Research On Clustering Of Heterogeneous IoT Data Based On The Meta-Path

With the development of the Internet,information resources have grown rapidly,and massive amounts of data continue to interact,forming the heterogeneous information network(HIN)with complex relationships and mutual influence.The data objects in the heterogeneous information network form complex interaction relationships through different types of edges.These different types of edges(meta-path)often contain rich semantic information and have a certain degree of interpretability.It is of great significance to use the interpretability of meta-path to discover hidden structural information in the heterogeneous information networks.A typical paradigm of current Io T data application is Trigger-Action Pattern(TAP),the TAP mode is generally achieved by the combination of multiple sensor data or network services.These Io T data have the characteristics of heterogeneity.Therefore,using the heterogeneous information networks to model and express Io T data can achieve the complete preservation of data relationships,which is of great significance for mining the deeper value information of Io T data.How to design a suitable analysis method based on heterogeneous information network to mine the data of the Internet of Things has become a work that needs to be further explored.The heterogeneous IoT data set based on the TAP model has a large amount of "Trigger-Action" structural information after continuous interaction with its underlying object.Using these structural information can express and describe the requirements of the Internet of Things.With the continuous increase in the application requirements of the Internet of Things,the number of objects in the network and the complexity of the structure are also increasing.Using different semantic information in heterogeneous information networks to design appropriate clustering algorithms can obtain more reasonable network structure information.Therefore,in this thesis,we model the Internet of Things data set as a heterogeneous information network,and propose I-Rank Clus,which is a heterogeneous Internet of Things data clustering method based on the meta-path.Compared with the traditional heterogeneous information network clustering algorithm,I-Rank Clus proposes a ranking algorithm based on the integration of multiple meta-paths according to the degree of influence of each semantic meta-path in different Io T applications.The ranking value that better reflects the distribution of different objects in the heterogeneous information network of the Internet of Things data can achieve better clustering.And better clustering can be calculated to get a more reasonable ranking,and finally the application oriented to Io T is divided into multiple reasonable clusters.In the experimental results section,the influence of selecting different meta-paths on the clustering algorithm is given,and the influence of the fused meta-path and single meta-path on the clustering results.Compared with other clustering algorithms,the I-Rank Clus algorithm can better achieve data clustering and division for Io T application scenarios.