Tourism recommendation platform based on knowledge graph

Abstract

The invention relates to the technical field of tourism recommendation platforms, and discloses a knowledge graph-based tourism recommendation platform, which comprises a multi-role cognitive portrait construction module for constructing a structured multi-role cognitive portrait; the cognitive enhanced dynamic knowledge graph construction module is used for marking specific requirements of tourism resource entities on cognitive competence of the user; the multi-role demand fusion and conflict detection module is used for identifying potential conflicts and generating a coordination strategy; the cognitive adaptability itinerary generation and optimization module is used for generating an itinerary scheme giving consideration to cognitive demands of all members; and the experience feedback and cognitive model iteration module is used for calibrating cognitive portraits and resource cognitive attributes. According to the scheme, the dynamic knowledge graph fusing the user cognitive portraits and the resource cognitive attributes is constructed, refined modeling of multi-role group travel demands is achieved, the elastic travel taking consideration of cognitive suitability, space-time constraint and budget is generated, the model and resource annotation are continuously optimized by means of feedback closed loop, and the user experience is improved. And the heterogeneous group participation degree and the stroke feasibility are improved.

Description

Technical Field

[0001] This invention relates to the field of tourism recommendation platform technology, and in particular to a tourism recommendation platform based on knowledge graphs. Background Technology

[0002] Existing knowledge graph-based travel recommendation platforms typically integrate multi-source information such as scenic spots, hotels, restaurants, transportation, and activities related to a destination in a structured manner to construct a static or semi-dynamic knowledge graph. They then combine user input preferences, such as budget, travel dates, number of people, and interest tags, to perform entity retrieval and sorting, ultimately generating a list of attractions or simple itinerary suggestions. These systems generally rely on predefined entity attributes such as ticket prices, opening hours, ratings, and user historical behavior data, and achieve recommendations through collaborative filtering, graph embedding, or rule matching. Their core objective is to provide "reasonable" travel options under limited resource constraints.

[0003] However, existing technologies fail to model the differences in individual cognitive abilities within user groups in a fine-grained manner. Specifically, user profiles in existing systems typically only include superficial features such as age and gender, lacking quantitative characterization of cognitive dimensions such as text comprehension, visual / auditory perception, attention span, and abstract concept comprehension. At the same time, tourism resource entities are not labeled with metadata related to their cognitive interaction requirements. Therefore, even if the system identifies "family travel including elderly people and children," it cannot determine whether the elderly or children have the ability to understand specific exhibits, leading to a serious mismatch between recommendation results and actual experience. The direct consequence is that some members cannot participate effectively due to excessively high cognitive thresholds, which not only reduces the overall trip completion rate and satisfaction but also renders the system unsuitable for high-potential scenarios such as the silver economy and special education travel. Summary of the Invention

[0004] The purpose of this invention is to propose a knowledge graph-based travel recommendation platform that integrates individual cognitive characteristics as a key dimension into the entire process of knowledge graph construction and recommendation decision-making. Through multi-role modeling, cognitive needs mapping, and experience balance optimization, it achieves refined and inclusive itinerary planning for heterogeneous groups.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A knowledge graph-based travel recommendation platform includes: a multi-role cognitive profile construction module, which transforms user-inputted travel companion information into a structured multi-role cognitive profile; a cognitively enhanced dynamic knowledge graph construction module, which labels each travel resource entity with its specific requirements for user cognitive abilities; a multi-role demand fusion and conflict detection module, which identifies potential conflicts within the group regarding resource selection and generates structured coordination strategies; a cognitively adaptable itinerary generation and optimization module, which generates itinerary plans that meet budget, time, and space constraints and take into account the cognitive needs of each member based on the candidate travel resource set after conflict coordination; and an experience feedback and cognitive model iteration module, which collects user experience feedback after actual travel and calibrates the cognitive profile and resource cognitive attributes.

[0007] As a preferred technical solution of the present invention, the multi-role cognitive profile construction module includes: a role semantic parsing unit: the user fills in the identity description of each companion item by item on the platform input interface; a cognitive feature reasoning unit: the user calls the cognitive feature reasoning model trained on a large-scale cultural tourism user behavior dataset to predict key cognitive indicators; and a cognitive profile storage and indexing unit: the final output multi-role cognitive profile is stored in the user session context in the form of a structured object.

[0008] As a preferred technical solution of the present invention, the identity description in the role semantic parsing unit includes: role type, age range, educational background, interest tendency, and ability limitation statement.

