio-temporal behavior weight matrix, and the personnel interaction behavior weight matrix based on the Tanimoto coefficient; Preset a prediction weight threshold, and extract the entity feature values and attribute weight values of the spatio-temporal behavior weight matrix, the service provider spatio-temporal behavior weight matrix, and the personnel interaction behavior weight matrix based on the prediction weight threshold to obtain a personnel prediction behavior matrix based on user spatio-temporal behavior, service provider spatio-temporal behavior, and personnel interaction behavior. Load the user spatio-temporal behavior weight matrix, service provider spatio-temporal behavior weight matrix, personnel interaction behavior weight matrix, and personnel prediction behavior matrix. Use the LOAD CSV command to convert the data in the matrix into the triple form of the knowledge graph data layer, store the data in the form of a graph, and visually present it using Neo4j to complete the construction of the spatio-temporal behavior graph.

3. The service order identification and distribution method based on the tourist spatio-temporal behavior map according to claim 2, characterized in that: The entity extraction model is based on the BP neural network model. After the input layer of the basic model, a Transformer architecture is introduced. The Transformer architecture consists of an encoder and a decoder. The FreeLB adversarial training algorithm is introduced into the encoder. The hidden layer of the basic model is replaced by a spatio-temporal federated learning framework. Advanced statistical functions are introduced into the spatio-temporal federated learning framework. A CRF annotation sequence layer embedded with a Softmax layer is added between the spatio-temporal federated learning framework and the output layer. The sigmoid activation function is used to activate the output of the Transformer architecture. The entity extraction model is iteratively trained based on the FreeLB adversarial training algorithm combined with historical association data. The hyperparameters of the entity extraction model are adjusted by the Adma optimizer to minimize the cross-entropy loss function of the entity extraction model.

4. The service order identification and distribution method based on the tourist spatio-temporal behavior map according to claim 2, wherein: The prediction behavior influence weight is calculated by the following formula: Among them, q d represents the prediction behavior influence weight of the entity feature attribute, W s a , W s b , W s c are respectively the entity feature values of the entity feature attribute s in the user spatio-temporal behavior weight matrix A, the service provider spatio-temporal behavior weight matrix B, and the personnel interaction behavior weight matrix C. q a , q b , q c are respectively the attribute weight values of the entity feature attribute s. Sim T represents the Tanimoto coefficient of the user spatio-temporal behavior weight matrix, the service provider spatio-temporal behavior weight matrix, and the personnel interaction behavior weight matrix. φ a,b,c represents the embedding vector of the entity feature attribute s in the user spatio-temporal behavior weight matrix A, the service provider spatio-temporal behavior weight matrix B, and the personnel interaction behavior weight matrix C. T represents the number of entity features, |KU i +KU j | represents the user-service provider interaction degree value, |A i +B i +C i |, |A j +B j +C j | respectively represent the number of entity features of the interacting user i and the service provider j in the user spatio-temporal behavior weight matrix A, the service provider spatio-temporal behavior weight matrix B, and the personnel interaction behavior weight matrix C. |A∩B∩C| represents the matrix intersection, and |A∪B∪C| represents the matrix union.

5. The service order identification and distribution method based on the tourist spatio-temporal behavior map according to claim 3, characterized in that: The spatio-temporal behavior graph performs semantic understanding on the tourist demand data based on advanced statistical functions, including: Load the tourist demand data, preprocess the tourist demand data, identify the preprocessed tourist demand data, and judge the data type; If the data type is non-text data, the input layer of the entity extraction model extracts the signal feature of the non-text data, and performs denoising processing on the signal feature based on the inverse fast Fourier transform; If the data type is text data, the encoder of the Transformer architecture of the entity extraction model serially encodes the features of the text data to obtain a text encoding set; Load the signal reconstruction feature and the text encoding set. The spatio-temporal federated learning framework combines advanced statistical functions to perform fusion statistics on the signal reconstruction feature and the text encoding set to obtain a semantic output set. The CRF annotation sequence layer performs user label annotation and priority determination on the semantic output set, and outputs personalized semantic information; Using the personalized semantic information as an index, traverse the spatio-temporal behavior graph based on the retrieval enhancement generation technology. The spatio-temporal behavior graph outputs the user spatio-temporal behavior weight matrix, service provider spatio-temporal behavior weight matrix, and personnel interaction behavior weight matrix corresponding to the personalized semantic information in response to the personalized semantic information; Load the user spatio-temporal behavior weight matrix, service provider spatio-temporal behavior weight matrix, and personnel interaction behavior weight matrix corresponding to the personalized semantic information. The spatio-temporal behavior graph generates a personnel prediction behavior matrix corresponding to the personalized semantic information; Extract the triple entities and entity relationships corresponding to the personnel prediction behavior matrix in the knowledge graph data layer, and load the triple entities and entity relationships of the knowledge graph data layer into a preset work order template to generate a personalized service work order.

