Research On Personalized Tourism Recommendation Based On Data Mining

In today's highly developed information technology,in the face of increasingly rich species and amount of tourism resources,people can easily by tourism resources search engine,e-commerce sites,such as tourism information system for travel information,travel decisions.But with the explosive growth of tourism data,people increasingly drowned in a sea of digital,need to spend a lot of time and energy to filter information really interested in traveling.Data mining in the face of this phenomenon,how to tourism,find user preferences,and then carry on the personalized recommendation become the focuses on tourism information system construction at present.On the basis of analyzing the existing research,this paper argues that the user's travel decisions affected by their own conditions and characteristics of scenic spots comprehensive,hence put forward a way to use the user-resort space vector model for users interested in characterization methods.On the basis of the model,in this paper,based on mutual information credibility of weighted bayesian classifier to predict user ratings on to specific sites,and to the score height size and probability of attractions recommended to the user.Based on the preference recommendations based on bayesian classification,this paper studies the proposed algorithm of scenic spot recommendation using association rules.This paper collected a user visited all the attractions for tourism affairs database,the minimum support and minimum confidence and minimum degree under the condition of ascension,frequent pattern mining using FP-Growth algorithm,strong interesting association rules between sites.Combining the results of preference and the association rules found,this article will also add to the list of recommended destinations for the relevant sites.Based on the framework of Scrapy web crawler technology and web information extraction based on BeautifulSoup technology,starting from the day of cellular travel network data grabbed all the scenic spots in wuhan city,visited the attractions of user data,and the users comments on the spot data.The personal information network questionnaire was designed for users who were unable to automatically grasp the information.Based on a valid user filtering,using part of the user's data as a bayesian classification of sample data,residual user data as test data,using the recommended result accuracy and recall rate and evaluate the effectiveness of the recommendation algorithm.The experimental results show that the proposed combining bayesian classification recommended hybrid recommendation algorithm and association rules has better recommendation accuracy and coverage,can satisfy the demand for the user to the attractions.