FUSION
FUnctionality Sharing In Open eNvironments
Heinz Nixdorf Chair for Distributed Information Systems
 

Selecting, Tailoring and Aligning Formal Ontologies for Scientific Knowledge Management

Title: Selecting, Tailoring and Aligning Formal Ontologies for Scientific Knowledge Management
Presenter(s): Erik Fäßler, Friederike Klan, Alsayed Algergawy, Udo Hahn, Birgitta König-Ries
Event: GfÖ Annual Meeting
Date: 2015-08-31 00:00
Description:
Current database management technology allows to store the evergrowing set of heterogeneous scientific data. However, due to the semantically non‐uniform way
(custom abbreviations, non‐explicit relationships between data columns etc.) in which data is stored, the retrieval of relevant data for a given search query and the coherent integration into an easily human‐understandable result remains a challenge. Our research addresses this issue by making the underlying meaning of scientific data machine‐comprehensible and thus enables more accurate and comprehensive retrieval.
We represent the meaning of data in terms of formal knowledge representations
(ontologies). The scientific framework we are working in, the SFB 1076 “AquaDiva”,
incorporates many different disciplines, such as biology, chemistry, physics and
ecology. Accordingly, already existing ontologies in these areas have to be screened
in order to construct a common knowledge base for a semantically informed management system for scientific data. We currently develop a toolsuite which interactively supports the following ontology engineering steps:
‐ Selecting a minimal subset of relevant domain ontologies from BioPortal, the largest biomedical ontology repository, by matching domain terms delivered by scientists with the knowledge occurring in the portals’ ontologies;
‐ Tailoring identified ontologies by finding ‘minimal terminology coverings’, i.e., we
prune those portions of the selected relevant ontologies which contain ‐ from the
AquaDiva perspective ‐ irrelevant knowledge; these compressed ontologies are
smaller and thus computationally ‘cheaper’ to process for formal reasoning;
‐ Aligning these ‘minimal’ relevant ontologies to a single combined ontology; this
combined ontology has to be checked for consistency (avoiding contradictory
knowledge), specification gaps (avoiding incomplete knowledge),coherency (con‐
ceptual overlaps of the tailored ontologies).