A personalized travel route recommendation method based on tourists' trust

Abstract

The present invention discloses a personalized travel route recommendation method based on tourist trust, which relates to the field of smart tourism technology. The method comprises the following steps: obtaining travel data, calculating the real-time deviation value of tourist trust, and generating a trust crisis event list; judging whether there are continuous and intensive trust crisis events during the tourist's travel according to the trust crisis event list, and outputting a fuse control message; inputting the fuse control message into a cognitive graph generation engine to build a basic trust relationship network; retrieving path reference data from a trust model parameter library, and performing root cause matching in combination with pre-set attribution templates in an event attribution knowledge base to generate a visual trust repair report; and executing route correction based on the visual trust repair report. The present invention can systematically present the multi-dimensional associations between tourists and service providers by inputting the fuse control message into a cognitive graph generation engine and building a basic trust relationship network in combination with historical interaction data.

Description

Technical Field

[0001] The present invention relates to the field of smart tourism technology, and in particular to a method for recommending personalized tourist routes based on tourist trust. Background Art

[0002] With the continuous development of smart tourism and personalized recommendation systems, dynamic travel route planning based on user behavior data and real-time environmental information has become an important research direction in the field of smart tourism. Traditional travel recommendations mainly rely on static historical data, attraction ratings, and user preference tags to provide tourists with customized itinerary suggestions. In recent years, with the integration and application of technologies such as the Internet of Things (IoT), big data analysis, and artificial intelligence, tourism services have gradually evolved towards intelligent and real-time services. For example, some travel recommendations can now incorporate dynamic factors such as traffic conditions, weather changes, and tourist flow at scenic spots to make partial adjustments to existing routes to enhance the user experience. In addition, knowledge graph-based recommendation methods are also under exploration, leveraging structured semantic relationships to enhance the interpretability and relevance of recommendation results.

[0003] However, existing travel recommendations still have significant limitations in addressing unexpected trust crises: most lack a dynamic modeling mechanism for tourist trust, making it difficult to promptly identify trust biases caused by factors such as resource unavailability, misguided routes, and unexpected events. Existing travel recommendations typically employ trust assessment methods based on historical ratings, user feedback, or static preferences, using collaborative filtering, content recommendation, or hybrid algorithms to roughly estimate tourist trust. These methods primarily rely on static information such as historical user behavior data, attraction ratings, and social connections, lacking the ability to dynamically perceive real-time environmental changes, fluctuations in resource availability, and the impact of unexpected events. Summary of the Invention

[0004] The present invention provides a personalized travel route recommendation method based on tourist trust to solve the problem of failure to timely identify trust deviation caused by factors such as unavailable resources, wrong route guidance and sudden event interference.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] In a first aspect, the present invention provides a method for recommending personalized travel routes based on tourist trust, comprising: obtaining travel data, calculating a real-time deviation value of tourist trust, and generating a list of trust crisis events; the travel data includes location verification data, traffic status data, environmental constraint data, resource availability data, itinerary execution data, and user behavior data;

[0007] According to the trust crisis event list, it is determined whether there are continuous and intensive trust crisis events during the tourist's travel. If the number meets the preset threshold, the trust crisis real-time fuse mechanism is triggered and a fuse control message is output;

[0008] Input the circuit breaker control message into the cognitive graph generation engine to build a basic trust relationship network; retrieve the path reference data in the trust model parameter library, and combine it with the pre-set attribution template in the event attribution knowledge base to match the root cause and generate a visual trust repair report;

[0009] Based on the visual trust repair report, route correction is performed and the trust model parameter library is updated in a targeted manner to complete the trust recovery of personalized travel route recommendations.

[0010] As a preferred solution of the personalized travel route recommendation method based on tourist trust described in the present invention, the trust crisis event list includes the trust crisis event timestamp, tourist identifier, real-time deviation value of tourist trust and status label of potential trust crisis event.

[0011] As a preferred solution of the personalized travel route recommendation method based on tourist trust of the present invention, the method of judging whether there are continuous and intensive trust crisis events during the tourist's travel according to the trust crisis event list is as follows:

[0012] Extract status labels from the trust crisis event list;

[0013] Group the trust records in the trust crisis event list according to the tourist identifier to form an independent trust crisis event time series for each tourist;

[0014] Arrange the trust crisis event time series in the order of trust crisis event timestamps to obtain the interval between two adjacent trust crisis events;

[0015] Based on the tourist preference information in the user behavior data, a trust crisis event priority sorting mechanism is established to sort the trust crisis event time series in descending order of priority value;

[0016] When the interval between two consecutive high-priority trust crisis events is less than the time window threshold, it is determined to be a continuous and intensive trust crisis event.

