Although the Workshop did not include a systematic examination of the
methods for converting texts from paper (or from facsimile images) into
machine-readable form, nevertheless, various speakers touched upon this
matter. For example, WEIBEL reported that OCLC has experimented with a
merging of multiple optical character recognition systems that will
reduce errors from an unacceptable rate of 5 characters out of every
l,000 to an unacceptable rate of 2 characters out of every l,000.
Pamela ANDRE presented an overview of NAL's Text Digitization Program and
Judith ZIDAR discussed the technical details. ZIDAR explained how NAL
purchased hardware and software capable of performing optical character
recognition (OCR) and text conversion and used its own staff to convert
texts. The process, ZIDAR said, required extensive editing and project
staff found themselves considering alternatives, including rekeying
and/or creating abstracts or summaries of texts. NAL reckoned costs at
$7 per page. By way of contrast, Ricky ERWAY explained that American
Memory had decided from the start to contract out conversion to external
service bureaus. The criteria used to select these contractors were cost
and quality of results, as opposed to methods of conversion. ERWAY noted
that historical documents or books often do not lend themselves to OCR.
Bound materials represent a special problem. In her experience, quality
control--inspecting incoming materials, counting errors in samples--posed
the most time-consuming aspect of contracting out conversion. ERWAY
reckoned American Memory's costs at $4 per page, but cautioned that fewer
cost-elements had been included than in NAL's figure.