[0009] As a preferred technical solution of the present invention, in the cognitive feature reasoning unit of the multi-role cognitive profile construction module, the cognitive indicators include: text comprehension ability: reflecting the user's efficiency in acquiring written information such as display boards and explanatory signs; visual and auditory perception ability: used to evaluate the user's reception quality of image details and audio guide content; attention maintenance ability: representing the effective participation time of the user in a single activity; abstract concept comprehension ability: measuring the user's grasp of non-concrete content.

[0010] As a preferred technical solution of the present invention, in the multi-role cognitive profile construction module, if the user provides a clear capability limitation statement, the platform will prioritize using the capability limitation statement to cover the model's default inference results and mark it as a high-confidence feature.

[0011] As a preferred technical solution of the present invention, the cognitive-enhanced dynamic knowledge graph construction module includes: a multi-source data fusion unit: the platform integrates multi-source data, extracts basic information of tourist attractions and projects, and stores all data in a graph database after cleaning, deduplication, and entity alignment; an automated cognitive attribute annotation unit: supplements each entity with cognitive-related attributes; and a dynamic update and manual verification unit: the graph adopts a daily incremental update mechanism, regularly captures the latest comments and operational announcements, automatically adjusts cognitive attribute values, and opens a verification interface for cultural and tourism institutions.

[0012] As a preferred technical solution of the present invention, the multi-role demand fusion and conflict detection module includes: an individual suitability scoring unit: traversing the candidate tourism resource set and calculating the matching degree between each resource and each traveler; a group experience balance evaluation unit: the overall acceptance of the resource is determined by the score of the least suitable member; and a conflict classification and coordination strategy generation unit: classifying conflicts according to their root causes and providing solutions for each type of conflict through a built-in strategy library.

[0013] As a preferred technical solution of the present invention, the conflict root cause classification in the conflict classification and coordination strategy generation unit includes: cognitive rejection type conflict: the cognitive ability of some members is insufficient to meet the basic interaction requirements of the tourism resource; physiological-rhythm misalignment type conflict: the differences in members' physical endurance, activity rhythm and attention duration make it difficult to take into account the whole group in the itinerary arrangement; interest split type conflict: members have fundamental differences in their preferences for tourism themes and activity types, and it is difficult to satisfy them with a single resource.

[0014] As a preferred technical solution of the present invention, the cognitive adaptability journey generation and optimization module includes: a main journey skeleton generation unit: using a problem-solving algorithm that satisfies spatiotemporal constraints to search for feasible paths in a cognitively enhanced dynamic knowledge graph; a cognitive compensation micro-experience filling unit: inserting micro-experience units into the gaps between main journeys, and the design of the micro-experience units follows the principle of alternating cognitive load; and a flexible alternative solution generation unit: generating at least one set of alternative solutions for each main journey.

[0015] As a preferred technical solution of the present invention, the experience feedback and cognitive model iteration module includes: a structured feedback collection unit: after the trip, the platform pushes a lightweight feedback questionnaire to the user; a deviation detection and model calibration unit: the system compares the feedback data with the original cognitive profile and resource attributes to identify prediction deviations; and a cold start and template optimization unit: for new users, the platform provides several typical cognitive templates.

[0016] The present invention has the following beneficial effects:

[0017] This solution systematically integrates individual user cognitive ability differences into the entire recommendation decision-making process. By constructing a multi-role cognitive profile encompassing four dimensions—text comprehension, visual and auditory perception, attention maintenance, and abstract concept comprehension—and combining this with a knowledge graph that labels the cognitive attributes of tourism resource entities, the platform can quantitatively assess the cognitive fit between resources and users. In group itinerary planning, a balanced evaluation mechanism is introduced, using the rating of the least compatible member as the basis for overall acceptance, generating an inclusive plan that takes into account the needs of all members. The itinerary generation stage adopts a combination of a main framework and cognitive compensation micro-experiences, and provides dynamically switchable flexible alternatives to enhance adaptability in actual execution. In addition, a closed-loop feedback mechanism is established, continuously calibrating the cognitive profile and resource attributes through structured collection of real user experience data, enabling online iteration of the model and optimization of the cold start template. This design allows the system to adapt to the complex needs of heterogeneous user groups, improving the feasibility and participation of itineraries in scenarios such as family travel, senior citizen travel, and special education, providing a solution for smart tourism services that combines technological depth with humanistic care. Attached Figure Description

[0018] Figure 1 This is a structural diagram of a knowledge graph-based tourism recommendation platform proposed in this invention.

