Description (en)
As mechanisms emerge to certify the trustworthiness of
digital preservation repositories, no systematic efforts
have been devoted to assessing the quality and
usefulness of the preserved content itself. With generous
support from the Andrew W. Mellon Foundation, the
University of Michigan’s School of Information, in close
collaboration with the University of Michigan Library
and HathiTrust, is developing new methods to measure
the visual and textual qualities of books from university
libraries digitized by Google, Internet Archive, and
others and then deposited for preservation. This paper
describes a new approach to measuring quality in largescale
digitization; namely, the absence of error relative
to the expected uses of the deposited content. The paper
specifies the design of a research project to develop and
test statistically valid methods of measuring error. The
design includes a model of understanding and recording
errors observed through manual inspection of sample
volumes, and strategies to validate the outcomes of the
research through open evaluation by stakeholders and
users. The research project will utilize content deposited
in HathiTrust – a large-scale digital preservation
repository that presently contains over five million
digitized volumes – to develop broadly applicable
quality assessment strategies for preservation
repositories.