Error detection in data base systems

• Authors: Michael Hammer
• Venue: AFIPS '76 Proceedings of the June 7-10, 1976, national computer conference and exposition
• Year: 1976

Incorrect data poses a serious impediment to the effective use of computerized data bases. Conventional approaches to the design and implementation of automated data error detection systems are inadequate for large and complex data bases. Partly, this derives from the inherent intricacy of the problem, with decisions being required as to what checks to perform, how and when to do the checking, and how to respond should an error be found; writing procedures to accomplish these functions is a difficult programming task. Also at fault is the unrealistic and overly simplistic view of data correctness embodied in most contemporary systems. "Intelligent" data checking systems are required, which possess more extensive knowledge of the data base environment. They will need to understand the structure of the world which the data base models; the way the data base is used, and the relative importance of its various components; the sources of the errors that might occur and the costs of detecting them; and the patterns and rates of errors that actually do occur. Such a system would then be in a position to detect a wide range of errors, allocating its resources in a systematic fashion and responding appropriately to different error situations.