Report Conclusion and Recommendations

From Semantic Sensor Network Incubator Group
Revision as of 00:22, 17 June 2011 by Llefort (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Conclusion and Recommendations

Directions for future work

The value of the Semantic Sensor Network Ontology developed by the group lies in its formal model, its documentation and examples, and its ability to be integrated with existing sensor web standards and external ontologies. It answers a demand for harmonisation which is now better understood, thanks to the review of existing ontologies, the identification of the four flagship use cases, and the numerous examples of applications published by the group participants and early adopters outside the group.

The ontology developed by the group can act as a reference model for both the Semantic Sensor Network and Semantic Sensor Web realms, and with its emphasis on interoperability with other ontologies, it is primarily aimed at developers of Linked Sensor Data applications or sensor applications powered by other semantic data management architectures. It is also possible to embed the SSN ontology definitions within standardised APIs between the Internet of Things and the Internet for Services, including, but not limited to, client-side APIs to access metadata information related to sensor assets. A third opportunity, reserved to developers of OGC standards willing to leverage semantic web technologies, is to take advantage of the strong ties between the Semantic Sensor Network Ontology and existing OGC Sensor Web standards to enrich them.

This suggests three main directions for future work:

  • Standardise the SSN ontology to use it in a Linked Sensor Data context
  • Standardise the SSN ontology to bridge the Internet of Things and the Internet of Services
  • Foster the adoption of the SSN ontology in the OGC community

Standardise the SSN ontology to use it in a Linked Sensor Data context: This approach faces two challenges. W3C normally refrains from standardizing vocabularies or ontologies for specific application areas unless they have foundational character (e.g., SKOS) or are an integral part of a W3C activity. Sensing is a transversal activity domain which does not fit well within the current split of the W3C Semantic Web activity into vertical applications. And the alignment of the SSN XG to DOLCE Ultra Lite triggers a dependency to a third party specification which is not endorsed by any standard development organisation and for which there is no credible alternative yet available.

Standardise the SSN ontology to bridge the Internet of Things and the Internet of Services: This approach raises two different issues. The ontology should cover other categories of “Things” and be bundled as an “Ontology and API” package following the model of what is done by the W3C Media Annotations working group. Two examples of “things” not yet modelled in the SSN ontology are actuators for applications such as industrial monitoring and control applications and smart labels (attached to physical artefacts) for logistics. Richer APIs giving access and possibly control for all types of devices are needed for new applications, such as the provision of sensor feeds from portable devices and fixed sensors to augmented reality applications.

Foster the adoption of the SSN ontology in the OGC community: There has been some preliminary work to extract linked sensor data from existing applications and to work on new sensor information model API on the basis of the SSN Ontology. As indicated in the Semantic Markup section of this report, it is possible to embed links to the SSN ontology in existing applications and to apply semantic web services standards like SAWSDL to inject these definitions prior to, or during, the triplification of the data. Some early adopters of the SSN ontology have directly used RDB2RDF transformation tools such as D2RQ to achieve the same goal. A preliminary attempt to select “the best parts of both SensorML and the SSNO [sic] in an alternative sensor information model” has also been made ([Malewski et al. 2011]). In both cases, one priority is to understand if the modelling approach used by the group, based on OWL, is compatible or not with the modelling principles currently applied by the OGC community: this is something which the ISO Technical Committee 211 has started to work on ([ISO/TC211 2009]).

Adoption of the SSN Ontology

A description of the major projects which have driven the SSN ontology development is available on the Participants page of the XG wiki. The examples section of this report also includes a short description of some of these projects.

Linked Sensor Data

The Semantic Sensor Web group at Kno.e.sis is converting raw sensor observations to RDF and linking with other datasets to build a Linked Sensor Data subset on the Linked Open Data cloud Linked Open Data Cloud. With such a framework, organizations can make large amounts of sensor data openly accessible, thus allowing greater opportunity for utilization and analysis. Several W3C members backing this group have also worked on Provenance management for sensor data, working with participants of the W3C Provenance incubator group and working group.

