--no_lower

Switch

--no_lower

Description

Preserves case when extracting n-gram/char-n-gram features, instead of lowercasing everything.

Argument and Default Value

No argument. Sets lowercaseonly to False; default (flag absent) is True (dlac.LOWERCASE_ONLY).

Details

Verified: with --no_lower, features keep their original case (e.g. 10AM, 128M show up as distinct features rather than being folded to lowercase), and the generated table name changes from feat$1gram$... to feat$1gramUp$... — the Up marks the table as case-preserved, so it never collides with the default lowercased table. Applies to --add_ngrams, --add_ngrams_from_tokenized, --add_char_ngrams, --use_collocs, and --gzip_csv.

Other Switches

Optional Switches:

• --add_ngrams, --add_char_ngrams, --use_collocs, etc.

Example Commands

Using the packaged dla_tutorial dataset (see Differential Language Analysis (DLA) Tutorial):

dlatkInterface.py -d dla_tutorial -t msgs -c user_id --add_ngrams -n 1 --no_lower