- WDI-Lab: Web Data Integration
- Business Intelligence with Integrated Instance Graphs
- LOD Link Discovery
- Mashup data integration
- Object Matching
- Schema and Ontology Matching
- Evolution of Ontologies and Mappings
- Frühere Projekte
Generic Ontology Matching and Mapping Management (GOMMA)
GOMMA - top results at OAEI competition
The Ontology Alignment Evaluation Initiative (OAEI) yearly compares ontology and schema matching systems w.r.t. their efficiency and effectiveness. The evaluation covers different domains such as matching bibliographic or life science ontologies.
In OAEI 2012 GOMMA achieves the best results for the Anatomy and Library track. Overall GOMMA produced mappings of very high quality with good runtimes for all 6 SEALS tracks. For details, please check the OAEI results page and our paper GOMMA results for OAEI 2012.
The OAEI 2011.5 evaluation results appeared in April 2012. GOMMA is among the Top2 systems for matching anatomy and very large life science ontologies. For matching large biomedical ontologies (results), GOMMA achieves the best matching quality (F-measure >92%) for the “FMA-NCI small overlapping” subtask. The organizers of the track conclude that the Top2-Systems GOMMA and LogMap “are a bit ahead” w.r.t. the other participating systems (F-Measure <76%). Both systems also achieve very good execution times, even for matching full FMA against full NCIT each comprising more than 60,000 concepts. For the anatomy track, GOMMA is the best system w.r.t. F-Measure (91.2%) among all participating systems in 2011.5. In previous campaigns only Agreement Maker achieved a better F-Measure of 91.7%. Note that GOMMA was able to produce this result in less than one minute while Agreement Maker (as most other systems) run more than 10 minutes (see Runtimes and Scalability at result page).
Ontologies are increasingly used to structure and semantically describe entities of domains, such as genes and proteins in life sciences. Their increasing size and the high frequency of updates resulting in a large set of ontology versions necessitates efficient management and analysis of this data.
GOMMA is a generic infrastructure for managing and analyzing life science ontologies and their evolution. The component-based infrastructure utilizes a generic repository to uniformly and efficiently manage many versions of ontologies and different kinds of mappings. Different functional components focus on matching life science ontologies, detecting and analyzing evolutionary changes and patterns in these ontologies. These components are used by several analysis and online applications, such as for ontology matching and the Ontology Evolution Explorer (OnEX). The figure below shows the system architecture of the component-based infrastructure.
This infrastructure enables scalable applications and analyses based on a generic data management of heterogeneous ontologies, entity sources and mappings. The modular design of the infrastructure components allows for a flexible extensibility of GOMMA’s functionalities to meet changed requirements and to realize further applications in future.
You can get access to GOMMA for non-commercial use. Please write an email containing detailed information on the name of your institution, the purpose of using GOMMA (Non-commercial research project, Commercial use, Other, …) to Anika Groß .
IMPORTANT: USE THE PROTOTYPE ON YOUR OWN RISK. THERE ARE NO GUARANTEES AND NO SUPPORT SERVICES. DISTRIBUTION OF THE SOFTWARE ONLY FOR EDUCATIONAL, RESEARCH AND NON-COMMERCIAL PURPOSES. IF YOU INTEND TO USE (RESULTS OF) THE PROTOTYPE FOR ANY KIND OF PUBLICATION YOU HAVE TO
- INFORM US IN ADVANCE (write an email including detailed information to Anika Groß)
- CITE OUR GOMMA PUBLICATIONS
The complete system is programmed in Java. After getting access, you can test the software using the GOMMA package and a MySQL database as backend. You will find instructions in the read-me as well as helping comments in an exemplary application. We welcome your feedback, but please have in mind, that GOMMA is an academic research prototype.