SPITFIRE

The EU-funded SPITFIRE project aims at integrating application-level protocols, software, development environments, and evaluation methodologies from the Web and the Internet of Things. To achieve this goal of interoperability, the SPITFIRE project works at different layers: from the low-level inter-servicing protocol and network-agnostic communication, to the semantically enhanced description of sensor data and real-world semantic entities, providing semantics-driven service composition and query facilities to end-users. Further research is expected in this area because Internet-connected objects will be an integral component of the future internet and must therefore become integrated into emerging internet service delivery models, such utility and cloud computing.

SemSorGrid4Env

The EU FP7 project SemSorGrid4Env integrates the SSN XG ontology at the core of its ontology suite ([Garcia-Castro et al. 2011]), alongside domain ontologies describing the services and datasets which are integrated with sensor observations in the end-user applications, e.g., flood defence assets, topographic objects, metocean forecast models, etc.

The SSN ontology - specifically the Observation and Upper models - are the key link used to bridge the technical infrastructure (including sensor networks such as the Channel Coastal Observatory ones) with the information concepts applied by end users. Application developers can then access data services such as sensor network observations by querying them in terms relevant to the user domain (using the stSPARQL query language).

The SemSorGrid4Env project has also developed simplified web APIs to access observations made by sensors with RESTful access to traditional formats such as OGC WFS, OGC O&M, and GeoJSON, married to a linked data interface for RDF representation. More recent developments provide a more generally applicable service, incorporating data from any source that can be applied to the Observation model, including streaming data services, and utilising the latest version of the SSN ontology. These APIs have been exercised in the development of a flood response and planning web application for Southern UK coastal regions.

CSIRO SSN TCP

The Australian CSIRO Sensors and Sensor Networks Transformational Capability Platform is a strategic initiative to create technologies and capability that will transform the process of scientific discovery. Its aim is to radically increase the availability and accessibility of empirical data about the natural world. The outcomes of its research are equipping scientists with new tools for data driven scientific discovery and improved understanding and management of the environment and our impact on it. The SSN ontology is being used in the Phenonet agricultural meteorology network for plant phenomics and other applications are planned [Taylor and Leidinger 2011].

Other applications of the SSN ontology

The XG participants and early adopters have also published many research papers describing applications and possible extensions of the SSN ontology. A selection of these papers is available as a Tagged Bibliography on the SSN XG wiki, through BibBase and through the ssn-xg-public group hosted by Mendeley.

Recommendations

The first recommendation of the group is to initiate the process for the creation of a W3C community group focused on the maintenance and extension of the SSN Ontology. The group has not discussed if the inclusion of new themes like actuation should be envisaged but has agreed that for this work to continue in the best conditions, it must target a broader community of developers and adopters with common interests. We estimate, on the basis of the list of publications we have collected, that this new community should attract support from multiple W3C member organisations, including several which were not involved in the XG. A community group is also the best formula to build bridges between W3C and other standard organisation interested by the SSN ontology. To target more granular APIs and to increase the impact of the SSN ontology in new areas like Augmented Reality, it may be important to maintain or even refine the modular structure of the ontology described in this report.

The second recommendation is to encourage W3C peers to continue to work on the generic issues raised by the decision to align the SSN ontology to an upper ontology (e.g. DOLCE Ultra Lite). Such an alignment is a key feature of the SSN ontology for developers wishing to build a set of interoperable ontologies (or foundry) around it for a specific application or science domain. One of the findings of the group is that this alignment raises specific challenges for the publication, packaging and maintenance of the SSN Ontology which are not frequently addressed by other W3C groups publishing recommendations focusing on ontologies.

The third recommendation of the group is to encourage its participants and followers to join the Provenance working group to work on sensor-specific issues. Provenance is one of the flagship use cases identified by the group because this type of data is important for users of sensor data to understand its origin and the conditions in which it has been produced so that they can assess its quality. For governmental or scientific organisations operating sensor assets, provenance data and the data citations which can be derived from it will provide a critical incentive to encourage the sharing of data.

The fourth recommendation of the group is to request the W3C to promote uptake of the SSN ontology in the W3C community, especially through adoption by other W3C activities and also application by member organisations that produce relevant device or software products.

Finally, this incubator activity has allowed W3C members interested in semantic web, sensor web and geospatial standards to explore some issues which are mostly outside the core areas of competence of W3C and OGC. The final recommendation of this report is to encourage W3C and OGC to coordinate their activities in this area and especially to build a larger pool of experts by targeting organisations which are already or could be interested in becoming dual members.