Long term preservation of digital information

Abstract

The preservation of digital data for the long term presents a variety of challenges from technical to social and organizational. The technical challenge is to ensure that the information, generated today, can survive long term changes in storage media, devices and data formats. This paper presents a novel approach to the problem. It distinguishes between archiving of data files and archiving of programs (so that their behavior may be reenacted in the future). For the archiving of a data file, the proposal consists of specifying the processing that needs to be performed on the data (as physically stored) in order to return the information to a future client (according to a logical view of the data). The process specification and the logical view definition are archived with the data. For the archiving of a program behavior, the proposal consists of saving the original executable object code together with the specification of the processing that needs to be performed for each machine instruction of the original computer (emulation). In both cases, the processing specification is based on a Universal Virtual Computer that is general, yet basic enough as to remain relevant in the future. Copyright 2001 ACM.