Version 2.0 (Released 30/06/2006)
=================================
This version is an almost complete rewrite of the original code. While
the old version worked I got a lot of feedback from one kind user showing
me places where it failed miserably or didn't do as expected. These
problems led to drastic rewrites and hence this version. The major new
features include: the option to leave element annotations in the document
when they are also part of a compound, a uri feature, and a mapping between
element names and symbols (the kind feature tells you which was actually
found and so the other is inferred and should not necessarily be relied
upon to be accurate).

Version 1.2 (Released 24/11/2005)
=================================
Now loads resources from the plugin directory rather than from inside the JAR
file.  Also now explicitly releases references to the current document after
processing, which should help when working with a corpus in a datastore.

Version 1.1 (Released 26/03/2004)
=================================
This release is to fix a mistake I made in the clean up process which
means the tagger can take a lot longer to run than it really should do
especially on long documents. This version won't tag anything that v1.0
didn't but it will do the job faster!

Version 1.0 (Released 09/02/2004)
=================================
Currently compounds, elements and ions are tagged. Compounds which are
made up only of capital letters are not tagged as it introduces too
many false positives. For the same reason elements appearing as the
first token in a sentence are not tagged.

If a text contains "element token (Compound Formula)" then "element token"
will be tagged as a compound name, e.g. sulphur dioxide is tagged in the
text "sulphur dioxide (SO2)" (note that this also works if the formula is
surrounded by commas instead of parenthesis). Hopefully support for
compound names will be improved in a future release.
