Skip to content

Commit

Permalink
experimental - philosophy updated
Browse files Browse the repository at this point in the history
  • Loading branch information
eubinecto committed Jul 12, 2022
1 parent 76b1609 commit cda5e6d
Showing 1 changed file with 8 additions and 1 deletion.
9 changes: 8 additions & 1 deletion explore/explore_politely_rules_as_dict.py
Original file line number Diff line number Diff line change
@@ -1,3 +1,10 @@
"""
Some philosophies to follow:
1. Rules account for 70% of the solution. Strive to hard-code the rules with intuitions and sweat. Keep the rules as general as possible.
2. The remaining 30% is based on the context. Don't use rules for that. It may be effective but it's inefficient (you won't get much from specific rules).
Disambiguate contexts with masked language models (e.g. Word2Vec, BERT, etc).
3. Every new feature must be tested. If writing Unit tests are not viable, then try to write a small demo.
"""
# The symbol to use for separating tags from texts
from typing import Dict, Tuple
import re
Expand All @@ -12,7 +19,7 @@
ALL_NO_CAPTURE = rf"{SEP}(?:[^\s{SEP}]+{TAG}EF)"
kiwi = Kiwi()

# --- all candidates for different styles of politeness --- #
# --- all EF's for different styles of politeness --- #
CASUAL = {
f"어{TAG}EF",
f"다{TAG}EF",
Expand Down

0 comments on commit cda5e6d

Please sign in to comment.