6. The service order identification and distribution method based on the tourist spatio-temporal behavior map according to claim 5, wherein: When performing denoising processing on the signal feature: Load the original signal and identify the original amplitude and phase corresponding to the original signal frequency, and construct a screening window with adjustable length and positive and negative directions; The screening window identifies the conventional amplitude, discrete amplitude, and high-energy amplitude in the original amplitude based on the clustering spectrum algorithm and suppresses the discrete amplitude; The Euclidean distance between the high-energy amplitude and the conventional amplitude is calculated using the clustering spectrum algorithm to determine whether the Euclidean distance between the high-energy amplitude and the clustering center of the conventional amplitude exceeds the conventional amplitude threshold interval; If the Euclidean distance between the high-energy amplitude and the clustering center of the conventional amplitude exceeds the conventional amplitude threshold interval, the high-energy amplitude is suppressed. If the Euclidean distance between the high-energy amplitude and the clustering center of the conventional amplitude does not exceed the conventional amplitude threshold interval, the current high-energy amplitude is retained; The original signal frequency after denoising the signal features is loaded, and the original signal is reconstructed based on the inverse fast Fourier transform to obtain the signal reconstruction features of non-text data.

7. The service order identification and distribution method based on the tourist spatio-temporal behavior map according to claim 5, wherein: The method for the in-degree interaction matrix corresponding to the personalized service work order and the service provider based on the ant colony optimization algorithm and fuzzy clustering analysis includes: Load the incoming degree interaction matrix and the associated personalized service work order. Based on the personalized service work order, identify the predicted spatio-temporal point information in the personnel prediction behavior matrix. Use the ant colony optimization algorithm to find l paths from the current location point l1 to the predicted spatio-temporal point l n Among these l paths, the ant colony optimization algorithm iteratively searches to find the optimal path l best ; Based on the optimal path l best Extract the service providers within the spatio-temporal behavior, calculate the initial interaction degree between the personalized service work order and the service provider using the Pearson correlation coefficient, and generate the initial interaction degree sequence; The fuzzy clustering analysis algorithm characterizes the weighted interaction degree between the personalized service work order and the service provider through the Minkowski distance to obtain the weighted interaction degree sequence; The initial interaction degree sequence between the personalized service work order and the service provider and the weighted interaction degree sequence between the personalized service work order and the service provider are loaded. The time series objects, the personalized service work order and the service provider, are locally compressed on the time axis and the space axis, and the weighted interaction degree sequence is optimally mapped to the initial interaction degree sequence to obtain the interaction degree between the personalized service work order and the service provider. Based on the interaction degree between the personalized service work order and the service provider, the service providers are sorted to obtain the ranking position of each service provider. The ant colony optimization algorithm performs initial pheromone weighted update according to the ranking of the service providers to obtain the optimal service provider.

8. The service order identification and distribution method based on the tourist spatio-temporal behavior atlas according to claim 7, characterized in that: The interaction degree between the personalized service work order and the service provider is calculated by the following formula: Among them, J best , J0, respectively represent the interaction degree, the initial interaction degree, and the weighted interaction degree. δ is the spatio-temporal expectation degree of the service provider in the optimal path l best . represents the in-degree interaction matrix and the mean value of the in-degree interaction matrix. k is the number of service providers in the spatio-temporal behavior. AG i , AG j respectively represent the weight matrix of the in-degree interaction matrix corresponding to user i and service provider j respectively represent the in-degree interaction value between user i and service provider j in the in-degree interaction matrix.

9. A service work order identification and distribution system based on the tourist spatio-temporal behavior map is used to implement the service work order identification and distribution method based on the tourist spatio-temporal behavior map as described in any one of claims 1-8, and is characterized in that: The service work order identification and distribution system based on the tourist spatio-temporal behavior map includes: The knowledge graph module is used to crawl the historical associated data related to the tourist spatio-temporal behavior and pre-construct the tourist spatio-temporal behavior map in combination with the historical associated data; The demand identification module is used to collect multi-source heterogeneous tourist demand data in real time. The spatio-temporal behavior map performs semantic understanding on the tourist demand data based on advanced statistical functions, triggers the service work order generation instruction, and generates a personalized service work order in response to the generation instruction; The work order index module loads the personalized service work order, extracts the work order priority, user label, and work order information features based on the pre-trained entity extraction model and performs weighted fusion to obtain the fusion feature set. The entity extraction model traverses the spatio-temporal behavior map to obtain the user-service provider interaction matrix, user-associated person interaction matrix, and service-associated person interaction matrix based on the fusion feature set, and combines the interaction matrices with weights to obtain the in-degree interaction matrix; The work order distribution module is used to input the in-degree interaction matrix into the entity extraction model, analyze the in-degree interaction matrix corresponding to the personalized service work order and the service provider based on the ant colony optimization algorithm and fuzzy clustering, and distribute the personalized service work order to the corresponding optimal service provider based on the interaction degree.

Images