Research Of Personalized Recommendation Algorithm For Ancient Village Tourism

With the economic improvement,travel becomes an important part of life.It enriches people's life and stimulates economic growth by creating jobs.Travel information online enables people to identify problems with information overload.Therefore,the recommendation of individual tourism is the growing trend of future tourism website and Apps.In China,there are many ancient villages with unique region cultures or records of China's transformations.However,villages are dying out in the process of modernization.The protection is becoming urgent.It's important to enable more and more tourists get to know and travel among interesting villages.Based on the collaborative filtering algorithm,this paper studies ancient-village-tourism oriented recommendation model primarily in three perspectives:Firstly,The propose of collaborative filtering algorithm on similarities.To reduce the data sparsity's influence on Item-Based collaborative filtering algorithm,this paper adopts User-Item matrix on different villages to build a reliable relation modeling.It improves recommendation quality and accuracy through delivering similarities among villages,improving calculation.Secondly,The integration of Latent Factor Model(LFM)with time weight.Generally,LMF is one of matrix decomposition models to solve data sparsity.But it ignores users' changing preferences over time.The paper has solved this problem by minimizing Root Mean Square Error(RMSE)to the model.Thirdly,The hybrid of model recommendations.Due to inaccurate or beyond calculation caused by data sparsity in collaborative filtering algorithm,similarity relation transfer is followed by ancient-village-feature vector.Through LFM,the vector modifies the similarity.Information lost in time-weight LFM is solved by integration of collaborative filtering algorithm.Yet,too many parameters lead to issues like occupying a large amount of memory and overfitting.The problem needs to be solved by to optimizing the model and iterating constantly.The paper is intended to offer practical theories on ancient-village-tourism recommendations.On the issues,like data sparsity in User-Item matrix and tourists' changing preferences,it studies on calculations and proves its availability with the use of experiments.