[0019] Figure 2 Flowchart for building a multi-role cognitive profile module;

[0020] Figure 3 A flowchart for the construction module of a cognitively enhanced dynamic knowledge graph;

[0021] Figure 4 A flowchart for the multi-role requirement fusion and conflict detection module;

[0022] Figure 5 A flowchart for the cognitive adaptability trip generation and optimization module;

[0023] Figure 6 The flowchart for the experience feedback and cognitive model iteration module. Detailed Implementation

[0024] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0025] Please refer to the appendix. Figure 1 A knowledge graph-based travel recommendation platform, comprising the following modules:

[0026] Multi-role cognitive profile construction module: Please refer to the appendix Figure 2This module is responsible for transforming the user-input information about fellow travelers into a structured multi-role cognitive profile, providing fine-grained evidence for subsequent personalized recommendations. Traditional systems only record age or identity tags such as "elderly" or "child," while this module further infers the specific levels of each member in dimensions such as text comprehension, perception, attention maintenance, and abstract concept comprehension, forming a computable cognitive feature vector, specifically including the following units:

[0027] The semantic role parsing unit requires users to fill in the identity description of each companion on the platform's input interface, such as "Mother, 65 years old, rarely reads books," "Son, second grade, likes hands-on activities," and "Myself, 30 years old, history enthusiast." This unit first performs word segmentation and named entity recognition on the input text to extract key elements, including age values, educational stage cues such as "second grade," interest keywords such as "hands-on activities" and "history," and explicit ability declarations such as "rarely reads books" and "poor hearing." Subsequently, through a pre-trained semantic role annotation model, the above elements are classified into standard fields, including role type, age range, educational background, interest tendencies, and ability limitation declarations.

[0028] Cognitive Feature Inference Unit: This unit calls upon a cognitive feature inference model, which is trained on a large-scale cultural tourism user behavior dataset. The dataset includes multimodal signals such as guided tour dwell time, exhibition board gaze trajectory, interactive device completion rate, and questionnaire self-assessment results. The model employs a multi-task learning architecture and simultaneously predicts four cognitive indicators, specifically:

[0029] Text comprehension ability: Reflects the user's efficiency in acquiring written information such as display boards and explanatory signs. The value ranges from 0 to 1, where 0 indicates that the user is completely unable to read the text and 1 indicates that the user can fluently understand academic texts.

[0030] Visual and auditory perception ability: used to assess the quality of a user's reception of content such as image details and audio guides, based on a statistical model of age-related sensory degeneration and user self-report.

[0031] Attention span: This refers to the effective engagement time of a user in a single activity. Children typically engage for less than 20 minutes, while adults can engage for more than 45 minutes.

[0032] Abstract concept comprehension ability: measures the user's grasp of non-concrete content such as historical background and cultural symbols, and is strongly correlated with educational background and life experience.

[0033] It should be noted that if a user provides a clear statement of ability limitations, such as "Grandma is completely illiterate," the system will prioritize using this statement to override the model's default inference and mark it as a high-confidence feature. That is, the cognitive feature inference unit is based on the user's input role description and automatically infers the numerical level of the user in the four cognitive dimensions through a pre-trained model. This approach is an implicit, probabilistic estimation based on group statistical laws and is suitable for ordinary users who do not provide detailed information. Ability limitation statements are made when the user actively provides a clear description of their abilities. The system treats this as high-confidence individualized prior knowledge and directly overrides the model's default inference. This indicator is an explicit, deterministic, and individual-specific correction signal. If the statement involves a certain cognitive ability, such as "illiterate" → text comprehension ability is 0, then the system will force that indicator to be set as the declared value.

[0034] Cognitive Profile Storage and Indexing Unit: The final output multi-role cognitive profile is stored in the user session context as a structured object, containing each member's unique identifier, role type, and four cognitive indicator values. At the same time, a reverse index is established to support downstream modules in quickly filtering applicable resources based on cognitive ability thresholds.

[0035] This module is the first to introduce the multidimensional ability model from cognitive science into the field of tourism recommendation, enabling the system to distinguish users with the same superficial identity but significant differences in their internal cognitive abilities. For example, when faced with "two 70-year-olds", the system can recommend a historical document exhibition suitable for those with high text comprehension and a physical restoration theater suitable for those with low text comprehension, respectively, avoiding the experience gap caused by coarse-grained overall recommendations and significantly improving the participation and satisfaction of disadvantaged groups.