[0017] As a preferred embodiment of the personalized travel route recommendation method based on tourist trust of the present invention, the following steps are implemented: if a preset number threshold is met, a trust crisis real-time fuse mechanism is triggered, and a fuse control message is outputted: when the number of consecutive and intensive trust crisis events exceeds a number threshold set based on historical trust crisis events, the trust crisis real-time fuse mechanism is triggered, personalized travel route recommendations are suspended, and a fuse control message is generated;

[0018] The fuse control message includes a visitor identifier, a fuse triggering timestamp, a fuse reason description, and recommended countermeasures.

[0019] As a preferred solution of the personalized travel route recommendation method based on tourist trust in the present invention, wherein: the fuse control message is input into the cognitive graph generation engine to build a basic trust relationship network, specifically as follows:

[0020] Extract historical interaction data based on the visitor identifier in the circuit breaker control message;

[0021] Retrieve interaction records from historical interaction data, extract service provider information, service link information, and interaction feedback information, and form service provider nodes, service link nodes, and interaction feedback nodes;

[0022] Through the association rules of interaction records, the connection relationship between service provider nodes, service link nodes and interaction feedback nodes is determined, and integrated into a graph structure to output the basic trust relationship network.

[0023] As a preferred solution of the personalized travel route recommendation method based on tourist trust of the present invention, the path reference data in the trust model parameter library is retrieved and combined with the attribution template preset in the event attribution knowledge base to perform root cause matching, specifically as follows:

[0024] Extracting path reference data matching the tourist identifier from the trust model parameter library;

[0025] Overlaying the path reference data onto the basic trust relationship network to form an extended trust relationship network;

[0026] Extract the circuit breaker reason description from the circuit breaker control message to obtain the trust crisis event description information;

[0027] A root cause matching model based on random forests is used to process the triggering causes in the trust crisis event description information, the real-time deviation value of tourists' trust, and the type of attraction to generate a comprehensive feature vector.

[0028] Compare the comprehensive feature vector with the pre-set attribution templates in the event attribution knowledge base to match the root cause;

[0029] The matching root causes are marked on the nodes of the extended trust relationship network, and a visual trust repair report is generated.

[0030] As a preferred solution of the personalized travel route recommendation method based on tourist trust in the present invention, the root cause matching model based on random forest is used to process the triggering cause in the trust crisis event description information, the real-time deviation value of the tourist trust and the type of attraction to generate a comprehensive feature vector, as follows:

[0031] Using text segmentation methods, the trust crisis event description information is separated by fields, and the trigger cause field, the real-time deviation value field of tourist trust, and the attraction type field are identified;

[0032] Perform category coding on the triggering reason field and map it to the triggering reason code value;

[0033] Extracting the real-time deviation feature value from the real-time deviation value field of the tourist's trust;

[0034] Encode the attraction type field into a category and map it to the attraction type code value;

[0035] The combined trigger cause coding value, real-time deviation characteristic value and scenic spot type coding value are converted into a comprehensive characteristic vector.

[0036] As a preferred solution of the personalized travel route recommendation method based on tourist trust described in the present invention, the route correction based on the visual trust repair report refers to adjusting the service links in the personalized travel route recommendation by utilizing the root cause annotations and recommended countermeasures in the visual trust repair report.

[0037] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the method for recommending personalized travel routes based on tourist trust as described in the first aspect of the present invention is implemented.

[0038] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the personalized travel route recommendation method based on tourist trust as described in the first aspect of the present invention.

[0039] The beneficial effects of the present invention are as follows: by inputting the fuse control message into the cognitive graph generation engine and building a basic trust relationship network in combination with historical interaction data, the multi-dimensional relationship between tourists and service providers can be systematically presented; the cognitive graph generation engine realizes the structured expression of service links, feedback information and responsible entities through node modeling and weight distribution, thereby improving the accuracy and explainability of the attribution of trust crisis events; at the same time, with the help of the extended trust relationship network formed by the superposition of path reference data, the dynamic tracking capability of the evolution of tourists' trust status is further enhanced; finally, it provides solid support for the generation of visual trust repair reports, effectively promoting the trust recovery and service quality optimization of the personalized tourism recommendation system. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0041] Figure 1 Flowchart of the personalized travel route recommendation method based on tourists' trust. DETAILED DESCRIPTION

[0042] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0043] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0044] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0045] Reference Figure 1 , is an embodiment of the present invention, which provides a personalized travel route recommendation method based on tourist trust, comprising the following steps:

[0046] S1. Obtain tourism data, calculate the real-time deviation value of tourists' trust, and generate a list of trust crisis events.

[0047] Furthermore, tourism data includes location verification data, traffic status data, environmental constraint data, resource availability data, trip execution data, and user behavior data.

