In the Internet of Things (IoT), all devices are supposed to communicate among themselves, worldwide. So what are they saying to each other? Recently, former Siemens CTO Siegfried Russwurm got to the core of the issue: “Industry 4.0 needs first of all semantics. We can only get through interfaces and breaking points using unified semantics.” Apparently not only civil servants in cross-border projects or industry supply chain managers need semantic interoperability. The billions and billions of IoT devices need it as well.
The Must of Semantic Technologies
Sebastian Tramp, coordinator of the Linked Enterprise Data Services (LEDS) project, nicely explains why the vision of the IoT and Industry 4.0 cannot be realized without semantics. If the meaning of IoT devices is not clear, it’s hard for them to interact or even communicate. For this, the devices and their relevant metadata must be clearly defined. If, for example, some value is supposed to be measured, the data stream needs to contain information which sensor took the value when and where. But also what this value is all about. The power of the IoT is based on combining data from different sources. To link this data in a meaningful way you need interfaces in form of shared knowledge, i.e. ontologies. That’s what semantic technologies deliver.
Human language plays a surprisingly big role in the IoT. For example, a visual sensor’s image Exif information records under [Flash mode] the value “flash, red eye, no strobe return”. Another device processing this textual metadata needs to understand what “…red eye, no strobe…” actually means. Crucially, if it can’t provide specific processing for the strobe usage, it should conclude the more generic fact that a flash was active. To make things even more complex, it might say this in Chinese or German, depending on where the device was built.
Leverage Terminologies, Taxonomies, and Ontologies
Luckily Multilingual Knowledge Systems (MKS) like Coreon deliver the required semantic and linguistic intelligence for the communication of IoT devices. Companies can leverage existing resources such as word lists, multilingual termbases, and taxonomies to build their metadata concepts with corresponding labels in one or more languages. The metadata concepts need to be semantically structured at least in broader-narrower relations. Through auto-taxonomisation, a provisional graph is suggested which is reviewed and finalised by subject matter experts. Knowledge resources often require coverage of several languages. Mono- and bilingual term extraction, and text and translation memory harvesting algorithms, reduce this effort significantly.
This way a knowledge graph is created with each node representing a metadata meaning expressed by one or more labels. When shared, this graph becomes the interface for IoT devices.
Semantics for the IoT
Without semantic interoperability, IoT devices fail to communicate with each other. If human intervention is necessary, the Internet of Things with billions of devices remains a buzzword for a great vision. Multilingual Knowledge Systems are a proven solution to make data repositories, systems, organizations, and even countries interoperable. They will provide the unified semantics for the Internet of Things, globally.