Changes between Version 1 and Version 2 of Preparing the documents for keying

Timestamp:

Aug 20, 2009, 4:38:32 PM (15 years ago)

Author:

Klaus Thoden

Comment:

--

Preparing the documents for keying

@@ -1 +1 @@
 = Workflow =
 
+
+== Viewing the images ==
+After it has been made clear which books are to be typed, the scans of
+said book are being looked at. The following issues are important:
+
+=== Quality of the scans ===
+This issue is not necessarily part of the workflow as the library's
+digigroup normally takes care of that (the major part of books are
+scanned at the MPIWG). However, some scans might be faulty
+nevertheless. It also happens that in the process of uploading the
+files to the ftp-Server some jpg-files are broken (or in the process
+of being downloaded by Formax). This is about ten pages in a work
+order of 5.000 pages. This seems to be a problem of the jpg format.
+
+
+=== Structural organization of the book ===
+The book in question has to be examined according to the document DE
+Specs (latest version, which can be found on the versioned file
+folder). The DE Specs document should cover most of the cases that
+might occur in a book.
+
+The instructions are kept simple and cover even the very basic issues
+how a book should be typed. It is recommended that one is familiar
+with the contents of the DE Specs.
+
+It is not necessary that each page of a book is examined closely.
+After the first few pages of the main matter it should be obvious how
+the book works. The rest can be viewed via the thumbnails (if one
+views the images using digilib -- which is recommended due to the
+documentation in the wiki (see below).
+
+Looking at the thumbnails is mostly enough to spot difficult parts of
+a book. One can easily tell pages with pure text from pages containing
+other structures, e. g. tables, indexes or images.
+
+Difficulties in pages with plain text can basically only arise on the
+level of unknown or unreadable characters. 
+
+
+=== Printing, special characters ===
+Spotting difficult characters is of course an issue where the page has
+to be examined very closely. The same goes for damages in
+the book. They might occur here and there. In the DE Specs there is
+a flow chart which checks the eventualities what to do
+with an unknown character on the side of the data entry firm (check if
+in Unicode, check if in unknown characters).
+
+== Documenting in the wiki ==
+
+The wiki is the central place for storing
+information about the texts. For each work order sent to the data
+entry firms, there should be an information page. Each of these
+overview pages contains links to the images of every book in the work
+order. These pages are based on a template which documents the whole
+work flow. It contains the following elements:
+ - links to the image scans and to the overview page of the work order
+ - information which version of the DE Specs was used and if special
+   instructions were given
+ - when the images were sent and when the text was returned
+ - information about expected difficulties
+ - Questions by the data entry firm and answers given
+ - Analysis of the results
+ - Post processing
+
+=== Expected difficulties ===
+During the examination of the images (see above), difficult structures
+should be noted. This may apply to things not covered in the DE Specs
+and which have to be dealt with in the future or with things that have
+gone wrong a few times in the past. Links to the difficulties should
+be provided by using the features of the digilib tool.
+
+The digilib tool is sometimes irresponsive, lacks features
+(buttons) or is not available at all (heavy traffic?). A possible
+solution is to download the images from the server to view them
+directly. However, marking and linking things in the wiki is not
+possible then.
+
+=== Analysis of the results ===
+At some point, the data entry will be finished and a respective *.txt
+file will have returned. Accompanying that file might be an updated
+version of the unknown characters list. Said list should be examined
+as to which characters were added. 
+
+One obvious part of this analysis is of course to check how the data
+entry firm fared with the expected difficulties. This is done by
+looking up the passages in the received text file and judging how they
+were typed. In any case, the results should be written down in the
+wiki. Lateron, a decision has to be made what to do if there is an
+error in the typing. The possibilities are:
+ - correct the mistake silently
+ - ask the data entry firm to redo the part
+
+/Add something about the Special Instructions/
+