Project Support

Project Support for : Ph.D/M.E/M.Tech/M.Phil/B.E/B.Tech/MCA/Msc/Diploma

Department : Computer Science/ Information Technology/ Electronics


v Intropedia

The Semantic Web Portal working group has developed and continues to develop a research portal ontology. This research portal ontology shall enable DERI to publish and document its research activities in a Semantic Web. While the work on the portal ontology has seen considerable progress in many areas during the last months, an important part of it is still neglected: the research topics. So far research topics are seen as mere string value attributes of concepts such as publication, thereby reducing the possibilities of describing complex and networked research contents. Value attributes would soon lead to a de-normalization of research topics described, because many documents and activities relate to the same research topics, which would then be stored again and again as more or less similar string value properties of instances. In the end, besides agents, however, research topics might become one of the most important access keys to information stored in DERI’s Semantic Web Portal. Research topics would best be separated into a special domain ontology to increase re-usability in other contexts than a portal ontology application. A radical position of knowledge management would even maintain that a research topic ontology could hold itself all the information of DERI’s research activities. Instead of using such an ontology to annotate research results stored in documents defined by a portal ontology, it could itself be the store for knowledge about the research domain. 
However, as DERI does not yet seem prepared to take such a radical step forward in knowledge management, it should suffice to provide a research domain ontology to annotate the instances of the concepts of the portal ontology. The domain ontology proposed is different from the portal ontology in that the former might easily become subject of more heated debates on how concepts and instances representing theoretical ideas are to be defined and related to each other, while the latter seems to be more or less a mere reflection of the visible world expressible in general concrete concepts – such as person, organization, and activity – and their specific sub-concepts. The processes of the portal ontology creation have followed a rather classical ontological engineering method. In short, given suitable anthologies were integrated, new ontology concepts were proposed by chosen experts, and discussions were stimulated to reach sufficient consensus. Even though new concepts and proposals to improve the portal ontology will arise in future, the ontology engineering can be centralized and be taken care of by only few people. The research topic ontology would be different in this respect as most members of DERI are active in a quite fluent research process. Moreover, one often finds that many people at DERI hold different views on research topics. To express it in another way: all research fellows of DERI will become ontology engineers in the creation of a research domain ontology. Of course, it would be quite easy to just decide to leave the development of the research topic ontology to a few selected people, or to restrict its scope considerably as done in the current portal ontology. One might argue that even the portal ontology will be engineered by all members of DERI, Because instances belonging to the ontology are to be created via the portal, for example in annotations; and research topics can be defined in instances instead of concepts as well, thus enabling a core ontological engineering process and a following controlled extension and development of the research topic knowledge base on instance level. From my experience with Artificial Memory, which includes a medium-sized knowledge base of several thousand entries dealing with the Semantic Web online, it is easy to start with a term-oriented approach that sketches the basic areas of a domain. Such a concepts-and-relations ontology base might be constructed as a starting point clinging to the traditional ontological engineering process.

v The Goal:

The goal of the Semantic Portal Service is to provide connection, fusion, and analysis services on academic research information to enable scientists to effectively obtain information. In order to help scientists to access relevant information, we have developed the KISTI Reference & Academic Ontology. The ontology models research entities (e.g., persons and institutions), their accomplishments (e.g., articles), publications which indicate specific journal issues or proceedings, locations and topics. In contrast with previous research-related ontologies, our ontology connects researchers to the affiliations they were members of at the time they had their accomplishment. It also connects institutions to their locations such as countries, states and cities. The ontology schema is composed of 16 classes and 89 properties, and supports inference.

1. Semantic Web technology
The Semantic Portal Service of the Academic Research Information follows several Semantic Web standards such as RDF, RDFS, OWL and SPARQL. KISTI Reference & Academic Ontology is described in OWL (more specifically, OWL-Lite), and the request for service is represented and executed in SPARQL internally.

2. Conclusion and further work

This Semantic Portal Service shows how Semantic Web technologies can be utilized for information connection and fusion in the academic research information service sector. KISTI will continue to test the services on much larger data and formulate new connection and fusion services. Moreover, by connecting with similar external services synergy can be gained.

3. Other Works

Based on this service platform, namely Onto Frame, we are developing another advanced services, which are listed here with a link to their brief descriptions.