Lehrgebiet Informationssysteme

In the last years, the modeling and querying facilities required by advanced applications have consolidated, and standardizations like SQL3 and ODMG have emerged. Consequently, increasing attention should be paid to improving system performance through adequate processing models and implementation technologies. KRISYS, a knowledge base management system developed in our group pursues this endeavor.

According to the requirements of advanced data processing, KRISYS is conceived for a client/server environment. While processing at the server is performed in the data model of the server DBMS, client-based processing is carried out in the KOBRA object model. The KOBRA model is comparable to other object-oriented data models as, e.g., those known from standards as the ones mentioned above. In contrast to the latter, KOBRA features not only the abstraction concepts of classification and generalization, but also those of association and aggregation. For exploiting locality of reference in client processing, the client component of KRISYS hosts an application buffer, the Working Memory, which serves to loosely couple server and client components of KRISYS.

Retrieval and modification operations are formulated in KOALA, a descriptive, set-oriented query language constituting the user interface of KRISYS. KOALA features two powerful operations, ASK to query the knowledge base, and TELL to change the state of the knowledge base. Based on the KOALA language, KRISYS additionally provides features as, e.g., rules or a powerful framework for modeling as well as monitoring semantic integrity constraints.

The client/server environment has a strong impact on query processing in KRISYS. Hence, apart from the adequate processing of a powerful object model, the topics of interest for this project also include issues like vertically distributed query processing or descriptive buffer management. Additionally, aspects of parallel query processing are at the center of interest.