[0036] Cognitive Enhancement Dynamic Knowledge Graph Construction Module: Please refer to the appendix Figure 3 This module, building upon the traditional tourism knowledge graph, adds a cognitive interaction attribute layer. It annotates each tourism resource entity with its specific requirements for user cognitive abilities, constructing a knowledge graph with both a resource view and a cognitive demand view. The resource view describes the basic attributes of entities, while the cognitive demand view depicts their matching relationship with user cognitive abilities. This module specifically includes:

[0037] Multi-source data fusion unit: The platform integrates data from multiple sources such as online travel platforms, official websites of cultural and tourism departments, museum databases, and user review communities, and extracts basic information of entities such as scenic spots, exhibition halls, tourist attractions, and restaurants, including geographical location, opening hours, ticket prices, and service facilities. All data is cleaned, deduplicated, and entity aligned before being stored in the graph database.

[0038] Cognitive Attribute Automated Labeling Unit: This unit is responsible for supplementing each entity with cognitively relevant attributes, such as:

[0039] For museum exhibits, the system calls on the optical character recognition engine to analyze high-definition images of the exhibition panels, count the number of Chinese characters per unit area, and calculate the text density index by combining the average font size. At the same time, it processes the official audio guide through speech recognition technology, counts the proportion of exhibits it covers, and evaluates the completeness of the voice assistance. In addition, it uses a large language model to perform fine-grained sentiment and semantic analysis on user comments, identify feedback such as "the text is too small to see" and "the child cannot understand the explanation", and inversely infer the visual or abstract understanding requirements of the exhibit.

[0040] For interactive projects, the system analyzes the project instructions and operation process videos to assess the user's attention span, hand-eye coordination accuracy, and rule understanding complexity required to complete the task. For example, a number game that requires rapid screen tapping has high requirements for attention maintenance and operational precision; while a pottery experience that can be freely touched does not require text comprehension ability but has a positive effect on tactile perception.

[0041] Restaurants analyze menu images to determine font size, layout clarity, and ease of ordering. If a menu only offers a foreign language or uses artistic fonts, the system will determine that its text readability is low.

[0042] Dynamic updates and manual verification units: The map adopts a daily incremental update mechanism, regularly capturing the latest comments and operational announcements, and automatically adjusting cognitive attribute values. At the same time, the platform opens a verification interface for cultural and tourism institutions, allowing scenic area administrators to log in to the backend to correct or supplement the cognitive adaptation information of their venues, such as adding sign language guidance services, setting up large-print guidebooks, adding rest seats, etc. All modifications must be reviewed by the platform before taking effect to ensure the authority of the data.

[0043] This module transforms tourism resources from static information carriers into cognitive interactive objects, enabling the system not only to answer questions about available activities but also to determine the suitable groups and activities for each activity. During the recommendation phase, the system can proactively filter out resources with cognitive barriers exceeding the user's capabilities or prioritize recommending inclusive projects that support multiple channels, such as text, voice, and touch, thereby fundamentally reducing negative experiences caused by cognitive mismatch.

[0044] Multi-role requirement fusion and conflict detection module: Please refer to the appendix. Figure 4 This module is responsible for integrating user preferences, multi-role cognitive profiles, and cognitive enhancement maps to identify potential conflicts in resource selection within a group, and generating structured coordination strategies to ensure that the itinerary plan takes into account the core needs of all members. Specifically, it includes:

[0045] Individual Fit Scoring Unit: The system traverses the candidate resource set and calculates the matching degree between each resource and each fellow user. The matching degree is composed of two weighted parts: First, interest and preference matching degree, which is calculated by using a pre-trained text embedding model in the cultural tourism field to calculate cosine similarity based on the semantic similarity between the user's input of scenic spot characteristic preferences such as "like natural scenery" and the resource's theme tags; Second, cognitive fit degree, which is calculated based on the gap between the cognitive attributes of the resource and the user's cognitive ability. If the user's ability in a certain aspect is lower than the minimum requirement of the resource, such as the resource requiring a text comprehension ability of no less than 0.5 while the user's is only 0.3, then the score for that aspect is zero. Finally, the comprehensive score of each member for each resource is obtained.