[0048] The location verification data and itinerary execution data are used to match the tourist's actual location with the location of the scenic spot in the itinerary plan, identify the specific scenic spot location where the tourist is currently located, and calculate the difference between the tourist's actual arrival time at the scenic spot and the recommended arrival time in the itinerary plan as the actual arrival time deviation. The actual arrival time deviation is expressed as:

[0049] ;

[0050] in, is the actual arrival time deviation (time unit, such as minutes or hours), The actual time when tourists arrive at the scenic spot. Recommended arrival time for trip planning;

[0051] The traffic status data is associated with the path between the tourist's current location and the next attraction. The expected travel time under normal circumstances is estimated by combining historical route traffic information. This is compared with the actual travel time, and the difference between the actual travel time and the expected travel time is calculated as the actual travel time deviation. The actual travel time deviation is expressed as:

[0052] ;

[0053] in, is the actual travel time deviation, The actual travel time of tourists, that is, the actual time spent from the current location to the next attraction, The expected travel time under normal circumstances is estimated based on historical route traffic information (such as traffic conditions and historical traffic records);

[0054] Analyze the weather information in the environmental constraint data to determine whether the current weather has an impact on tourists' travel. If an extreme weather event is detected, mark the time period when the extreme weather event is detected as a special environmental interference period.

[0055] Extract rainfall, wind speed, temperature and visibility indicators from the weather real-time information in the environmental constraint data; compare the extracted rainfall with the preset rainfall threshold (based on meteorological standard definition), such as rainfall exceeding 10 mm per hour; compare the extracted wind speed with the preset wind speed threshold (based on wind force level standard definition), such as wind speed exceeding 10 meters per second; compare the extracted temperature with the preset temperature range (based on human comfort temperature range and tourism activity suitability definition), such as temperature below 0 degrees Celsius or above 35 degrees Celsius; compare the extracted visibility with the preset visibility threshold (based on aviation safety standard definition); The system compares the rainfall, wind speed, temperature, or visibility indicators with the corresponding threshold values ​​(defined by the local standard), for example, visibility is less than 100 meters; if any of the indicators of rainfall, wind speed, temperature, or visibility exceeds the corresponding rainfall threshold, wind speed threshold, temperature range, or visibility threshold, the current weather is determined to be an extreme weather event; the timestamp of the extreme weather event is recorded, and the time period affected by the extreme weather event is determined, for example, 30 minutes from the timestamp of the extreme weather event; the time period affected by the extreme weather event is marked as a special environmental interference period, and the marking information including the timestamp of the extreme weather event and the duration of the time period affected by the extreme weather event is stored in the environmental constraint database;

[0056] Compare the operating status of the attractions during the planned visit time and the actual visit time based on the opening status of the attractions in the resource availability data. If the attraction is found to be temporarily closed without prior notice, mark the temporary closure event as a resource anomaly and record the timestamp of the temporary closure event.

[0057] Different weights are assigned to arrival time deviation, actual travel time deviation, resource anomalies, and special environmental interference periods, and the weighted scores are used to form a comprehensive trust score of tourists at the current moment. The comprehensive trust score at the current moment is expressed as:

[0058] ;

[0059] in, A comprehensive score for the trustworthiness at the current moment. Resource exception mark (a value of 1 indicates an exception, and 0 indicates no exception). It is a mark for the special environmental interference period (a value of 1 indicates interference, and 0 indicates no interference). is the weight of the actual arrival time deviation, is the weight of the actual travel time deviation, is the weight of the resource exception mark, The weights marked for special environmental disturbance periods;

[0060] Compare the current trust comprehensive score with the historical average trust, and calculate the change rate between the current trust comprehensive score and the historical average trust as the real-time deviation value of the tourist trust. The real-time deviation value of the tourist trust is expressed as:

[0061] ;

[0062] in, is the real-time deviation value of tourists’ trust, is the historical average trust;

[0063] A trust threshold is set based on the historical tourist trust deviation value. When the real-time deviation value of tourist trust exceeds the trust threshold, it is determined to be a potential trust crisis event;

[0064] Record the timestamp of the potential trust crisis event, the corresponding tourist identifier, the real-time deviation value of the tourist's trust, and the triggering cause into the trust crisis event database;

[0065] Processing potential trust crisis event records in a trust crisis event database; specifically, extracting all potential trust crisis event records from the trust crisis event database, each potential trust crisis event record containing the timestamp of the potential trust crisis event, the corresponding tourist identifier, the real-time deviation value of the tourist's trust, and the triggering cause (including excessive deviation in actual arrival time, excessive deviation in actual travel time, resource anomalies, or special environmental interference period); arranging the potential trust crisis event records in ascending order according to their timestamps to generate a chronologically sorted list of potential trust crisis event records; attaching an event status label to each potential trust crisis event record, where the event status label is divided into active and inactive states; active status indicates that the time difference between the timestamp of the potential trust crisis event and the current time is less than 1 hour; inactive status indicates that the time difference between the timestamp of the potential trust crisis event and the current time is greater than or equal to 1 hour; adding the event status label to each potential trust crisis event record in the form of a field, where the field value is 1 for active state and 0 for inactive state; storing the sorted list of potential trust crisis event records back into the trust crisis event database, containing the timestamp, tourist identifier, real-time deviation value of the tourist's trust, the triggering cause, and the event status label;

