Why not just build your own? Presumably you are interested only
in those which end in full stop. Go through a corpus and make a list
of all strings followed by full stop. Count and uniq them and
compare them against the overall frequencies of the same strings (i.e.
including cases where they are not followed by full stops). In other
words, just find those strings with the highest mutual information
with full stop (perhaps weighted for absolute frequency). It should
be about 10 lines of Perl.
You have to be slightly careful cos some corpora don't use full stop
on any abbreviations.
Pete
-- E-mail: pete@sharp.co.uk \ Pete Whitelock Sharp global mail: SLEMV1::PETE \ Sharp Laboratories of Europe Ltd phone: +44 (0)1865 747711 \ Oxford Science Park fax: +44 (0)1865 714170 \ Oxford, OX4 4GA, England