CN-122019879-A - Personalized travel route planning method based on knowledge graph

CN122019879ACN 122019879 ACN122019879 ACN 122019879ACN-122019879-A

Abstract

The invention belongs to the field of travel management and artificial intelligence, and provides a personalized travel route planning method based on a knowledge graph, which comprises the steps of constructing a scenic spot knowledge graph; the method comprises the steps of calculating the grasping degree of a user on scenic spots, giving grasping degree attributes to nodes of a scenic spot knowledge graph based on the grasping degree of the user on the scenic spots, carrying out breadth-first search on nodes to be recommended from a first node with each grasping degree larger than a first preset value to form a second scenic spot knowledge graph, determining the current position of the user on the second scenic spot knowledge graph, carrying out breadth-first search on the nodes to be recommended based on the current position of the user to obtain a first node to be recommended, searching on the nodes to be recommended from the first node to be recommended, and determining a plurality of nodes to be recommended based on user planning time and cost constraints to form a recommended route.

Inventors

ZHOU XIANGBING
LI XI
Ran Xiaojuan
ZHANG CHENXI
LI XIAOFENG
DU SIYUAN
HUANG LIYAO
XUE DONG

Assignees

四川旅游学院
澳门城市大学

Dates

Publication Date: 20260512
Application Date: 20260202

Claims (10)

1. A personalized travel route planning method based on a knowledge graph, which is characterized by comprising the following steps: s1, constructing a scenic spot knowledge graph, wherein the scenic spot knowledge graph comprises nodes and relationship edges, the nodes are scenic spots, the nodes are used for representing various accessible travel destinations, each scenic spot node comprises attribute information, and the relationship edges are used for representing different types of connection relations among the scenic spots; s2, calculating the grasping degree of the user on the scenic spots, wherein the grasping degree is calculated based on historical access records, search behaviors, social interactions and scoring feedback information of the user; s3, giving a mastery degree attribute to the nodes of the scenic spot knowledge graph based on the mastery degree of the user on the scenic spot; S4, performing breadth-first search from each first node with grasping degree larger than a first preset value, marking nodes with grasping degree smaller than a second preset value and distance from the first node smaller than a third preset value and same as the first node as nodes to be recommended, and forming a second scenic spot knowledge graph; S5, determining the current position of the user on the second scenic spot knowledge graph, and performing breadth-first search on the nodes to be recommended based on the current position of the user to obtain a first node to be recommended; And S6, searching the nodes to be recommended from the first node to be recommended, and determining a plurality of nodes to be recommended based on the user planning time and the cost constraint to form a recommended route.
2. The personalized travel route planning method according to claim 1, wherein the attribute information comprises scenic spot type, geographical location, average cost, play time, open time, and tourist traffic.
3. The personalized travel route planning method according to claim 1, wherein the connection relationship comprises traffic connection relationship, geographic proximity, tourist behavior association, topic similarity, and recommended route.
4. The personalized travel route planning method based on knowledge graph according to claim 1, wherein the calculation method of the grasping degree is: Score_mastery(U, D) = weight_3 * Score_visit + weight_4 * Score_search + weight_5 * Score_social + weight_6 * Score_rating Wherein: score_ mastery (U, D) indicates the user's U mastery of the attraction D; Score_visit is the historical access Score of the user; Score_search is the search behavior Score of the user for the sight; score_association is a Score for the user's interactions on the social platform; Score_rating is the scoring feedback of the user for the sight; weight_3, weight_4, weight_5, weight_6 are the weights of the factors.
5. The knowledge-based personalized travel route planning method according to claim 1, wherein the mastery degree attribute is expressed as: Mastery _level (U, D) which indicates the grasping degree of the scenic spot D by the user U; The value range is [0,1], wherein: 0. indicating that the user is completely unfamiliar with the scenic spot; 1. indicating that the user is fully familiar with the attraction; The higher the value, the deeper the user's knowledge of the attraction.
6. The personalized travel route planning method according to claim 1, wherein the determining manner of the current position of the user comprises any one of the following steps: based on real-time GPS positioning; Based on a departure point set by a user; The record is accessed based on the history of the user.
7. The knowledge-graph-based personalized travel route planning method according to claim 1, wherein the performing breadth-first search on the nodes to be recommended based on the current position of the user to obtain a first node to be recommended comprises: In breadth-first search, all candidate nodes D_ recommend meeting the search conditions calculate a comprehensive Score score_ recommend, and finally the node with the highest Score is selected as a first node to be recommended, and the calculation formula is as follows: Score_recommend = alpha * Distance_factor + beta * Accessibility_score Wherein: Calculating the geographic Distance between the D_ recommend and the current position of the user, wherein the score is higher when the Distance is closer; accessibility _score, calculating whether the D_ recommend has an efficient traffic mode reachable or not based on the knowledge graph, and giving priority to less transfer; Alpha and beta are weight coefficients; and determining the candidate node with the highest score as the first node to be recommended.
8. The knowledge-based personalized travel route planning method according to claim 1, wherein the user planning time and cost constraint based comprises: Calculating the expected stay time of the user at each scenic spot according to the time constraint, and considering the travel time of different traffic modes to ensure that the total time does not exceed a user plan; cost constraint, calculating scenic spot cost and traffic cost, and ensuring that the total cost does not exceed a user plan.
9. The knowledge-graph-based personalized travel route planning method according to claim 1, wherein the searching for the node to be recommended from the first node to be recommended uses breadth-first search or a search algorithm.
10. The knowledge-graph-based personalized travel route planning method according to claim 1, wherein after the nodes to be recommended are screened, an optimal travel sequence is calculated based on a travel business problem optimization algorithm, and a shortest path is ensured.

