The semantic Geospatial web services environment developed in SWING consists of 6 components, each of which is being developed in a separate work package. As of April 2007, first prototypes of each component have been established. Material illustrating these prototypes is available below. Ongoing work enhances the prototypes, and integrates them.

 Workpackage 1: MiMS

MiMS (Mineral resources Management System) is the Application prototype. This prototype shows dynamic discovery, composition and invocation of geospatial web services within the domain of sustainable exploitation of natural resources. MiMS is an end-user graphical interface, using the other components developed by the partners to provide the semantic functionalities.

As of now, the Graphical User Interface has been established. It has basic functionalities regarding the creation and maintenance of projects. It interfaces with components and provides advanced functionalities for the publication of results in the form of web sites. On top of that, the semantic functionalities have been added for annotating new services.

End-User key operations now available are:

• Register Web Features Services into the catalogue
• Semantically annotate Web Features Services with the domain Ontology
• Discover previously registered web services (Keyword or semantic query)
• Edit settings for publication (Layer styles, attributes)
• Publish the resulting information as a web site.

On the other side, all resources involved or required by the MiMS (Ontologies, Web Services, Publishing space...) have been made available under the http://swing.brgm.fr web site.

A movie showing the prototype and its usage is available here: MiMS Application, Flash movie.

 Workpackage 2: WSMX

WSMX is the Semantic Discovery & Execution component, which provides the basic Semantic Web Service infrastructure. It is based on the WSMX technology developed in the DIP project, adapted to the geospatial domain. The two main roles WSMX will play in the SWING environment are: semantic disvovery of Geospatial web services; and semantic execution of composed Geospatial web services.

In discovery, semantic annotations will be exploited to increase precision and recall in answering user queries searching for web services. The semantic annotation and discovery will be based on logic programming, namely WSML-flight, which is based on a light-weight logic programming paradigm (datalog with locally stratified negation), and hence provides a good trade-off between expressivity, computational efficiency, and ease of use. A first demo of semantic discovery and annotation using WSMX is shown with the WP3 Demo below.

In execution, semantic annotations can be exploited to ease the buden of the human programmer in the creation of composed Geospatial web services, when several existing web services must be combined in order to accomplish the desired functionality. Namely, semantic composition techniques can potentially be used to provide guidance to the human programmer, e.g. suggesting which web services to incorporate for a given task. On the other hand, semantic mediation techniques, in particular data mediation techniques, can potentially make it much easier to resolve the scale and format conflicts that are omnipresent in the Geospatial domain.

Semantic execution in WSMX is based on formalizing the composed web services in terms of Abstract State Machines (ASM), which provide a very powerful and flexible way of defining transition systems. Our ongoing development further develops, and adapts to the Geospatial domain, WSMX's component executing ASMs. Powerpoint slides explaining the component are available here: ASM execution in WSMX, Powerpoint presentation. A movie showing an ASM and how it is executed is available here: ASM Execution, Flash movie; ASM Execution, AVI format (note that this movie is without Audio track).

 Workpackage 3: ONTO

ONTO is the Geospatial Ontology component, which serves the ontologies used for semantic annotation, semantic discovery, and semantic execution. Recently, ONTO was equipped with a Concept Repository (CoRe) which provides maintenance functionality for ontologies. This repository provides access to a set of domain ontologies, providing a formal vocabulary for the three SWING use cases. In use case 1 information about aggregate production and consumption must be collected and processed. Use case 2 requires ontologies about constraints to opening new quarries, like natural protected zones, regions of ground water supply, and historical monuments. Based on these two, the third use case requires ontologies specifying transport networks.

Web Feature Services (WFS) are annotated with feature type ontologies, which are derived directly from the XML descriptions of the WFS. The connection between feature type ontologies and domain ontologies is made via a set of semantic mappings connecting instances of feature types and domain ontologies. The ontologies and their mappings are used for semantic discovery. The domain ontologies are also supposed to form the basis of semantic execution. Discovery for geospatial processing functionality, in form or Web Processing Services (WPS), was developed using a similar (yet more complex) approach. Operations including their in- and outputs are described by semantic annotations and axioms specify input constraints and dependencies between in and outputs.

