Requirements-driven data engineering

In the early 1990s, the effectiveness and ef®ciency of the information systems (IS) supporting the US Department of Defense's non-combat operations was questioned. As had many organizations, the support had evolved into multiple, redundant, unintegrated, undocumented, stove-piped IS. These systems require unnecessarily large non-combat IS expenses, supporting war ®ghting efforts. Lack of integration hindered the Department from effectively providing mission support information. DOD's efforts to re-engineer the non-combat IS is one of the ®rst attempts to apply requirements-driven data engineering to a large systems environment. Its application to DOD's non-combat IS data environment has provided tangible results: (1) from the top down, an enterprise model (EM) now speci®es Department-wide requirements capable of guiding future integration and development efforts; (2) from the bottom up, non-combat IS are being signi®cantly reduced, simplifying the overall problem; and (3) data quality engineering methods, guided by the EM, are being developed and applied to the remaining IS. This success has achieved a prerequisite necessary to increase the effectiveness and ef®ciency of the systems. # 1999 Elsevier Science B.V. All rights reserved