Research On Data Processing Method Of Internet Of Things Based On Hierarchical Modeling And Multi-choice Meta-path

The emergence and rapid development of the Internet of Things(Io T)have enabled various devices and services to be connected through the network,such as web applications,smart devices,etc.The rise of communication technologies such as 5G and IPv6 has lowered the threshold for access to the Internet,which has led to a spurt of growth in the number of interconnected devices and services in the Internet of Things.Accompanying the proliferation of the Internet of Things entities is a considerable amount of Internet of Things data.The interaction between Internet of Things entities has produced many entity interaction data and the human demands that drive these interactions.To achieve more fine-grained processing and full utilization of Io T data,and further benefit various Io T data processing tasks,such as service recommendation for numerous potential Io T devices and functional combinations of services,Io T entities classification.According to the Internet of Things data source,our research divides the Internet of Things data into two levels——the interaction context of the machine level and the demand description of the human level.The combined interaction between different Io T entities' different functions generates an interaction context to meet specific functional requirements on the machine level.These basic Internet of Things applications are often described in a paragraph of natural language at the human level——the demand description of the human level.Our research aims to realize the integration of Io T data by using the mutual mapping and complementary characteristics of these two data levels,which is rarely paid to by traditional methods.However,the realization of two Io T data levels in different forms poses a challenge to our research.Simultaneously,how to effectively represent the Internet of Things data in different forms after integration has become an urgent problem for us.To this end,we designed a novel heterogeneous information network representation learning system for Io T data modeling.In this system,we use the graph structure to model the interaction context from the machine level and the demand description from the human level.Then,we view the graph structure as a bridge and use the mapping relationship between the two levels of data to integrate the two networks into Io T Heterogeneous Information Network(Io T-HIN).Finally,we apply graph representation learning technology to the Io T heterogeneous information network to obtain a vector representation after data integration.However,the heterogeneity of the Internet of Things heterogeneous information network and the diversity of semantics in the network make the general graph representation learning methods unable to use the complete information fully.Therefore,we propose a graph representation learning method based on Multi-choice Meta-Path(Mc MP)to capture richer semantic information between network nodes and learn the effective vector representation of nodes.We have conducted experiments on a real-world Io T dataset to verify our system.The experiment results proved the practicability and effectiveness of our system and provided a basis for the rationality of our research on Io T data.