We illustrate ONTO with a video explaining the ontologies and their development, as well as showing the prototypical semantic discovery. The video is available here: ONTO, Flash movie; ONTO, AVI format.

A video introducing the Ontology Testing Framework (OTF), an eclipse plug-in enabling unit testing in ontology development is available here: OTF Installation, AVI format; OTF Guide, AVI format.

 Workpackage 4: ANNOT

The software being developed in WP4, Visual OntoBridge (a.k.a. VOB), provides a graphical user interface that supports the user in the annotation task. On one hand, VOB provides functionality to annotate WFS schemas manually. This means that the user has the ability to browse the domain ontology, select concepts relevant for the annotation at hand, and interconnect them as appropriate. On the other hand, the user can enter a set of Google-like queries to retrieve concepts and domain-relation-range triples potentially relevant for the annotation. VOB employs text mining techniques, Page Rank-like algorithms, and consults a Web search engine to discover relevant entities. This means that VOB does not rely on (partial) string matching and goes well beyond simple heuristics such as synonymy resolution and edit distance.

VOB plays an important part in the overall SWING architecture: it is used by MiMS to annotate and discover Web feature services and compositions (WP1), by Composition Studio to discover services to be composed and annotate compositions (WP6), and by Concept Repository (a.k.a. CORE; WP3) to collect user feedback regarding the domain ontology (e.g. reports about missing concepts, incorrect conceptualizations, and so on). VOB is generally used in two contexts: as a tool for establishing annotations and as a tool for defining discovery goals. As a tool for establishing annotations, VOB turns user-created graph-like annotations into WSML formalisms and registers them in the Catalog (WP5). As a tool for defining goals, VOB similarly provides facilities to define a (partial) annotation which is then used as a discovery goal by the caller (either MiMS or the Composition Studio). In both modes, VOB tries to minimize the effort of the annotator and the time required to formulate annotations.

Demo videos / demos:

• In this video, we demonstrate the term-matching algorithm, the main building block of the module that provides recommendations with respect to a set of Google-like queries. Here you can find the corresponding commentary.
• In this demo, we demonstrate the cross-language capabilities of VOB.
• In this video, VOB is demonstrated alongside with MiMS.

 Workpackage 5: CAT

CAT is the Catalogue component, which provides a standard web service registry interface storing entries to classical geospatial and non-spatial services. In addition, it utilises the underlying components to provide semantically enhanced discovery functionality.

Our ongoing development enhances a standard Catalogue with a semantic query interface and with connections to WSMX for the semantic part of the discovery. Further, this WP develops the necessary interface between a SWING end user -- a Geospatial domain expert -- and WSMX's semantic execution (see WP2 above). Given a composed Geospatial web service, WSMX needs the parameters for its execution (for execution of the respective ASM) in WSML format, while the end user should of course be able to use the composed service like any standard OGC service. So an OGC adapter is needed, transforming OGC parameters to WSML input, and transforming WSML output to OGC results. The OGC adapter serves to be able to treat composed Geospatial services (WSMX ASMs) like and other OGC services inside the Catalogue.

As yet, we have developed such an OGC adapter for WFS services. A movie showing the adapter tool, and explaining its use in the interplay between end user and WSMX, is available here: WFS Adapter, Flash movie.

 Workpackage 6: DEV

DEV is the Development Environment component, which is based on IBM's Open Source Eclipse Development Environment. It consists of a number of plug-ins that integrates and hides the complexity of the other components. Application and service developers can use the Development Environment to discover semantic services, to semantically describe their services and to compose multiple services.

As of now, the current version of the Development Environment allows the import of ontologies, it connects to a query component for geospatial and semantic discovery. It features a version of a UML-based graphical web service composer. A movie illustrating the tool and its use is available here: DEV, AVI format.