Font Size: 

When building data repositories for biodiversity research, two competing requirements need to be met: On the one hand, the systems need to be built in such a way that scientists are actually willing to upload their data to the system. A - maybe the - major prerequisite to achieve this is to make data upload as easy and effortless as possible. On the other hand, in order to efficiently support scientists in answering the questions asked by biodiversity research today, systems should support data integration and synthesis. This, quite obviously, is possible only, if data in the system adheres to some standards, the same names are used to denote the same concepts etc.
One way to balance the two requirements is to allow users to upload their data with whatever schema they use locally, but to encourage them to annotate the concepts used there with concepts from an ontology. Part of the problem described above can be addressed by providing an intuitive and attractive user interface. However, such an interface alone will not be sufficient it is also necessary to reduce the underlying complexity of the task. One key aspect here is that it is difficult and cumbersome for a user to find an appropriate term to annotate something with in any given ontology, since these tend to be large and complex.
One solution to overcome complexity is splitting the ontology in several domain ontologies based on the disciplines of the contributing sub projects. But even domain ontologies can contain an immense number of concepts that might be unmanageable by a user who tries to find suitable concepts describing its data. Our idea to overcome the problem of large ontologies is based on the use of sub-ontologies. A sub-ontology contains only specific parts of the whole ontology.
The creation of a sub-ontology is based on one or more pre-defined topics which represent specific user categories (e.g. the belonging to a research discipline). A topic is mapped to a sub-ontology based on the global ontology with the intention that each mapping contains relevant parts of the ontology with regard to the corresponding topic. The result of the topic mapping process is a topic-based ontology which contains only a small part of the global ontology. By combining such topic-based ontologies we can get a sub-ontology that covers several topics but still is less complex than the global ontology. The idea of our semantic annotation approach is to present the ontology as a simple class hierarchy to the users and not to upset them with a complex and overloaded form of visualization.
In our presentation we, first, provide an overview of our user-based ontology hierarchy generation that is followed by a more detailed look into the underlying tasks: mapping topics to ontology terms by domain experts, selection of relevant topics by a user, and the generation of the user-based ontology hierarchy. We conclude with a summary and an outlook to future work.