Integrating and querying information sources in the context of large-scale ontological knowledge

点击翻译按钮获取中文摘要

Information abounds, and the introduction of commercially viable computer systems with mass storage to help manage information has prompted the development of database systems together with an underlying technology. By far the most successful are systems that are based on the relational data model. Such systems provide a clean conceptual domain specific interface to managed information. This shields application developers from the need to understand low level interfaces to existing data sources and their physical data organization and from the need to choose appropriate data structures and algorithms. Hence, for relational database systems, discovering efficient execution strategies, called plans, for data manipulation requests, called queries and updates, has been the focus of research for more than thirty years. The current practice for database system technology has proven to be very effective for information systems that correspond to business-like applications of database systems. In such cases, the conceptual interface is straightforward and the data sources are comparatively simple, for example, corresponding to files of records residing on mass-storage media in which record values have no intrinsic underlying structure. However, the application to other areas of data management has proven to be much more difficult when the conceptual interface relies on an ability to capture more sophisticated and extensive ontological knowledge of an application domain (e.g., in health informatics or in information retrieval over the semantic web), or when record values in data sources might be ordered or correspond to pointers that implement linked data structures (e.g., in data integration for main memory data sources). The objective of the proposed research plan is to broaden the scope of database technology in ways that overcome many of these limitations. In particular, the plan is to continue ongoing collaborative work on a number of research topics that can lead to new and more comprehensive approaches to the fundamental problems of translating user queries and updates to efficient plans over low level data sources.