Crash Course - Annotation NA KdK

[Dutch version]

When users of the National Archive stumble upon the image of a line of text (a line strip), either through browsing or by means of the clustering algorithms of the Scratch search engine, they might be willing to annotate that line, given a sufficient level of "Wiki-motivation".

Upon finishing the annotation of that line, at least three goals are met:

  1. from this time on, other users may find that line through keyword-based free-text search;

  2. the resulting new text annotation(s) can be used to retrain the pattern-recognition and cluster algorithmes opnieuw, thereby improving the accuracy of the results on future search actions;

  3. thirdly, the line-strip annotations are useful because they allow for a performance evaluation of the pattern-recognition algorithms. Thus, even if the annotations are not used for training the system, they are very useful!

Rules

  1. The text is labeled in reading order, from upper left to lower right.
  2. The ink trace of the target text is blackened. The irrelevant fragments on the margins of the line strip are displayed somewhat lighter and do not need to be labeled.
  3. For the first line of a paragraph, the following elements may be encountered (Dutch):
    • [month] Jan,Feb,Maart,April,Mei,Juni,Juli,Aug,Sept,Octb,Novb,Decb
    • [day]
    • [number]
    • Rappt
    • [department] FD, MD, BD, MarD of BlD
    • [day]
    • [month]
    • no (numero) ==> no
    • [number]
    • The content of the line. Please seek advice if you need help with the Dutch spelling

Example

Jan 14 19 Rappt FD 12 Jan no 49 Op adres van het Be

More instructions

  • Type what is written, not what you think you read
  • The two horizontal lines in front of "Besluit fiat" are coded as: _-_
    (so: underscore,minus,underscore)
  • People did not use a hyphen at the end of the line: please do not add it in the annotation.
  • Please only annotate (label) only the line on the scan number which was allotted to you. This demonstrator has no facilities for record locking or multiple annotators, as yet.

The interface

The interface is rather rudimentary and intended for usage by technical people rather than by the end users. Should you get lost on this site, please go to Monk.