## # automatic WSD # semyhe configuration file # kaarel kaljurand # Tue Oct 2 02:24:55 EET 2001 ## # NOTE: this is basically a Perl file, so # Perl syntax applies. File must end with 1; # window_size defines how many context words are # taken into consideration while disambiguating. # window size can be redefined on the commandline: # > semyhe --file myfile.txt --pos n --window 3 $window_size = 5; # should we consider the multiword units (detected # manually at the moment). this should always be set, # even if no multiword unit info is present $arvestan_kasitsi_tehtud_infot = 1; # vector that sets weights on the importance # of the context words, first element corresponds # to the closest neighbour of the word under analysis, # and so on... (experimental) #@relevance_vector = (0.2, 0.3, 0.4, 0.5, 0.7); #@relevance_vector = (0.9, 0.7, 0.5, 0.3, 0.2, 0.1, 0.1, 0.1); # should we use depth info (experimental) $use_depth = 1; # each selection of senses gets evaluated, best result ($best) # is the biggest result, others (result $x) can also make it to # the output iff $best/$x <= $threshold #$threshold = 1.2; #$threshold = 1.0; $threshold = 1.1; # how verbose should we be? # 0 means not verbose at all $debug_mode = 0; # conversion file declarations $convert_char_file = 'convert_char.cnv'; $convert_n_file = 'convert_n.cnv'; $convert_v_file = 'convert_v.cnv'; 1;