[0046] Group Experience Balance Assessment Unit: The platform introduces the principle of minimum member experience guarantee, which means that the overall acceptance of a resource by a group is determined by the score of the least suitable member. If the minimum score is lower than the preset threshold, the resource is judged to have a high risk of conflict and should not be included in the main solution. This principle ensures that even if most members are satisfied, the basic experience of the weakest member will not be sacrificed.

[0047] Conflict Classification and Coordination Strategy Generation Unit: The system classifies conflicts based on their root causes, including:

[0048] Cognitive rejection type: Some members' cognitive abilities are insufficient to meet the basic interactive requirements of a certain tourism resource, resulting in their inability to effectively participate in or understand the content of that segment. Typical manifestations include: users with low text comprehension ability being recommended to a history museum with dense display boards and lack of audio assistance, or children with weak abstract concept comprehension ability facing a highly symbolic art exhibition. The core of this type of conflict is that the cognitive threshold of the resource is higher than the actual ability of the users, which can easily cause a gap in the experience.

[0049] Physiological-rhythm mismatch conflict: refers to the difficulty in accommodating the whole group due to differences in physical endurance, activity rhythm or attention span among members. For example, young adults may want to visit multiple attractions in a row, while the elderly need to rest frequently, or children may have short attention spans and find it difficult to adapt to long periods of static explanation. This type of conflict is due to a mismatch between physiological rhythm and itinerary intensity, which can easily lead to fatigue or dropout.

[0050] Interest-splitting conflict: refers to fundamental differences in preferences among members regarding tourism themes or types of activities, which are difficult to satisfy with a single resource. For example, one person may prefer in-depth historical and cultural experiences, while another may prefer natural scenery or entertainment projects. Or, teenagers may be enthusiastic about interactive technology installations, while older people may be more interested in traditional handicrafts. This type of conflict reflects a structural differentiation in value orientations or aesthetic tastes, and forcibly unifying arrangements may reduce overall satisfaction.

[0051] For each type of conflict, the platform's built-in strategy library provides solutions. For example, for cognitive rejection conflicts, the system recommends alternative resources, such as replacing text-intensive exhibitions with immersive live performances; for physiological-rhythm misalignment conflicts, it suggests inserting buffer nodes into the itinerary, such as arranging tea breaks or free time; for interest-splitting conflicts, it suggests adopting a group activity model, with individuals experiencing things separately in the morning and meeting up to share in the afternoon.

[0052] This module abandons the traditional approach of pursuing average group satisfaction and instead adopts the principle of prioritizing the weakest link, forcibly guaranteeing the basic experience of the most vulnerable members. In this way, when traveling with families or mixed-age groups, it effectively avoids the low mood or trip interruption caused by the inability of some members to participate, thereby improving the overall harmony and completion rate of the trip.

[0053] Cognitive Adaptability Trip Generation and Optimization Module: Please refer to the appendix. Figure 5 This module generates multi-day itinerary plans based on the candidate resource set after conflict coordination, which meet budget, time, and space constraints while taking into account the cognitive needs of each member, and supports dynamic and flexible adjustments. Specifically, it includes:

[0054] Main Path Skeleton Generation Unit: This unit utilizes a spatiotemporal constraint satisfaction problem-solving algorithm to search for feasible paths in the cognitive enhancement graph. The constraints include:

[0055] The total cost shall not exceed the user's budget;

[0056] The total daily activity time shall not exceed the weighted upper limit of each member's attention span, with the weights allocated according to the number of members;

[0057] The geographical distance between adjacent activities should not exceed the physical endurance threshold of the weakest member, such as an elderly person walking no more than 1.5 kilometers at a time.

[0058] During the path search process, the algorithm prioritizes resources with higher minimum experience scores for the group, ensuring that the main route has a good inclusive foundation.

[0059] Cognitive Compensation Micro-Experience Filling Unit: This unit inserts micro-experience units into the gaps between the main program. The design of these units follows the principle of alternating cognitive load: if the previous stage requires a high level of text comprehension, the subsequent stage will be arranged to rely on tactile or kinesthetic activities such as pottery making or plant printing; if a member's fit score is low in a certain stage, the system will insert highly fit activities before and after them to balance the emotional curve. The micro-experience library contains hundreds of lightweight activities, all of which have been labeled with cognitive attributes and can be called on demand.

