Name		Name	Last commit message	Last commit date
parent directory ..
BUILD.bazel		BUILD.bazel
README.md		README.md
README.txt		README.txt
aux_dictionary.tsv		aux_dictionary.tsv
collocation.txt		collocation.txt
collocation_suppression.txt		collocation_suppression.txt
connection_single_column.txt		connection_single_column.txt
dictionary00.txt		dictionary00.txt
dictionary01.txt		dictionary01.txt
dictionary02.txt		dictionary02.txt
dictionary03.txt		dictionary03.txt
dictionary04.txt		dictionary04.txt
dictionary05.txt		dictionary05.txt
dictionary06.txt		dictionary06.txt
dictionary07.txt		dictionary07.txt
dictionary08.txt		dictionary08.txt
dictionary09.txt		dictionary09.txt
dictionary_filter.tsv		dictionary_filter.tsv
evaluation.tsv		evaluation.tsv
id.def		id.def
reading_correction.tsv		reading_correction.tsv
suffix.txt		suffix.txt
suggestion_filter.txt		suggestion_filter.txt

README.md

data/dictionary_oss

AUX dictionary is a mechanism to add word entries to data/dictionary_oss/dictionary0x.txt without running the internal training pipeline.

Note

We recommend to update TSV files in data/dictionary_manual as an easier way. Please consider using data/dictionary_manual/ at first.

key	value	base_key	base_value	cost_offset
あるぱか	アルパカ	かぴばら	カピバラ	-1

aux_dictionary.tsv is a list of additional words to be added to Mozc's word dictionary.
key and value (i.e. アルパカ) are fields for the new word.
base_key and base_value (i.e. カピバラ) are fields of the reference word already in the dictionary.
cost_offset is the difference from the base cost. (i.e. -1 makes the new word prioritized than the base word).
other fields of the new word (e.g. lid, rid , cost) are copied from the reference word.

The entries in this file are used to filter out the entries in the dictionary.

key	value
あるぱか	或るパカ

key and value are the regexp patterns of words to be removed.
Each line is extended to the following regexp pattern:
- '^{key}\t\\d+\t\\d+\t\\d+\t{value}(\t.*)?\n$'