Description

Personalized travel route planning method based on knowledge graph Technical Field The invention belongs to the field of travel management, and particularly relates to a personalized travel route planning method based on a knowledge graph. Background In recent years, with rapid development of the travel industry and wide application of information technology, personalized travel recommendation systems become an important means for improving user experience. The existing travel recommendation methods mainly depend on rule-based recommendation, collaborative filtering recommendation and deep learning recommendation, so that the accuracy of recommendation is improved to a certain extent, and the following defects still exist: traditional recommendation methods mainly recommend according to historical access records or popular scenic spots, but the methods can not accurately measure the grasping degree of a user on a certain scenic spot, namely whether the user only browses related information or has deep knowledge and does not need recommendation. Existing deep learning recommendation methods (such as neural network recommendation model and reinforcement learning recommendation model) generally require a large amount of user behavior data to train, including browsing records, scoring feedback, search behaviors, social interactions and the like of users. However, in the context of travel recommendation, the access data of users are sparse, and the travel behavior of each user is highly personalized, so that the training data is difficult to cover all personalized requirements, and the generalization capability of the recommendation system is further affected. On the other hand, the deep learning recommendation model has higher calculation complexity, and relates to large-scale matrix calculation, vector embedding and multi-layer neural network, so that the system faces calculation resource bottleneck when processing real-time personalized recommendation. Because the existing recommendation algorithm is mostly based on collaborative filtering or similar user behaviors, the problem of repeated recommendation easily occurs, namely the system recommends scenic spots which have been removed to the user for many times. Disclosure of Invention In order to solve the problems in the prior art, the invention provides a personalized travel route planning method based on a knowledge graph, which comprises the following steps: s1, constructing a scenic spot knowledge graph, wherein the scenic spot knowledge graph comprises nodes and relationship edges, the nodes are scenic spots, the nodes are used for representing various accessible travel destinations, each scenic spot node comprises attribute information, and the relationship edges are used for representing different types of connection relations among the scenic spots; s2, calculating the grasping degree of the user on the scenic spots, wherein the grasping degree is calculated based on historical access records, search behaviors, social interactions and scoring feedback information of the user; s3, giving a mastery degree attribute to the nodes of the scenic spot knowledge graph based on the mastery degree of the user on the scenic spot; S4, performing breadth-first search from each first node with grasping degree larger than a first preset value, marking nodes with grasping degree smaller than a second preset value and distance from the first node smaller than a third preset value and same as the first node as nodes to be recommended, and forming a second scenic spot knowledge graph; S5, determining the current position of the user on the second scenic spot knowledge graph, and performing breadth-first search on the nodes to be recommended based on the current position of the user to obtain a first node to be recommended; And S6, searching the nodes to be recommended from the first node to be recommended, and determining a plurality of nodes to be recommended based on the user planning time and the cost constraint to form a recommended route. Further, the attribute information includes scenic spot type, geographical location, average cost, play time, open time, and tourist traffic. Further, the connection relationship comprises traffic connection relationship, geographic proximity, tourist behavior association, theme similarity and recommended path. Further, the method for calculating the mastery degree comprises the following steps: Score_mastery(U, D) = weight_3 * Score_visit + weight_4 * Score_search + weight_5 * Score_social + weight_6 * Score_rating Wherein: score_ mastery (U, D) indicates the user's U mastery of the attraction D; Score_visit is the historical access Score of the user; Score_search is the search behavior Score of the user for the sight; score_association is a Score for the user's interactions on the social platform; Score_rating is the scoring feedback of the user for the sight; weight_3, weight_4, weight_5, weight_6 are the weights of the factor