use util;
#Config file for acronym extraction. EDIT THIS FILE, it should
#not be overridden by the software. It's read by GSDL using perl's
#'eval' function, so pretty much anything that's valid in perl is 
#valid here.

#Quite a few things here are defined in terms of recall and precision
#which are the key measures from Information Retreval (IR). If you
#don't understand recall and precision, any good IR textbook should
#explain them fully 

#the maximum range to look for acronyms (raise to raise precision)
$local_max_offset = 30;

#acronyms must be upper case (0 = false, 1 = true (high precision))
$local_upper_case = 1;

#acronym case must match (0 = false, 1 = true (high precision))
$local_case_match = 1;

#minimum acronym length (raise to raise precision)
$local_min_def_length = 3;

#let definitions be all capitals
$local_allow_all_caps = 0;

#minimum acronym length (raise to raise precision)
$local_min_acro_length = 3;

#minimum acronym length saving (raise to raise precision)
$local_min_length_saving = 4;

#allow recusive acronyms (0 = false (high precision), 1 = true)
$local_allow_recursive = 0;

#stop words-words allowed in acronyms (the multi-lingual version
#slows down acronym extraction slightly so is not the default)
#@local_stop_words = split / /, "A OF AT THE IN TO AND VON BEI DER DIE DAS DEM DEN DES UND DE DU A LA LE LES L DANS ET S";
@local_stop_words = split / /, "OF AT THE IN TO AND";

#the file to collate acronyms into
$acronym_accumulate_file = &util::filename_cat($ENV{'GSDLCOLLECTDIR'}, "etc","acronym_definitions.pm");

# any acronym definitions which should always be marked up can be copied here
# from the acronym_accumulate_file file ...
# 
# 
#