Establishing and verifying authentic performances of digital objects : a framework and process for evaluating digital preservation actions

Abstract

Museums, archive and libraries are often confronted with the challenge to preserve not just analogue data, but also digital objects for the long term. While one challenge is to store the data for a long term, an even more difficultly one is to keep it accessible. Digital objects have to be rendered as close as possible to how they were originally used. But not only cultural objects, also business processes and scientific data have to be kept for future generations, either for legal purposes or to reproduce scientific experiments. Two major strategies exist to keep digital objects useable over the long term. Either a digital object is migrated from an obsolete format to a format that can be rendered using modern hardware and software, or the rendering environment of the object is preserved by creating a virtual version of the environment. Virtualizing an environment means replacing various layers in the stack of software and hardware rendering the object with different representations of this layer. For example, the hardware of a system can be replaced by a software emulator of this hardware, allowing us to render the object in a new hardware environment. Independently of the strategy used to preserve a digital object, the actual preservation action has to be evaluated. We have to determine, if the significant properties of a digital object are changed by applying the preservation action, and to what degree. Only if the influence of a preservation action on the significant properties of an object is known, a preservation planner can decide if the preservation action is valid, or if the properties are changed in a way that is not acceptable for the preservation purpose. Previous research focussed on how to compare digital objects before and after a migration action, concentrating on the properties stored in the format of the object. Frameworks for performing an evaluation of objects rendered in a virtual environment do not exist. In this thesis we create a framework that allows the comparison of renderings of a digital object in different rendering environments, called the -Preservation Action Evaluation Framework-. First, we show that even for migrated objects a comparison has to be done on the level of the object being rendered, as not only the object properties change, but, identically to an emulation strategy, layers in the view-path used to render the object change. We describe the information that has to be collected about an object and its original environment. We show the different forms of a rendered object that exist in a virtual environment, to determine when and where a digital object-s significant properties can be compared. Next, we show the steps necessary to ensure that any differences in the rendering are caused by the rendering environment and not by changed external events influencing the rendering of the object. We also show, how the framework can be used in the different steps of a preservation workflow to evaluate the rendering of an object in the plan, preserve and re-deploy phase. As automatic evaluation has to be supported by the rendering environment to some degree, we show guidelines to be considered when developing a virtual environment, including data exchange between a host and a guest system. Next, we introduce an obsolete home computer used for video games and business processes. We show the implementation of a tool that allows us to migrate data stored on audio tapes to non-obsolete formats without use of the original system. As some of this data are programs, we develop an emulator that is able to execute these programs, considering the guidelines for evaluation and data exchange. We then show how the framework is used to evaluate the rendering of different objects in the emulator in the context of a preservation workflow. We also discuss the framework-s application on two more more recent objects, a scientific process and a digital artwork. The successful evaluation of the case studies shows the validity of our framework and its implementation in a virtual environment. Finally, we discuss current and future work connected to the work shown in this thesis.