POI Recommendation Based On Attraction Transfer Probability Model

With the development of China's economy and the affluence of people's lives,tourism has become an important way for people to entertain.In the face of the booming tourism industry,there are a large number of travel websites.The travel websites tend to be open to the public and lack of personalized services.Therefore,tourists have to spend a lot of time and energy collecting information and making their own travel plans before travelling,which is difficult to meet the personalized travel needs of tourists.which is difficult to meet the needs of tourists for personalized travel.The emergence of the recommendation system effectively solves this problem,and the user obtains a better service experience while avoiding spending a lot of time retrieving information.However,the traditional recommendation algorithm has some shortcomings,such as the problem that the recommendation accuracy rate decreases in the case of data sparseness.Some personalized recommendation algorithms consider the geographical location and time factors,but the recommendation effect is not ideal.Aiming at these issues,this article has conducted in-depth analysis and made improvements.The main work is as follows:(1)Firstly,the scenic spots are pre-processed to establish the word bag model of the comment,and then the characteristics of the scenic spots are acquired using the Latent Dirichlet Allocation(LDA),and the user portraits are constructed with the characteristics of the scenic spots;(2)Proposing the point transition probability model.According to the tourist routes obtained from Ctrip.com,the database of scenic spots is established,and the attraction transition probability model is obtained by combining user interest and the current time.The model gives the probability that the user selects an attraction at a certain moment;(3)According to the attraction transition probability model,travel route generate based on Attraction transfer probability model algorithm is used to recommend the travel route for the user under the constraint of the planned travel time of the user;(4)Combining the algorithm model proposed above with Android development technology,a novel attraction recommendation system APP client is realized.Experiments show that the algorithm can be used for POI recommendation for users to meet the individual needs of users.