| With the rapid development of the tourism industry and the continuous improvement of people’s living standards,tourism users are increasingly pursuing high-quality and personalized tourism services.Therefore,research on tourism recommendation has attracted the attention of researchers.However,in the field of tourism route planning research,existing route planning methods usually focus on the shortest path or minimum cost as a single goal,planning the fastest or most economical travel route for users.These methods are difficult to meet the needs of tourism users for personalized and multifunctional travel routes.Therefore,in response to this phenomenon,thesis investigates a personalized scenic tourism route planning model based on urgency,and on this basis,generates predictive rating values for nearby scenic spots along the road to guide users to selectively sign in to scenic spots along the road.The specific research content is as follows:(1)In response to the problem that existing route planning methods focus solely on the shortest path or minimum cost,which makes it difficult to meet the personalized and diverse needs of tourism users,thesis designs a personalized scenic tourism route planning model based on urgency.This model first utilizes user historical check-in data,basic information data of scenic spots,and public road network data to extract information such as user preferences,relationships between scenic spots,and edge scenic score.Then,the planned path function is determined based on the urgency value,which can ensure that the travel route meets the urgency level of the user’s travel and does not delay their journey.Finally,numerical experiments were conducted on road network datasets in Xi’an and Wuhan using improved genetic algorithms based on gene replacement and gene splicing operators.The experimental results show that the algorithm proposed in this paper can not only plan routes with different functions for different users,but also has the ability to personalized route planning based on user preferences.(2)In response to the problem of poor accuracy in rating prediction caused by severe sparsity of data and inability to model multi-source data in existing rating prediction methods,thesis designs proposes a deep tourist scenic spot rating prediction model based on multi-source heterogeneous data.This model uses heterogeneous network representation learning module,matrix factorization module,and text processing module to simultaneously process users’historical check-in data,rating data,and comment data,and extract user preference vectors and attraction feature vectors respectively from them.Then,the feature fusion strategy of the fusion module is used to fuse the three types of feature vectors mentioned above,obtaining the final embedded representation of user and scenic spots features,and achieving rating prediction through a latent factor model.Finally,comparative experiments were conducted on the datasets of Xi’an,Wuhan,New York,and Tokyo,respectively.The experimental results showed that the proposed model to some extent alleviated the problem of poor prediction accuracy caused by sparse data and improved the accuracy of the prediction model.(3)In order to verify the effectiveness and usability of the scenic route planning model and the scenic spot rating prediction model mentioned above,thesis designs designs and develops a smart tourism platform based on the actual needs of tourism users.Among them,the scenic tourism route planning module and the scenic spot rating prediction module are the two core modules of the platform,which are implemented based on the scenic route planning model and the scenic spot rating prediction model,respectively.They are applications of these two models in real tourism scenarios,which can plan travel routes for users based on query conditions and display the predicted rating values of each scenic spot along the path.In summary,thesis first transforms the actual needs of tourism users into specific research questions and designs corresponding models for solution.Secondly,through a large number of experiments,it has been proven that the scenic route planning model can plan scenic routes with high scenic values and features that meet users’ preferences based on their query conditions,and the overall performance is good;The scenic spot rating prediction model has improved the accuracy of rating prediction to a certain extent.Finally,the correctness,effectiveness,and usability of the model were further verified through the smart tourism platform.Therefore,this study has certain practical application value. |
