Data management in large-scale AI systems

It is argued that the development of large-scale AI (artificial intelligence) systems, like that of all large software systems, must incorporate the principles and techniques of data management. In particular, developers of knowledge-gased expert systems, making use of heuristic knowledge, must take care that the complexity and sheer size of the knowledge bases do not render impossible system development, maintenance, and execution. Traditional database management techniques and procedures alone are considered insufficient and not always applicable to the knowledge bases used in expert systems. However, the developers of large-scale knowledge-based systems still need to adhere to the principles of these procedures, such as maintainability, clarity, portability, and efficiency, to ensure successful development. The distinct knowledge-based processing paradigm imposes additional data management requirements on system development, including the ability to trace the knowledge acquired for and used in the system