Research On Location-sensitive Personalized Service Recommendation Model

TTSS(Transportation Travel Service System)plays an important part in ITS(Information Transportation System),which provides travelers with timely,accurate and reliable travel information to improve the intelligent level of services.Due to the unclear demands,diversified services,and inaccurate market-orientation,it is necessary to construct a location-based model for personalized service recommendation.By identifying the traveler's resident behavior,the service pattern involved in the trajectory was extracted,and then the similar traveler was determined.Futher,the service recommendation algorithm was constructed with a description of dynamic reputation.First,in order to find out the stay point,the CSP(Constraint Stay Point Recognition Algorithm),and LSP(Limited to Stay Point Clustering Algorithm)were presented for the construction of location-based travel trajectory.Then,considering a place having different service categories,the POS(Points of Service)was built to enable to mine the service-using behaviors.The measurement of travelers' similarity was also given.Finally,the dynamic reputation was introducted to discover a reliable similar recommender and make the personalized service recommendation.The Geolife,POI datasets covering the 60% region of Beijing and the regional GPS dataset in Xi'an were used in exprements.The experimental results shown that LSP can improve the recognition accuracy of the stay point,and the similarity algorithm based on travel trajectory can be superior to other tranditional algorithm.Also,the dynamic reputation model can reduce the recommendation error and improve the performance of the recommended.