Research On Location Recommendation Algorithm In Consideration Of Time And Geographical Factors

With the development of science and technology and the popularization of the Internet,users can easily and quickly share information,such as words,pictures,video and geographical location.However,with the increase of users,the social network has accumulated a huge amount of information,and it is increasingly difficult for users to choose the content they are interested in.Because each user has different preferences,they may have different opinions about the same information.Therefore,the research of the personalized recommendation algorithm gradually into the line of sight of people,find each user interest preference,and provides specific recommendations for them choose,is the key research direction in the field of data mining.Currently,recommendations for LBSN generally include the following four aspects:friend recommendation,location recommendation,social recommendation,and activity recommendation.Among them,the research on location recommendation algorithm is the focus of this direction.Location recommendation services typically dig into user check-in records,location information,and user social relationships to recommend a list of places where users are most likely to sign up in the future.There are many studies on location recommendation services,but there is not much research on the impact of time factors on the recommended location.Time factor in the position to recommend plays a very important role,because people tend to be different in different time of the day to visit the site,for example:the user has larger probability to choose to go to a restaurant at noon,at night,choose to go to the gym.Furthermore,people tend to visit nearby locations,for example,after a user may choose to walk in a nearby park after dinner.Therefore,geographical factors also play an important role in position recommendation.Conducted in-depth research in this paper,the recommendation algorithm,this paper discusses the influence of recommendation system related algorithms,and at the Point of Interest?Point of Interest,POI?,on the basis of collaborative recommendation algorithm is suggested to consider a time factor of the improved algorithm.The specific improvement method is different in solving the matrix sparse problem.After considering the time factor,the data processed is changed from the previous two dimensional data to three dimensional data,which must be accompanied by the problem of sparse matrix.The relevant scholar^[48],when solving this problem,smoothes the current time slot with the sign-in data of the adjacent time slot,but the adjacent time slot is not necessarily the most similar time slot^[49].Therefore,the top-n time slots of each time slot are calculated first,and the most similar top-n time slots are used for smoothing to solve the problem of sparse matrix.In addition,the time factor and geographical factors are integrated into the recommendation algorithm by linear weighting method.Finally,structures,Spark distributed computing cluster,the experiments conducted on real data set algorithm,and adopt scientific to evaluate standard,the experimental results show that the presented algorithm is better than the benchmark to a certain extent.