An IoT Ontology Class Recommendation Method Based on Knowledge Graph

Ontology is a formal representation of a domain using a set of concepts of the domain and how these concepts are related. Class is one of the components of an ontology for describing the concepts of the system. It is used to create, update, search or delete instances which are digital representations of physical things. With the development of the IoT (Internet of Things) technology, developers create and manage the corresponding IoT instances on IoT platform. With the user's query of a few key words, how to find the ontology classes accurately is a hard problem. IoT Ontology classes recommender system can help developers find the ontology classes that they want to use efficiently. In a general recommender system, user's historical usage records, background features and input keywords are used for making personalized recommendations. However, the newly established IoT platforms do not have a large number of user usage records to optimize recommendation results. And recommendation based on input words' semantics lacks relevance between the IoT ontology classes. This paper proposed a method for recommendation of IoT ontology classes based on knowledge graph building and semantics to introduce more auxiliary information and relationships for the recommendation. And the result shows that our proposed recommendation method can recommend more related IoT ontology classes and have better performance in results' accuracy.