Font Size: a A A

Research On Association Rules Mining Method Of Indoor POI Based On Moving Object Trajectory

Posted on:2021-03-24Degree:MasterType:Thesis
Country:ChinaCandidate:H E WangFull Text:PDF
GTID:2518306032966919Subject:Surveying and Mapping project
Abstract/Summary:PDF Full Text Request
In the Internet era,the rapid development of e-commerce has brought a huge impact on offline real economy industries such as large indoor malls,airport shopping areas,and railway station shopping areas.Personalized services,better perfect indoor location services,and then improve their competitiveness is an urgent problem to be solved.With the continuous development and popularization of mobile terminals and the continuous advancement of indoor positioning technologies such as radio frequency identification,ultra-wideband,Bluetooth,and Wi-Fi,the trajectory data of indoor mobile users has shown explosive growth.The trajectory data of indoor mobile users completely records the user's movement process in the indoor space,and the user's behavior and habits are also hidden in the trajectory data.Association rules mine indoor mobile user trajectory data,detect hidden indoor POI association patterns,obtain users'behavior habits,serve indoor location applications,and help offline real economy industries develop better.Therefore,this article relies on the national key research and development plan of the Ministry of Science and Technology "Indoor high-precision mapping and real-time GIS technology"(project number:2016YFB0502104),using indoor moving object trajectory data as the research object,to study indoor POI association rule mining methods,the main work is as follows:First,analyze the problems of the original trajectory data of indoor moving objects,and clean the abnormal points such as floors,time,space in the data.Secondly,a semantic method of indoor moving object trajectory is proposed.Firstly,an indoor spatio-temporal agglomeration hierarchical clustering algorithm Indoor-AGNES(Indoor AGglomerative NESting)is proposed for indoor trajectory data with time sequence and multi-floor spatio-temporal features,to identify indoor user stay points,and then analyze the characteristics of indoor space Divide to better mark the semantic information of the stay point,and convert the indoor trajectory data into the user's semantic trajectory.Third,in the process of mining indoor POI association rules in the semantic trajectory,a new index-association degree that reflects the association strength between the spatial entity sets in the association rules in the indoor environment is defined.The network embedding algorithm node2vec can comprehensively consider the potential correlation information such as space and semantics between indoor spatial entities,so as to better measure the correlation strength between POIs.On this basis,a new indoor association rule mining algorithm R-FP-growth(Relation Frequent Pattern Growth)is proposed,which greatly improves the quality of mining results and indoor application value.Fourth,this paper uses real Wi-Fi location trajectory data of a shopping mall to conduct experiments.According to the experimental results,the algorithm is verified from two aspects:quantitative index and non-quantitative index.Experimental results based on quantitative indicators show that the correlation calculation based on cosine similarity has the best calculation effect,and the accuracy rate is 87%,and it is 19%higher than the traditional algorithm FP-Growth.The analysis based on the experimental results of non-quantitative indicators shows that the results of association rule mining match the true situation.
Keywords/Search Tags:Data mining, Indoor trajectory, Network embedding, Association rules, Association degree, R-FP-growth
PDF Full Text Request
Related items