[0066] A trust crisis event list is generated based on the sorted potential trust crisis event record list; the trust crisis event list includes the timestamp of the trust crisis event, the visitor identifier, the real-time deviation value of the visitor's trust, and the status label of the potential trust crisis event.

[0067] S2. Determine whether there are continuous and intensive trust crisis events during the tourist's travels based on the trust crisis event list. If the number meets the preset threshold, trigger the trust crisis real-time fuse mechanism and output a fuse control message.

[0068] Furthermore, a trust crisis event list is received, and a timestamp of the trust crisis event, a visitor identifier, a real-time deviation value of the visitor's trust, and a status label of the potential trust crisis event are extracted from the trust crisis event list;

[0069] The trust records in the trust crisis event list are grouped according to the tourist identifier to form an independent trust crisis event time series for each tourist. The trust record includes the timestamp of the trust crisis event in the trust crisis event list, the tourist identifier, the real-time deviation value of the tourist's trust, and the status label of the potential trust crisis event.

[0070] Arrange the time series of each tourist's trust crisis events in the order of their timestamps, and obtain the interval between two adjacent trust crisis events;

[0071] Based on the tourist preference information in the user behavior data, a trust crisis event priority sorting mechanism is established. Specifically, the following steps are taken: extract tourist preference information from the user behavior data, such as preference for historical and cultural attractions or natural scenic spots; determine whether the triggering cause of the potential trust crisis event involves the type of attraction preferred by tourists, such as if a resource anomaly occurs at a historical and cultural attraction preferred by tourists; if the triggering cause involves the type of attraction in the tourist preference information, the potential trust crisis event is marked as high priority with a priority value of 2; if the triggering cause does not involve the type of attraction in the tourist preference information, the potential trust crisis event is marked as low priority with a priority value of 1; sort the potential trust crisis events in the trust crisis event time series in descending order of priority value, with high-priority events given priority; when the interval between two consecutive high-priority potential trust crisis events is less than the time window threshold (set based on the tourist travel density and historical travel time interval in the user behavior data, for example, 20 minutes), it is determined to be a continuous and intensive trust crisis event;

[0072] A quantity threshold is set based on historical trust crisis events, and a real-time trust crisis circuit breaker mechanism is triggered according to the number of consecutive and intensive trust crisis events to generate a circuit breaker control message; specifically: historical trust crisis event records are extracted from the trust crisis event database, the total number of potential trust crisis events for each tourist identifier in the past 30 days is summarized, and the average number of potential trust crisis events per day is determined; based on the average number of potential trust crisis events per day, a fixed value is selected as the quantity threshold, for example, if the average number is 2 times per day, the quantity threshold is selected as 3; the number of consecutive and intensive trust crisis events is compared with the quantity threshold; if the number of consecutive and intensive trust crisis events exceeds the quantity threshold, for example, more than 3 times, the circuit breaker mechanism is activated; the circuit breaker mechanism includes suspending the personalized travel route associated with the current tourist identifier Recommendation: record the timestamp of the fuse trigger; generate a description of the fuse reason, which includes the number of trust crisis events that occur continuously and intensively and the triggering reasons, such as "the interval between three consecutive high-priority trust crisis events is less than 20 minutes, and the triggering reasons include resource anomalies and excessive deviation in actual travel time"; extract tourist preference information from user behavior data, such as preference for historical and cultural attractions, screen alternative attractions that match the tourist preference information from the resource availability data, and generate recommended response measures, such as "it is recommended to replace the next preferred attraction with historical and cultural attraction X"; integrate the tourist identifier, fuse trigger timestamp, fuse reason description and recommended response measures to form a fuse control message; among them, the number of trust crisis events that occur continuously and intensively is determined based on the time interval of high-priority potential trust crisis events.

[0073] S3. Input the circuit breaker control message into the cognitive graph generation engine to build a basic trust relationship network; retrieve the path reference data in the trust model parameter library, and combine it with the preset attribution template in the event attribution knowledge base to match the root cause and generate a visual trust repair report.

