| In recent years,with the continuous development of mobile Internet technology,and the popularity of intelligent devices,a new social network service which is called the location-based social network(LBSN)gradually formed.Thanks to the progress of mobile computing and wireless network technology,LBSN users are able to share information in mobile space,which results in the data covered is also growing,In addition,with the rapid development of LBSN,the related technology has been paid attention to by some researchers gradually,consequently making the field a popular researching direction.This paper mainly studies the point-of-interest recommendation system oriented to LBSN,which is based on project of degree period.Two parts of this paper mainly studies are:the incentive mechanism in the crowd sensing and the POI algorithm,but the data fusion work in the project is not the content of this paper.Main viewpoints of this research are as following:(1)We analyzed present research situation of the recommendation system in location-based social network in this paper.Besides,a few researching ideas and relative merits of several typical recommendation methods are also discussed here.All the work above provides a theoretical basis for the following research.(2)We researched about the incentive mechanism in the crowd sensing,leading to several incentive methods which improve authenticity of the sensing data uploaded by users.According to the mobile crowdsourcing and reward payment incentive methods,the BARIM algorithm has been proposed and improved.Finally,by considering all the factors above,a incentive mechanism based on reverse auction is presented in this paper,making more users participate in the POI recommendation system to improve the sparseness of user data.(3)Aiming at the phenomenon that point-of-interest recommendation effect may be influenced by temporal,social and distance,the paper deeply analyzes these three factors as well as putting forward relevant solutions.Eventually,by joining all the factors together,a multi-dimensional fusion approach for point-of-interest recommendation system is brought up here in this paper.All the datasets studied in this paper are acquired from a real LBSN website which is called Foursquare,Whrrl.And these datasets are used to compare the proposed approaches with some typical recommended methods,finally leading to the conclusion which proves the effectiveness and feasibility of the proposed approaches,either from a better precision rate or a better recall rate. |
