Ontronic: a General-purpose Ontology-based Meta-data Manager and Integration by Sang Su Lee Tool

The Semantic Web provides a compelling vision for a common framework that allows data to be shared, understood by machines and humans and reused across applications, enterprises, and community boundaries. But it raises many research challenges such as the availability of content, ontology development and evolution, scalability, multilinguality, visualization to reduce information overload, and stability of Semantic Web languages. To address those problems, we have been actively investigating these challenges, focusing on efficient ontology building and managing techniques, learning ontologies, and matching ontologies.

We introduce Ontronic, which that provides general functionality for the engineering, discovery, management, and presentation of ontology-based metadata. The main goal of Ontronic is to suggest an ontology modeling methodology which is capable to increase the level of semantic interoperability and to provide the high accessibility to users. To achieve this goal, Ontronic firstly supports a high-level conceptual modeling methodology based on CIOM (Classified Interrelated Object Model) which is compatible with RDF and OWL. Therefore, the generated ontologies in Ontronic can be represented as various kinds of metadata languages such as RDF, OWL, and DAML+OIL. Also, Ontronic allows cooperative multi-author to develop and to share ontologies in the web-based environment. Figure 2 illustrates the overall architecture of Ontronic.

Figure 1. The overall architecture of Ontronic

Another purpose of Ontronic is the identification of best semantic matches among the similar domain ontologies. Different terminologies can be used to describe same domain concepts, attributes or instances. Consequently, although we deal with same domain, there might be multiple domain ontologies or multiple ontologies might have overlapping domains. Thus, it is essential to align disparate ontologies for the purpose of data integration. Toward this end, we are extending Ontronic to identify ontologies mapping according to the semantic correspondences among their concepts, attributes, and instances using semantic-based wrappers.

Ontronic will allow information analysts to extend the feature space by including representational elements of their choice, together with training examples of instances of these features. To do so, they will have to be able to focus on particular areas of interest and to explore possible unexpressed relationships between information units. Since it is very difficult at this time to build a federated dynamic ontology completely automatically, the interface must support multiple perspectives to allow the analyst to assist the system in establishing cross-correlations among the information.
To make it possible, Ontronic includes prior ontology matching research. Ontronic exploits both schema-level matching and instance-level matching. These previous techniques, however, often fail to cover desirable results for the matching process. Subsequently, the necessity for other information such as external evidence beyond two disparate schemas arises. We provide the corpus ontology which includes rules for the federation in order to compensate for current matching techniques. The corpus ontology is generated by domain experts and plays a role as a dictionary in matching global ontology with relational databases. The corpus ontology combined with current matching techniques will offer better matching results. We will use pattern recognition and trend analysis techniques to analyze the data for important discoveries – both scientifically and clinically. This will in turn affect the ontology, so that it will be dynamic and can learn from its usage.

Figure 2. Research Area and Approach

Currently, we focus on ontology management for the following various domains:
• Earthquake Science Research
• Crisis Management for Homeland Security
• Neuroscience Research