[0074] Furthermore, the circuit breaker control message is fed into the cognitive graph generation engine;

[0075] Extract historical interaction data related to the tourist identifier in the cognitive graph generation engine to form node information of service providers, service links and interactive feedback; specifically: retrieve the interaction record matching the current tourist identifier from the historical interaction data, which contains the interaction record between the tourist and the service provider; extract the service provider information in the interaction record, including the name and type of the service provider, such as travel agency A or scenic spot manager B; extract the service link information in the interaction record, including specific service content, such as ticket booking, tour guide service or scenic spot maintenance; extract the interactive feedback information in the interaction record, including tourist evaluation text and rating, such as "the service response is slow, the rating is 3 points" "; convert the service provider information into a service provider node, and the service provider node attributes include the service provider name and type; convert the service link information into a service link node, and the service link node attributes include the service content and occurrence time; convert the interactive feedback information into an interactive feedback node, and the interactive feedback node attributes include the evaluation text, score and feedback time; assign unique identifiers to the service provider node, service link node and interactive feedback node, and the unique identifier is generated based on the visitor identifier and the timestamp of the interaction record, such as "visitor ID_timestamp"; store the service provider node, service link node and interactive feedback node in the temporary database of the cognitive map generation engine;

[0076] Based on the association rules in the historical interaction data, the connection relationship between the service provider, service link and interactive feedback is established to generate a basic trust relationship network; specifically: extract the service provider node, service link node and interactive feedback node from the temporary database of the cognitive map generation engine; extract the association information of the interaction record from the historical interaction data, and the interaction record contains the correspondence between the service provider, service link and interactive feedback; according to the correspondence of the interaction record, determine the connection between the service provider node and the service link node, and the service execution attribute of the connection is the service link executed by the service provider; according to the correspondence of the interaction record, determine the connection between the service link node and the interactive feedback node, and the feedback association attribute of the connection is the feedback of the tourist on the service link; assign a weight to each connection, and the weight is based on the interactive feedback. The score of the feedback node ranges from 1 to 5, and the weight is selected from a predefined score weight table (defined based on the score distribution in the historical interaction data). The higher the score of the interactive feedback node, the greater the weight. If there is no score in the interaction record, the weight is selected by the default value. The connection relationship between the service provider node, service link node and interactive feedback node, and the service provider node, service link node and interactive feedback node are integrated into a graph structure. The nodes in the graph structure are service provider nodes, service link nodes and interactive feedback nodes, and the edges are the connections between nodes. The edge attributes include service execution attributes, feedback association attributes and weights. The graph structure is stored in the temporary database of the cognitive graph generation engine and named as the basic trust relationship network. The basic trust relationship network is output as the basis for the subsequent formation of an extended trust relationship network.

[0077] Extract interaction records between tourists and service providers from historical interaction data; generate path reference data based on the interaction records, the path reference data including the trend of changes in tourist trust under normal circumstances, the expected performance of service links, and the typical reaction patterns of tourists; store the path reference data in a trust model parameter library, which is a dedicated database used to record historical patterns related to tourist trust; retrieve path reference data matching the current tourist identifier from the trust model parameter library, the path reference data including the trend of changes in tourist trust (e.g., the average change in ratings with service links), the expected performance of service links (e.g., the response time standard for ticket booking), and the typical reaction patterns of tourists (e.g., common feedback on delayed services); output the path reference data for use in forming an extended trust relationship network;

[0078] Overlaying the path reference data onto the basic trust relationship network to form an extended trust relationship network that includes a comparison between the normal trust path and the current trust status;

[0079] Extract the description information of the trust crisis event of the fuse reason description from the fuse control message, use the root cause matching model based on random forest, compare the preset attribution template in the event attribution knowledge base, and match the root cause corresponding to the current tourist trust crisis event; specifically: extract the fuse reason description from the fuse control message; use the text segmentation method to separate the description information of the trust crisis event by field, identify the triggering cause field (such as "resource abnormality" and "actual travel time deviation is too large"), the real-time deviation value field of tourist trust (such as "real-time deviation value 0.6") and the attraction type field (such as "historical and cultural attractions"); classify the triggering cause field and map each triggering cause to a triggering cause code value. The triggering causes include the actual arrival time deviation is too large (coded as 0), the actual travel time deviation is too large ( The triggering cause code value, real-time deviation feature value, and attraction type code value are combined into a comprehensive feature vector, for example, [2, 0.6, 1] represents resource anomaly, real-time deviation value 0.6, and historical and cultural attraction; the comprehensive feature vector is stored in the trust crisis event database, including the visitor identifier and the trust crisis event timestamp;

