Research And Application Of Spatio-temporal Association Rules Based On CAN-tree

Spatio-temporal data is a type of data that contains time attributes,spatial attributes,and other attributes.With the development of science and technology,spatio-temporal data is almost everywhere.It exists in many fields such as geography science,climate science,social science,epidemiology,transportation,and shopping mall consumption.However,spatio-temporal data has complex characteristics such as multidimensionality,dynamic and autocorrelation.Therefore,it has important theoretical significance and application value to research on spatio-temporal data mining.This thesis takes spatio-temporal data as the research object,mainly studies association rules algorithm based on spatio-temporal data,and applies it to the spatio-temporal recommendation system.The main research content of this thesis includes the following two points:1.Association rule is a common temporal data mining methods,widely used.Most of the research on association rules focuses on the multidimensional attributes of spatio-temporal data,while ignoring the dynamic growth of it.For incremental data,the existing association rules require loop iterations for each operation,resulting in a relatively low mining efficiency.In the specific scenario of spatio-temporal data,this thesis proposes a CAN-tree based spatio-temporal association rule mining algorithm(CSAR).The simulation results show that this algorithm can effectively improve the efficiency of spatio-temporal data mining under incremental problems.2.Currently,the most popular and widely used collaborative filtering algorithm still has some problems in the mining of spatio-temporal data,for instance,easily ignoring the internal many-to-many connections between items and items.Collaborative filtering algorithm with association rules can improve the recommendation accuracy of recommendation algorithm,however,the existing research directly applies traditional association rules to recommendation algorithm,whose operation efficiency is not high and the recommendation accuracy remains to be improved.Analyzing the shortcomings of the current recommendation algorithm,applying the CSAR algorithm proposed in Point 1 to the recommendation algorithm and optimizing the recommended algorithm,this thesis proposes a collaborative filtering recommendation algorithm based on CSAR(CSAR-CF),which can improves the accuracy of the recommendation algorithm and at the same time guarantees the operation efficiency of the algorithm to some extent.Finally,by comparing and analyzing the experimental simulation,the effectiveness of the recommendation algorithm is proved,and the recommendation accuracy of the recommendation algorithm on the spatio-temporal data can be improved to a certain extent,and at the same time,the operational efficiency is ensured.