Semantic Modeling and Transformation

Semantic modeling represents an important step towards the platform-independent description of geospatial data. In this way, the structure and the semantics of the geospatial data can be defined independently from specific data formats. However, to use heterogeneous geospatial data jointly within the same application, often a semantic transformation is required. During the transformation process the geospatial data are restructured such that they correspond to the structure and the semantics of a different data model.

In the context of various research projects the Chair of Geoinformatics focuses in particular on the subject of model-driven transformation of geospatial data which takes place at the level of the data models.

Description

Geospatial data represent simplified copies of the real world which have been created for a specific area of application. A semantic model of the real world (Universe of Discourse) is obtained by abstracting a certain view of reality which mostly is influenced by the particular area of application.

Semantic modeling of geospatial data can be carried out in two different ways: at the level of the data transfer format or at the conceptual level which is independent from specific data transfer formats. Conceptual modeling languages are employed for creating models at the conceptual level. In the geospatial domain the modeling language UML (Unified Modeling Language) is very common; it was e.g. used for defining the German AAA Reference Model or the INSPIRE Data Specifications. A model which is created using a formal modeling language is also referred to as a conceptual schema.

In particular in cross-border applications the scenario may arise that although the geospatial data of the individual countries describe a part of the same region, the view of this region differs and, thus, also the information which is provided by the models of the individual countries. For this reason, it is often necessary to transform the geospatial data into a common destination schema, i.e., the source data have to be restructured such that they correspond to the meaning (semantics) and the structure of the destination schema. This includes e.g. modifying attribute names or combining the values of several source attributes to form one destination attribute. This kind of restructuring is referred to as semantic transformation; thereby it is important that the semantics of the data is not falsified during restructuring.

Semantic transformation can be carried out in two different ways as well, format-driven and model-driven. Format-driven means that the geospatial data is transformed directly from the source data transfer format into the destination data transfer format. The rules which define how the transformation is to be executed are defined between specific formats such as Shape or GML. In contrast, using the model-driven approach the transformation is defined at the level of the conceptual schemas. First, transformation rules are defined between one or several source schemas and one destination schema; afterwards, the data are automatically mapped from the source structure into the destination structure based on these rules. The advantage of this approach is that the transformation rules only need to be defined once, due to the use of encoding rules the transformation can be carried out into any transfer format. The encoding rules define how a schema is mapped to the transfer format.

Projects

• Model-driven approach for accessing distributed spatial data using Web Services – demonstrated for cross-border GIS applications (mdWFS) [mehr...]
• Prototypical transformation of spatial data to INSPIRE in the cross-border Lake Constance region [mehr...]

Contact person

Dipl.-Inf. Tatjana Kutzner