[0080] The comprehensive feature vector is input into the root cause matching model based on random forest; specifically, the comprehensive feature vector is extracted from the trust crisis event database; the root cause matching model based on random forest is loaded, and the root cause matching model is a pre-trained random forest classifier. The root cause matching model is trained using historical trust crisis event records and historical interaction data in the trust crisis event database. The training data includes historical triggering causes, real-time deviation values ​​of tourist trust, attraction types, and manually annotated root causes; the comprehensive feature vector is used as input to the root cause matching model; the comprehensive feature vector is processed based on the root cause matching model, and the root cause category corresponding to the comprehensive feature vector is predicted through multiple decision tree classification. The root cause category comes from the preset attribution templates in the event attribution knowledge base, including information update delay, service response delay, experience deviation from publicity, and communication failure; the root cause matching model outputs the root cause category with the highest confidence, such as information update delay, with a confidence of 0.85; the predicted root cause category and confidence are recorded in the trust crisis event database, including the tourist identifier, trust crisis event timestamp, and comprehensive feature vector, and the predicted root cause category is output;

[0081] It should be noted that the training process of the random forest classifier is as follows: historical trust crisis event records are extracted from the trust crisis event database. The historical trust crisis event records include triggering causes, real-time deviation values ​​of tourist trust, types of attractions, and manually annotated root causes; interaction records between tourists and service providers are extracted from historical interaction data. The interaction records include interaction frequency and feedback sentiment as auxiliary features; the triggering causes in the historical trust crisis event records are categorized, for example, the actual arrival time deviation is coded as 0, the actual travel time deviation is coded as 1, resource anomalies are coded as 2, and special environmental interference periods are coded as 3; the types of attractions are categorized, for example, historical and cultural attractions are coded as 1, natural scenery attractions are coded as 2, and theme parks are coded as 3; the coded triggering causes, real-time deviation values ​​of tourist trust, coded types of attractions, interaction frequency, and feedback sentiment are combined into a training comprehensive feature vector, for example, [2, 0.6, 1, 5, 0.8] Indicates resource anomalies, deviation value 0.6, historical and cultural attractions, interaction frequency 5 times and feedback sentiment 0.8; the manually labeled root causes are used as target variables and encoded as target values, for example, information update delay is encoded as 0, service response delay is encoded as 1, experience deviation from publicity is encoded as 2, and communication failure is encoded as 3; the training comprehensive feature vector and target variables are divided into training set and test set, with the training set accounting for 80% and the test set accounting for 20%; use the random forest classifier based on existing machine learning libraries, such as scikit-learn, to set 100 decision trees with a maximum depth of 10; input the training set into the random forest classifier for training to generate a root cause matching model based on random forest; use the test set to evaluate the model performance and verify the matching accuracy of the root cause categories output by the model with the manually labeled ones, for example, the accuracy reaches 85%; save the trained root cause matching model based on random forest to process the comprehensive feature vector to predict the root cause category.

[0082] The root causes of the random forest model output are marked on the key nodes of the extended trust relationship network, marking the specific service links that caused the trust crisis, such as the service provider's failure to update the opening status of the scenic spot in a timely manner;

[0083] Based on the annotated extended trust relationship network, a visual trust repair report is generated; the visual trust repair report includes visitor identifiers, trust evolution paths, deviation analysis, root cause annotations, and recommended countermeasures.

[0084] S4. Based on the visual trust repair report, perform route correction and perform targeted updates on the trust model parameter library to complete the trust recovery of personalized travel route recommendations.

[0085] Going a step further, receive a visual trust repair report;

[0086] Based on the root cause annotations in the visual trust repair report, locate the specific service link that caused the trust crisis. The specific service link includes at least one of information update delay, service response delay, experience deviation from publicity, and communication failure. The specific service link is a specific cause selected from the root cause.

[0087] Based on the recommended response measures, perform route corrections on service links with trust risks in personalized travel route recommendations; route corrections include replacing service providers, adjusting the itinerary sequence, and changing the content of service links;

[0088] The occurrence time of this trust crisis event, tourist identifier, root cause type and route correction result are used as feedback data and stored in the trust model parameter library; specifically: the occurrence time of this trust crisis event is extracted from the visual trust repair report, and the occurrence time is the timestamp of the trust crisis event; the tourist identifier is extracted, and the tourist identifier is consistent with the tourist identifier in the trust crisis event list; the root cause type is extracted; the route correction result is extracted, and the route correction result is the correction operation performed based on the recommended response measures. The route correction result includes a specific description of replacing the service provider, adjusting the itinerary sequence or changing the content of the service link; the occurrence time, tourist identifier, root cause type and route correction result are integrated into feedback data, and the feedback data is organized in a structured format, for example, including the fields "occurrence time of the trust crisis event", "tourist ID", "root cause" and "correction description"; the feedback data is stored in the trust model parameter library, and the feedback data is added as a new record to the feedback data table of the trust model parameter library. The feedback data table and the path reference data table are stored together in the trust model parameter library;

