Processing Uncertainty Queries in Database Management Systems

1998 Annual FedLab Symposium, Advanced Displays & Interactive Displays Consortium |

Emerging applications, including many military applications, require explicit mechanisms to represent and process uncertainty in queries and in the data stored in databases. Most current approaches to supporting uncertainty in queries layer a reasoning component on top of existing relational database management systems. (DBMSs) which resolves the uncertainty in queries outside of the DBMS. While the layered approach is attractive due to its simplicity and since it requires minimal extensions to existing DBMS technology, it has some fundamental shortcomings which limit its usefulness to only simplistic application. This paper proposes an extended relational model together wit ha suitably extended relational algebra as an alternative mechanism to incorporating uncertainty in queries. In contrast to the layered approach, the proposed model allows uncertainty to permeate database processing overcoming many of its limitations. The paper identifies challenging research issues that we are currently addressing in developing the proposed framework.