Book People Archive

Re: Personal politics on this list; real issues that should be addressed

From: "Klaus Graf"
Sent: Friday, January 13, 2006 1:27 PM
> From an European/German point of view I fully agree. There are some
> very simple basics but where is the project which fulfills all?
> Funding organizations or taxpayers have to pay a lot of library people
> travelling around the world to visit "best digitization practices"
> workshops but I cannot see any real success.

Our facsimile server is very near fulfilling all of your points below
Klaus, although it is not the result of traveling around the world
in search of best digitization practices.

> Here are some simple points:
>
> Make high-quality scans!

Currently the standard page images are 300dpi full color although much
of the earlier content was 300dpi grayscale unless the original page
contained color information. We also offer higher quality wavelet only
compressed files for content that is rich in detailed images along with
the more highly compressed segmented version of the work.  For example:

http://fax.libs.uga.edu/F210xE341/
http://fax.libs.uga.edu/T427xV5/
http://fax.libs.uga.edu/E476x7xB24/
http://fax.libs.uga.edu/suff/
http://fax.libs.uga.edu/hmaps/

>
> Make it easy to download books for offline-use!
>

With a few exceptions, all of the works on our server can be downloaded
as a single layered PDF file, a single bundled DjVu file, or a single
uncorrected OCR text file from a browsable directory within the title's
directory.  This single file directory is always named "1f " for "one
file" and can be found via hyperlink on a title's option page or by
simply adding 1f to the title's directory, like this:

http://fax.libs.uga.edu/QD181xR1xS679/1f/

> Make an OCR text available for full text searching!
>

All of the titles are searchable with Boolean and proximity operators
either simultaneously or individually.

http://fax.libs.uga.edu/common/query.asp
http://fax.libs.uga.edu/common/qQD181xR1xS679.asp

Also the open source desktop viewer, WinDjView (~500K) has a very good
in-context search feature.

http://fax.libs.uga.edu/viewers/

> Give each image a simple URL as persistent identifyer (e.g. PURL, URN)
> in order to make it easy to cite a single page of a digitized book.

All works can be accessed via a persistent URL--for example:

http://purl.galileo.usg.edu/ugafax/QD181xR1xS679

This PURL string can also be used to target a single page-file in two
ways: either through the works index file so as to retain the ability
to page forward and back, or as a discrete file.  For example if you
go to the Search All option:

http://fax.libs.uga.edu/common/query.asp

and search for "single page" you will find one hit:

http://fax.libs.uga.edu/hd2951xc776/co27/co27067.djvu

This can also be fetched as:

http://purl.galileo.usg.edu/ugafax/hd2951xc776/co27/co27067.djvu

These simple single-file examples can be opened by a free graphics
program like IrfanView, an open source reader like WinDjView, or with
the standard DjVu browser plugin.  A more flexible way from the
standpoint of navigating the source title is to utilize the index
information like this:

http://fax.libs.uga.edu/hd2951xc776/co27/index.djvu?djvuopts&zoom=100&page=co27067.djvu

Which can also be expressed as:

http://purl.galileo.usg.edu/ugafax/hd2951xc776/co27/index.djvu?djvuopts&page=67

But this option will only work with a DjVu plugin equipped browser.

> Give appropriate meta-data including MARC-format (for library
> catalogs) and OAI-PMH (for OAI-harvester like OAIster)!

The Option Page for each work contains the MARC record, if available,
for that particular item.

bobk