Research On Behavior Analysis Of Objects Based On Spatio-temporal Data

With the advancement of Internet technology,the information about different types of users generated by various Internet applications in the era of big data has accumulated a lot.It is an urgent problem to extract the deep information of users from the massive data effectively and quickly to provide more personalized and intelligent service.In the field of spatio-temporal data research,the efficient mining of massive data and the realization of data-to-knowledge transformation are the key to getting out of the current predicament.Therefore,this paper starts with the massive spatio-temporal data generated by users,conducts in-depth analysis of user behavior,and explores the application of data mining in this direction.Based on the above considerations,this paper studies the following contents.First,beginning with the spatio-temporal data,we analyze the connotation of various types of data,and propose the metadata standard for establishing the target behavior model.Besides,the target behavior model is initially constructed.Secondly,mining the semantic patterns of moving objects based on spatiotemporal data is studied.As the range filling of some elements in the target behavior model,this part combines the information of other dimensions to clarify the target intent,and uses the method of sequential pattern mining to mine and analyze the trajectory with the target intent information.Thirdly,the research on the trend prediction of moving targets based on spatiotemporal semantics is carried out.This part is the application of the target behavior model.By understanding the connotation of the prediction problem,the spatio-temporal data prediction is raised to the trend prediction level.A prediction model based on deep learning is proposed to predict the target trend.Compared with various models on different performance metrics,the model performs well in solving the problem of the trend prediction.