{
'module' => {
'name' => 'N-gram statistics',
'program' => 'uplug-ngramstat',
'location' => '$UplugBin',
'stin' => 'text',
'stdout' => 'text',
},
'input' => {
'ngram freq' => {
'format' => 'DBM',
'key' => ['token']
},
},
'output' => {
'phrases' => {
'stream name' => 'ngram stat',
# 'format' => 'uwa tab',
# 'columns' => ['ngram'],
# 'write_mode' => 'overwrite',
},
},
'parameter' => {
'token' => {
'minimal frequency' => 2,
# 'delimiter' => '\s+',
},
'statistics' => {
'minimal score' => 0.3,
'measure' => 'dice',
},
'runtime' => {
'print progress' => 1,
},
},
'arguments' => {
'shortcuts' => {
'in' => 'input:ngram freq:file',
'infile' => 'input:ngram freq:file',
'informat' => 'input:ngram freq:format',
'out' => 'output:ngram stat:file',
's' => 'parameter:statistics:measure',
'stat' => 'parameter:statistics:measure',
'm' => 'parameter:statistics:minimal score',
'min' => 'parameter:statistics:minimal score',
'prec' => 'parameter:statistics:precision',
'lang' => 'parameter:token:language (target)',
}
},
}