[0060] Flexible alternative solution generation unit: The system generates at least one alternative solution for each main itinerary. Under the premise of keeping the core experience unchanged, the alternative solution replaces high cognitive load links with low load alternatives. For example, if the main itinerary includes a museum visit, the alternative solution can be replaced with an outdoor cultural sculpture park visit. Users can switch with one click during the itinerary execution, and the system will recalculate the subsequent arrangements in real time.

[0061] This module integrates multi-channel complementarity and load regulation theory into itinerary planning, making the recommendations not only a connection of locations, but also a carefully arranged cognitive experience. By alternating activity types and controlling the pace, it reduces fatigue caused by continuous high cognitive load, improves information absorption efficiency and emotional pleasure, and is especially suitable for groups with members who have significant differences in cognitive abilities.

[0062] Experience Feedback and Cognitive Model Iteration Module: Please refer to the appendix Figure 6 This module is responsible for collecting user feedback after their actual travel experiences. This feedback is used to calibrate cognitive profiles and resource cognitive attributes, enabling continuous learning and optimization of the system. This includes:

[0063] Structured Feedback Collection Unit: After the trip, the platform pushes a lightweight feedback questionnaire to users, asking about the members' level of understanding, participation experience, and whether there are any cognitive barriers such as not being able to see the display boards, not being able to hear the explanations, or not being able to understand the activity rules for each major part. Users can evaluate item by item and add text descriptions. The questionnaire design follows the principles of cognitive ergonomics and uses large fonts, voice broadcasts, and icons to ensure that elderly users can complete it independently.

[0064] Deviation detection and model calibration unit: The system compares the feedback data with the original cognitive profile and resource attributes to identify prediction deviations. For example, if multiple users marked as having low text comprehension ability give high scores to a certain exhibit, the system will lower the text comprehension threshold for that exhibit. If a user's actual performance is better than the profile prediction, the system records the deviation and uses it to adjust the parameters of the cognitive feature inference model. The calibration process uses an online learning algorithm to gradually optimize the model weights.

[0065] Cold Start and Template Optimization Unit: For new users, the platform provides several typical cognitive templates, such as "standard elderly users" and "school-aged children users" as initial profiles. The feedback data accumulated over a long period of time is used to optimize the template parameters to make them closer to the distribution of real people. At the same time, it discovers emerging cognitive demand patterns such as "digital divide elderly" to guide the age-friendly and child-friendly transformation of cultural and tourism resources.

[0066] This module establishes a closed-loop mechanism of "cognition-experience-feedback-learning," enabling the system to move from static rule-driven to dynamic data-driven. Over time, the platform's cognitive modeling of marginalized groups becomes increasingly accurate, and it gains insights into emerging cognitive needs patterns from massive amounts of feedback. These insights are then fed back to cultural and tourism institutions, providing data support for their age-friendly and child-friendly transformations. This extends the platform's value from simple personalized recommendations to promoting the inclusive upgrading of the entire cultural and tourism ecosystem.

[0067] Overall, through the organic synergy of the above modules, the platform can deeply integrate the dimension of individual cognitive differences into the entire chain of travel recommendations, realizing a shift from coarse-grained group tags to fine-grained cognitive adaptation. This not only solves the recommendation mismatch problem caused by existing technologies neglecting differences in cognitive abilities, but also ensures the basic sense of participation and satisfaction of every member, especially vulnerable members, in family or mixed-age group travel by prioritizing the weakest link principle. While improving user experience and increasing the completion rate of trips, it also provides solid technical support for high-potential scenarios such as the silver economy, parent-child study tours, and special education travel, possessing significant commercial value and social significance.

[0068] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A knowledge graph-based tourism recommendation platform, characterized in that, Includes the following modules: Multi-role cognitive profile building module: Transforms the user's input of fellow passengers into a structured multi-role cognitive profile, providing a basis for subsequent personalized recommendations; Cognitive Enhancement Dynamic Knowledge Graph Construction Module: Label each tourism resource entity with its specific requirements for users' cognitive abilities, obtain resource cognitive attributes, and construct a knowledge graph that coexists with resource views and cognitive demand views; Multi-role requirement fusion and conflict detection module: By integrating user preferences, multi-role cognitive profiles and cognitively enhanced dynamic knowledge graphs, it identifies potential conflicts in resource selection within a group and generates structured coordination strategies; Cognitive Adaptive Itinerary Generation and Optimization Module: Based on the candidate tourism resource set after conflict coordination, it generates itinerary plans that meet budget, time, and space constraints while taking into account the cognitive needs of each member; Experience Feedback and Cognitive Model Iteration Module: Collects user experience feedback after actual travel to calibrate cognitive profiles and resource cognitive attributes, enabling continuous learning and optimization of the platform.

