Journal
IEEE INTERNET OF THINGS JOURNAL
Volume 10, Issue 4, Pages 3322-3338Publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/JIOT.2022.3192259
Keywords
Semantics; Internet of Things; Interoperability; Distributed databases; Cloud computing; Servers; Scalability; Distributed representation; edge-cloud computing; entity tree embedding; Internet of Everything (IoE); Internet of Things (IoT); interoperability
Ask authors/readers for more resources
To promote IoT data and semantic interoperability, an edge-cloud framework is proposed, where the edge end handles customized data processing tasks and the cloud end deals with semantic information processing. An entity tree embedding algorithm is presented to convert IoT entities and attributes into embedding vectors in a tree-structured way, capturing both the semantic and structural information. Evaluation results show that the Whitening algorithm is the best method to compress embedding vectors and the proposed embedding algorithm achieves better clustering results compared with the original entity embeddings and the uncompressed averaging method.
Internet of Things (IoT) devices and services have become increasingly ubiquitous in recent years as they greatly facilitate our daily life. To promote IoT data and semantic interoperability, we propose an edge-cloud framework. The edge end is responsible for handling customized data processing tasks and transferring the processed data, while the cloud end deals with semantic information processing. Additionally, we present an entity tree embedding algorithm at the cloud end to convert IoT entities and attributes into embedding vectors in a tree-structured way. Consequently, entity embeddings could reflect the semantic information at both Class and Property levels, which ameliorates our previous entity embedding method, leading to better embedding results and clustering effects. Finally, the entity tree embedding algorithm and the corresponding compression algorithm are evaluated. Results indicate that the Whitening algorithm is the best method to compress embedding vectors. More importantly, the entity tree embedding algorithm captures both the semantic and structural information of entities and attributes. Additionally, the clustering experiments show that the proposed embedding algorithm achieves better clustering results compared with the original entity embeddings and the uncompressed averaging method.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available