[0089] In the trust model parameter library, based on the tourist identifier and the root cause type, the corresponding historical path reference data record is matched; specifically: the record matching the current tourist identifier is retrieved from the path reference data table of the trust model parameter library; the root cause type is extracted from the feedback data; the tourist identifier and the root cause type are used as query conditions to find the matching historical path reference data record in the path reference data table, the historical path reference data record contains the trend of changes in tourist trust under normal circumstances, the expected performance of the service link and the typical reaction pattern of tourists, such as the trust decline trend related to information update delay; if a matching record is found, the unique identifier and content of the record are extracted; if no matching record is found, a default record that matches the tourist identifier but has an empty root cause type is selected; the matching historical path reference data record is stored in the temporary cache of the trust model parameter library, the record format includes the tourist identifier, the root cause type and the path reference data content, and the matching historical path reference data record is output;

[0090] Use feedback data to perform targeted updates on matched historical path reference data records, enhancing the ability to predict trust change trends for corresponding tourist types and service links. Specifically, compare the root cause type and route correction results in the feedback data with the tourist trust change trend in the historical path reference data record, and update the change trend, such as adjusting the trust decline caused by information update delays to a more moderate change trend; compare the route correction results with the expected performance of the service link in the historical path reference data record, and update the expected performance, such as adjusting the expected response time of the tour guide service to a shorter time; compare the typical tourist reaction pattern with the route correction results in the feedback data, and update the reaction pattern, such as adjusting negative feedback on delayed service to neutral feedback; write the updated tourist trust change trend, expected performance of the service link, and typical tourist reaction pattern into the historical path reference data record to form a new historical path reference data record; store the new historical path reference data record in the path reference data table of the trust model parameter library, replacing the original matched historical path reference data record, ensuring consistency with the tourist identifier and root cause type; and output the updated historical path reference data record.

[0091] Based on the updated trust model parameter library, a personalized travel route recommendation algorithm is run to generate personalized travel route recommendation results that are consistent with the current trust status of tourists. Specifically: based on the trend of changes in tourists' trust, attractions and service providers that match the current trust status of tourists are screened. The current trust status of tourists is determined by the root cause type and route correction results in the feedback data, such as avoiding service providers with delayed information updates. Based on tourist preference information, attractions that meet the preferred attraction types and itinerary density are given priority, such as tourists who prefer historical and cultural attractions give priority to related attractions. Based on the expected performance of the service links, the order and content of the service links in the itinerary are adjusted, such as changing the response time. The service providers with shorter service time are arranged in the early stage of the trip; candidate tourist routes are generated, which include scenic spot sequences, service provider allocations and itinerary time arrangements; the candidate tourist routes are sorted according to the trend of changes in tourists' trust and tourists' typical reaction patterns, and routes that can enhance trust are given priority, such as routes where tourists have neutral reactions to delayed services; the candidate tourist route with the highest ranking is selected as the personalized tourist route recommendation result; the personalized tourist route recommendation result is stored in the trust model parameter library, which includes the tourist identifier, recommendation result and generation timestamp; the personalized tourist route recommendation result is output and used to display the trust repair completion status on the tourist terminal interface.

[0092] This embodiment also provides a computer device suitable for the personalized travel route recommendation method based on tourist trust, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the personalized travel route recommendation method based on tourist trust as proposed in the above embodiment.

[0093] The computer device may be a terminal, comprising a processor, memory, a communication interface, a display, and an input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system and computer programs. The internal memory provides an environment for the operating system and computer programs stored in the non-volatile storage media. The communication interface of the computer device is used to communicate with external terminals via wired or wireless communication. Wireless communication may be achieved via Wi-Fi, a carrier network, NFC (near-field communication), or other technologies. The display of the computer device may be a liquid crystal display or an electronic ink display. The input device may be a touchscreen overlay on the display, buttons, a trackball, or a touchpad on the computer device housing, or an external keyboard, touchpad, or mouse.

[0094] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for implementing personalized travel route recommendation based on tourist trust as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0095] In summary, the present invention can systematically present the multi-dimensional relationship between tourists and service providers by: inputting the fuse control message into the cognitive graph generation engine, and building a basic trust relationship network based on historical interaction data; the cognitive graph generation engine realizes the structured expression of service links, feedback information and responsible entities through node modeling and weight distribution, thereby improving the accuracy and explainability of the attribution of trust crisis events; at the same time, with the help of the extended trust relationship network formed by the superposition of path reference data, the dynamic tracking ability of the evolution of tourists' trust status is further enhanced; finally, it provides solid support for the generation of visual trust repair reports, effectively promoting the trust recovery and service quality optimization of the personalized tourism recommendation system.

