The BIG IoT project focuses on easy integration of the IoT data and services o ered by existing IoT platforms and services based on semantic technologies. To enable IoT applications to consume data and services provided by heterogeneous systems and di erent stakeholders, a common set of ontologies and an RDF triple store with querying functionalities are used in the service discovery phase. The subsequent communication between such applications or services (as consumer) and the provider can be optimized as well. The process of automatically transforming JSON-serialized responses from an IoT data provider to linked data format is discussed and the bene ts of this automation are explored in this poster contribution.
The Internet of Things (IoT) is rapidly growing both in terms of the number of entities
(sensors, actuators, etc.) placed onto networks and the number of applications and services
which operate them, collect sensor data and o er smart solutions. The growth of the IoT
and the corresponding market demands lead to an increased complexity of interactions
among the various IoT stakeholders, including thing manufacturers, platform providers,
service and application developers. To unlock the full potential of the connected world,
they need to consider the multi-domain nature of the IoT, heterogeneity of the protocols
and data formats used, competition and/or co-operation with other vendors and new
players entering the market. Thus, interoperability de ned as \the ability of two or more
systems or components to exchange data and use information" [1: 9] remains one of the
key problems to solve in the eld of IoT [
The idea of IoT ecosystems and ad-hoc collaboration among di erent stakeholders
is gaining popularity quickly to overcome the shortcomings of existing proprietary
vertical systems created for distinct product lines. Capable of reciprocal operation with
others, elements of such cross-platform and cross-domain ecosystems must utilize some
kind of shared semantics, i.e. either follow standardized semantic models or through use of
\on-the- y" concept alignment mechanisms. Sharing vocabularies and mapping varying
concepts to selected reference ontologies have been proposed as semantic
interoperability solutions long ago (see, for example, [
The goal of the BIG IoT (\Bridging the Interoperability Gap of the Internet of Things",
see more in [
One of the core concepts in the BIG IoT is the concept of o ering representing a set of IoT resources: sensor data, information or services. Providers register their o erings on the marketplace using semantically annotated o ering descriptions which are stored into an RDF triple store. O ering descriptions are grounded on semantic models common for all triples in a store: the semantics of each o ering follows the BIG IoT Semantic Core Model (e.g. de ning the o ering type, license, price, endpoint URI, protocol, input data needed to use its functionalities and output data returned when an o ering is consumed) enriched with domain independent and domain dependent semantic annotation.
Consumers discover the needed resources through the (o ering) query specifying the properties of the o erings of interest. For each query a set of matching o erings is returned, and a consumer is able to subscribe and get access to them at run-time. The pair o ering description { o ering query provides a mechanism of data/service discovery within the plentitude of registered providers (see the ow in Figure 1). It is important to stress that o ering description also contains an explicit mapping of the output elds to the corresponding semantic URIs (e.g. \longitude" : \http://schema.org/longitude").
Semantic Web technologies therefore form the core of the BIG IoT project: modeling providers, consumers and o erings according to a prede ned RDF schema allows to conceptualize all activities needed for the collaboration among the various stakeholders of an IoT ecosystem. The meta-data coming from di erent sources is combined and stored in an RDF triple store. Sophisticated semantic queries as well as semantic matching (even in cases where participants use di erent semantic models or formats) are enabled. 3
BIG IoT API and SDK To facilitate participation in the BIG IoT ecosystem, developers can download an SDK for the BIG IoT API. The SDK encompasses libraries for both providers and consumers, which translate function calls from the respective application or service code into interactions with the marketplace, or directly between a consumer and provider. The Provider Lib allows an IoT platform or service to register o erings. The Consumer Lib in turn allows an application or service to discover available o erings based on semantic queries, and to subscribe to o erings of interest.
New Facets of Semantic Interoperability
In the presented work, we make use of semantic annotations already provided in the o ering descriptions in order to empower consumers to use this knowledge when accessing data that are not yet semantically enriched by their providers. Speci cally, we extend the Consumer Lib in order to automatically transform the obtained data and semantically annotate them using linked data frameworks. 4
Semantic matching during the IoT service discovery phase might result in a set of o erings which { when accessed { may di er, for instance, in terms of output data returned to the consumer. Many already deployed platforms or services use JSON as data format. However, di erent providers use typically di erent property names for the data o ered, which, in conjunction with the lack of semantic information, makes it impossible for a consumer application to automatically integrate the data. As a consequence, the developer of a consumer application has to take care of the integration of data from di erent providers manually and at implementation time.
Transforming JSON-serialized data obtained from an external data source into linked data has many advantages from a consumer application point of view. Firstly, it enables it to compare (rank, lter) and fuse data coming from di erent providers, and to validate and check them for consistency. This is vital for future-proof IoT applications, which will be increasingly autonomous and thus rely on run-time selection of providers.
Secondly, the resulting data harmonization can help developers to minimize the e orts
needed to integrate new data providers, and to access data from various sources in a
uniform manner. Even if a consumer does not utilize the set of BIG IoT ontologies itself, it
is easier to translate a single pre-de ned vocabulary to the local schemata than to manage
potentially con icting keys from di erent sources. As JSON-LD is 100% JSON compatible
[
Thirdly, as JSON-LD represents an instance of an RDF data model [
In conclusion, mapping varying JSON attributes to a single set of unambiguous identi ers gives each portion of data a well-de ned structure, streamlines its processing, and makes data truly interoperable. For future-proof IoT consumer applications, which rely on dynamic (run-time) service discovery and integration of new data sources and services, we believe the proposed JSON { JSON-LD transformation is the best solution, as generic linked date frameworks can be used to integrate the data in a very exible manner and with the extra bene t that JSON-LD capable data providers can also be considered without additional e orts.