Study On The Visualization Method Of Sensor Ontology Alignment Based On Cognitive Load

Power Internet of Things(PIo T)combines the sensor network technology with the Smart Power Grid(SPG)to realize the interconnection of all the links in the power system.Semantic Sensor Web(SSW)further introduces the Semantic Web(SW)technique into PIo T to tackle the sensor data heterogeneity problem.The emergence of SSW is helpful to realize the data interoperation,information sharing and knowledge fusion among sensor systems in the PIo T.Sensor ontology is the core technique of SSW,which defines the sensor concepts and their relationships,which works as a shared and formal reference model for sensor information exchange.However,the heterogeneity problem also exists between different sensor ontologies,which hampers the cooperation between different intelligent sensor applications.Ontology matching is able to determine the similar entity correspondences,which is regarded as an effective technique to solve the sensor ontology heterogeneity problem.But,due to the huge entity scale and semantic richness,it is a cognitively demanding task check the problematic correspondences.Therefore,how to build an effective sensor ontology visualization method to improve the efficiency of user validation becomes the critical problem for an interactive ontology matching visualization system.To address this issue,the advantages and disadvantages of existing ontology visualization methods are first investigated and analyzed,which reveals reason behind the high error rate of user validation that they neglect the human cognition rule during the information visualizing procedure.To overcome this drawback,the main research work is as follows:(1)A mathematical model is constructed to define the sensor ontology alignment visualization formally,and a cognitive load based visualization mechanism is proposed to improve the efficiency of user validation.This mechanism provides enough information for users by appearance feature model,and helps users to construct correct internal representation model efficiently by combining working memory with working memory in user cognitive theory,so as to reduce the error rate of users in checking.(2)A multi-view Sensor Ontology Alignment Visualization method(MSOAV)is further designed and implemented.In particular,MSOAV combines the indenting list,node connection and other single-view visualization methods to effectively construct the internal and external representations for the user and improve the efficiency of the validating process.(3)In the experiment,the testing datasets provided by the Ontology Alignment Evaluation Initiative(OAEI)and three actual sensor ontologies are used to test the performance of MSOAV.First,the performance of MSOAV is compared with other cutting-edge ontology visualization methods through four types of user interaction experiments.The experimental results show that MSOAV is able to help the user efficiently construct the internal representation of the problematic concept correspondences,and reduce the error rate of user validation.Then the performance of MSOAV based interactive ontology matching method is tested by collaborating MSOAV with an automated ontology matching method based on the evolutionary algorithm.The experimental results show that the interactive ontology matching method using MSOAV can significantly improve the quality of sensor ontology alignment.