[0096] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A personalized travel route recommendation method based on tourist trust, characterized by: include, Acquire tourism data, calculate real-time deviation values ​​of tourist trust, and generate a list of trust crisis events; the tourism data includes location verification data, traffic status data, environmental constraint data, resource availability data, itinerary execution data, and user behavior data; According to the trust crisis event list, it is determined whether there are continuous and intensive trust crisis events during the tourist's travel. If the number meets the preset threshold, the trust crisis real-time fuse mechanism is triggered and a fuse control message is output; Input the circuit breaker control message into the cognitive graph generation engine to build a basic trust relationship network; Retrieve the path reference data from the trust model parameter library, combine it with the pre-set attribution templates in the event attribution knowledge base to perform root cause matching, and generate a visual trust repair report; Based on the visual trust repair report, the route is corrected and the trust model parameter library is updated in a targeted manner to restore trust in personalized travel route recommendations. The path reference data in the trust model parameter library is retrieved and combined with the preset attribution template in the event attribution knowledge base to perform root cause matching, as follows: Extracting path reference data matching the tourist identifier from the trust model parameter library; Overlaying the path reference data onto the basic trust relationship network to form an extended trust relationship network; Extract the circuit breaker reason description from the circuit breaker control message to obtain the trust crisis event description information; A root cause matching model based on random forests is used to process the triggering causes in the trust crisis event description information, the real-time deviation value of tourists' trust, and the type of attraction to generate a comprehensive feature vector. Compare the comprehensive feature vector with the pre-set attribution templates in the event attribution knowledge base to match the root cause; The matching root causes are marked on the nodes of the extended trust relationship network, and a visual trust repair report is generated.

2. The personalized travel route recommendation method based on tourist trust as claimed in claim 1, characterized in that: The trust crisis event list includes a trust crisis event timestamp, a visitor identifier, a real-time deviation value of the visitor's trust, and a status label of a potential trust crisis event.

3. The personalized travel route recommendation method based on tourist trust as claimed in claim 1, characterized in that: The method of judging whether a tourist has a series of trust crisis events during his / her travels according to the trust crisis event list is as follows: Extract status labels from the trust crisis event list; Group the trust records in the trust crisis event list according to the tourist identifier to form an independent trust crisis event time series for each tourist; Arrange the trust crisis event time series in the order of trust crisis event timestamps to obtain the interval between two adjacent trust crisis events; Based on the tourist preference information in the user behavior data, a trust crisis event priority sorting mechanism is established to sort the trust crisis event time series in descending order of priority value; When the interval between two consecutive high-priority trust crisis events is less than the time window threshold, it is determined to be a continuous and intensive trust crisis event.

4. The personalized travel route recommendation method based on tourist trust as claimed in claim 1, characterized in that: If the preset number threshold is met, the trust crisis real-time fuse mechanism is triggered. Outputting the fuse control message means that when the number of consecutive and intensive trust crisis events exceeds the number threshold set based on historical trust crisis events, the trust crisis real-time fuse mechanism is triggered, personalized travel route recommendations are suspended, and a fuse control message is generated; The fuse control message includes a visitor identifier, a fuse triggering timestamp, a fuse reason description, and recommended countermeasures.

5. The personalized travel route recommendation method based on tourist trust as claimed in claim 1, characterized in that: The fuse control message is input into the cognitive graph generation engine to build a basic trust relationship network, as follows: Extract historical interaction data based on the visitor identifier in the circuit breaker control message; Retrieve interaction records from historical interaction data, extract service provider information, service link information, and interaction feedback information, and form service provider nodes, service link nodes, and interaction feedback nodes; Through the association rules of interaction records, the connection relationship between service provider nodes, service link nodes and interaction feedback nodes is determined, and integrated into a graph structure to output the basic trust relationship network.

6. The personalized travel route recommendation method based on tourist trust as claimed in claim 1, characterized in that: The root cause matching model based on random forest is used to process the triggering cause in the trust crisis event description information, the real-time deviation value of the tourist trust and the type of attraction to generate a comprehensive feature vector, as follows: Using text segmentation methods, the trust crisis event description information is separated by fields, and the trigger cause field, the real-time deviation value field of tourist trust, and the attraction type field are identified; Perform category coding on the triggering reason field and map it to the triggering reason code value; Extracting the real-time deviation feature value from the real-time deviation value field of the tourist's trust; Encode the attraction type field into a category and map it to the attraction type code value; The combined trigger cause coding value, real-time deviation characteristic value and scenic spot type coding value are converted into a comprehensive characteristic vector.

7. The personalized travel route recommendation method based on tourist trust as claimed in claim 1, characterized in that: The performing route correction based on the visual trust repair report refers to adjusting the service links in the personalized travel route recommendation by utilizing the root cause annotations and recommended countermeasures in the visual trust repair report.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the personalized travel route recommendation method based on tourist trust as described in any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the personalized travel route recommendation method based on tourist trust are implemented as described in any one of claims 1 to 7.

Images