jwodder / scoregismu Goto Github PK

perl scoreGismu.pl [-g file | -G file] [-m num] [input file]

scoreGismu.pl implements steps 2 through 5 of the gismu scoring algorithm described in §4.14 of The Complete Lojban Language to produce a gismu from a given set of source words. Each line of input to the program specifies a set of Lojbanized source words and their language weights, e.g.:

iun 0.347 upiog 0.196 emploi 0.160 us 0.123 primin 0.089 amal 0.085

(If the same word appears more than once in the same line, its final weight is the sum of its listed weights.) Alternatively, a line may contain only Lojbanized source words, in which case the weights are the number of times each word appears on the line.

For each input line, scoreGismu.pl will calculate the gismu that match the given source words the best according to the scoring algorithm and output the top 5 scores (or an amount specified with the -m switch) and the gismu that obtained them. Note that calculations for separate lines do not take each other into account, and so if multiple gismu sets are produced at once, they may conflict with each other.

A file containing a list of pre-existing gismu, one per line, may optionally be specified with the -g switch. Any gismu listed in this file or that conflicts with a gismu in this file will be excluded from consideration for the algorithm.

A file containing a list of potential gismu, one per line, may optionally be specified with the -G switch. Only the gismu listed in this file will be considered by the algorithm. If both a -g switch and a -G switch are given, the -g switch is ignored.

This code was originally taken from [1], where it was credited to one "Hussell".

For reference, the language weights used in creating the official gismu were, as listed in §4.14 of the CLL:

Chinese     0.36
English     0.21
Hindi       0.16
Spanish     0.11
Russian     0.09
Arabic      0.07

Some later gismu instead used the weights:

Chinese     0.347
Hindi       0.196
English     0.160
Spanish     0.123
Russian     0.089
Arabic      0.085