2. The knowledge graph-based tourism recommendation platform according to claim 1, characterized in that, The multi-role cognitive profile construction module includes: Role semantic parsing unit: Users fill in the identity description of each fellow traveler on the platform's input interface; Cognitive Feature Inference Unit: Calls a cognitive feature inference model trained on a large-scale cultural and tourism user behavior dataset to predict key cognitive indicators; Cognitive Profile Storage and Indexing Unit: The final output multi-role cognitive profile is stored in the user session context as a structured object.

3. A knowledge graph-based tourism recommendation platform according to claim 2, characterized in that, The identity description in the role semantic parsing unit includes: role type, age range, educational background, interests, and ability limitation statement.

4. A knowledge graph-based tourism recommendation platform according to claim 3, characterized in that, In the cognitive feature reasoning unit of the multi-role cognitive profile construction module, the cognitive indicators include: Text comprehension ability: reflects the user's efficiency in acquiring written information such as display boards and explanatory signs; Visual and auditory perception capabilities: used to assess the quality of a user's reception of image details and audio guide content; Attention span: Indicates the effective duration of a user's engagement in a single activity; Abstract concept comprehension ability: measures the user's grasp of non-concrete content.

5. A knowledge graph-based tourism recommendation platform according to claim 4, characterized in that, In the multi-role cognitive profile construction module, if the user provides a clear capability limitation statement, the platform will prioritize using the capability limitation statement to override the model's default inference results and mark it as a high-confidence feature.

6. A knowledge graph-based tourism recommendation platform according to claim 1, characterized in that, The cognitive-enhanced dynamic knowledge graph construction module includes: Multi-source data fusion unit: The platform integrates multi-source data, extracts basic information about tourist attractions and projects, and stores all data in the graph database after cleaning, deduplication, and entity alignment. Cognitive attribute automated annotation unit: supplements each entity with cognitively relevant attributes; Dynamic Updates and Manual Verification Units: The graph adopts a daily incremental update mechanism, regularly captures the latest comments and operational announcements, automatically adjusts cognitive attribute values, and provides a verification interface for cultural and tourism institutions.

7. A knowledge graph-based tourism recommendation platform according to claim 1, characterized in that, The multi-role requirement fusion and conflict detection module includes: Individual suitability scoring unit: Traverse the set of candidate tourism resources and calculate the match degree between each resource and each traveler; Group experience equilibrium assessment unit: The overall acceptance of resources is determined by the ratings of the least compatible members; Conflict Classification and Coordination Strategy Generation Unit: Classifies conflicts according to their root causes, and provides solutions for each type of conflict through a built-in strategy library.

8. A knowledge graph-based tourism recommendation platform according to claim 7, characterized in that, The conflict root cause classification in the conflict classification and coordination strategy generation unit includes: Cognitive rejection conflict: Some members' cognitive abilities are insufficient to meet the basic interaction requirements of the tourism resource; Physiological-rhythm mismatch conflict: Differences in members' physical endurance, activity rhythm, and attention span make it difficult to arrange itineraries that cater to the whole group. Interest-splitting conflict: Members have fundamental differences in their preferences for travel themes and types of activities, which are difficult to satisfy with a single resource.

9. A knowledge graph-based tourism recommendation platform according to claim 1, characterized in that, The cognitive adaptability trip generation and optimization module includes: Main path skeleton generation unit: Using spatiotemporal constraint satisfaction problem-solving algorithms, feasible paths are searched in cognitively enhanced dynamic knowledge graphs; Cognitive Compensation Micro-Experience Filling Units: Micro-experience units are inserted into the gaps in the main course, and the design of the micro-experience units follows the principle of alternating cognitive load; Flexible Alternative Solution Generation Unit: Generates at least one alternative solution for each main drive.

10. A knowledge graph-based tourism recommendation platform according to claim 1, characterized in that, The experience feedback and cognitive model iteration module includes: Structured feedback collection unit: After the trip, the platform pushes a lightweight feedback questionnaire to the user; Deviation detection and model calibration unit: The system compares the feedback data with the original cognitive profile and resource attributes to identify prediction deviations; Cold Start and Template Optimization Unit: For new users, the platform provides several typical cognitive templates.

Images