Preparing a dictionary for Kirrkirr

This page describe the steps that are necessary to make a dictionary usable inside Kirrkirr. (Kirrkirr is a dictionary visualization program for XML dictionaries, particularly aimed at indigenous languages.) The foundations for being able to do this are having a well-formed XML representation of the dictionary, at least one XSLT file that will render dictionary entries in HTML, and another XML file that explains to Kirrkirr how to find things in your dictionary XML file. Along the way, other things that you might need to do are things such as adding and linking pictures and sound files.

Here are the steps in detail:

  1. Settting up the folder and properties file
  2. Getting the dictionary into a suitable XML format
  3. Building an XSLT stylesheet
  4. Building a DictionaryInfo XML file
  5. Building the Kirrkirr indices
  6. Selecting and running your new dictionary
  7. Advanced topics (fonts, splashscreen, ...)
  8. Troubleshooting problems

Examples

Here is a first complete but tiny example of what you need. You can either download TinyDict.zip which has all the files of the tiny example dictionary, or you can browse individual files in TinyDict.

TinyDict is a bit too minimal. It doesn't even exploit all the information that is present in its tiny dictionary: for instance, you don't see any links between words, and you can't view the dictionary by semantic domains. This is because we haven't put information in the DictionaryInfo specification file that explains to Kirrkirr where to look for this information. Here is a second example, which adds these features and a few more. You can either download MiniDict.zip which has all the files of the mini example dictionary, or you can browse individual files in MiniDict. MiniDict is probably about the right starting point for building up your own dictionary, but you may also want to look at some of the other dictionaries used with Kirrkirr.

The MiniWrl dictionary included in the Kirrkirr 4.0 download doesn't include many Warlpiri words in it, but it does include the same DictionaryInfo file as used with the complete dictionary, and all of the XSL files that we provide. These files exploit a number of the more advanced features that are available (but are probably too complex to be used as a starting point).

The TinyJapanese dictionary doesn't illustrate anything beyond TinyDict, but serves as a demonstration of Unicode fonts working or not working. You will probably have to fiddle to get it displaying everything correctly.

Good luck! Feel free to pass on information about any problems.

Back to the Kirrkirr home page.


Valid HTML 4.01! http://nlp.stanford.edu/kirrkirr/dictionaries/
Christopher Manning -- <manning@cs.stanford.edu> -- Last modified: Tue Jan 27 15:29:52 PST 2004