Skip to content

Commit

Permalink
Update human words.
Browse files Browse the repository at this point in the history
  • Loading branch information
emfomy committed May 6, 2020
1 parent 870cce1 commit 83cb7eb
Show file tree
Hide file tree
Showing 22 changed files with 4,236 additions and 7,238 deletions.
15 changes: 3 additions & 12 deletions ckipnlp/data/coref/__init__.py
Original file line number Diff line number Diff line change
@@ -1,12 +1,3 @@
#!/usr/bin/env python3
# -*- coding:utf-8 -*-

# pylint: disable=missing-docstring

__author__ = 'Mu Yang <http://muyang.pro>'
__copyright__ = '2018-2020 CKIP Lab'
__license__ = 'CC BY-NC-SA 4.0'

from ._human_words import *
from ._pronoun_words import *
from ._self_words import *
from ._human import *
from ._pronoun import *
from ._self import *
39 changes: 39 additions & 0 deletions ckipnlp/data/coref/_human/__init__.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,39 @@
from ._alumnus import ALUMNUS_WORDS
from ._aristocrat import ARISTOCRAT_WORDS
from ._candidate import CANDIDATE_WORDS
from ._fans import FANS_WORDS
from ._humanized import HUMANIZED_WORDS
from ._member import MEMBER_WORDS
from ._owner import OWNER_WORDS
from ._professional import PROFESSIONAL_WORDS
from ._recipient import RECIPIENT_WORDS
from ._religious_follower import RELIGIOUS_FOLLOWER_WORDS
from ._sage import SAGE_WORDS
from ._sick_patient import SICK_PATIENT_WORDS
from ._student import STUDENT_WORDS
from ._talent import TALENT_WORDS
from ._victim import VICTIM_WORDS
from ._warrior import WARRIOR_WORDS

from ._instance import HUMAN_INSTANCE_WORDS

HUMAN_WORDS = {
*ALUMNUS_WORDS,
*ARISTOCRAT_WORDS,
*CANDIDATE_WORDS,
*FANS_WORDS,
*HUMANIZED_WORDS,
*MEMBER_WORDS,
*OWNER_WORDS,
*PROFESSIONAL_WORDS,
*RECIPIENT_WORDS,
*RELIGIOUS_FOLLOWER_WORDS,
*SAGE_WORDS,
*SICK_PATIENT_WORDS,
*STUDENT_WORDS,
*TALENT_WORDS,
*VICTIM_WORDS,
*WARRIOR_WORDS,

*HUMAN_INSTANCE_WORDS,
}
5 changes: 5 additions & 0 deletions ckipnlp/data/coref/_human/_alumnus.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,5 @@
# 校友|alumnus

ALUMNUS_WORDS = {
'校友',
}
33 changes: 33 additions & 0 deletions ckipnlp/data/coref/_human/_aristocrat.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,33 @@
# 貴族|aristocrat

ARISTOCRAT_WORDS = {
'伯爵',
'侯',
'侯爵',
'侯爺',
'公侯',
'公爵',
'勳爵',
'國戚',
'大公',
'子爵',
'爵',
'爵士',
'王侯',
'王侯將相',
'王公',
'王公貴人',
'王孫',
'王爵',
'王爺',
'男爵',
'舍人',
'萬戶侯',
'親王',
'親貴',
'諸侯',
'貴戚',
'貴族',
'貴胄',
'邦伯',
}
8 changes: 8 additions & 0 deletions ckipnlp/data/coref/_human/_candidate.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,8 @@
# 人選|candidate

CANDIDATE_WORDS = {
'人選',
'候選人',
'應選人',
'被選人',
}
17 changes: 17 additions & 0 deletions ckipnlp/data/coref/_human/_fans.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,17 @@
# 迷|fans

FANS_WORDS = {
'影迷',
'戲迷',
'棋迷',
'樂友',
'樂迷',
'歌迷',
'洋迷',
'球迷',
'票友',
'粉絲',
'舞迷',
'詩癖',
'音樂迷',
}
257 changes: 257 additions & 0 deletions ckipnlp/data/coref/_human/_humanized.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,257 @@
# humanized|擬人

HUMANIZED_WORDS = {
'三極',
'上天',
'上帝',
'上蒼',
'主',
'主基督',
'二郎神',
'亡靈',
'亡魂',
'人子',
'人魚',
'仙',
'仙人',
'仙女',
'仙姑',
'仙子',
'仙家',
'仙童',
'仙翁',
'佛',
'佛佗',
'佛寶',
'佛爺',
'佛祖',
'佛陀',
'僵尸',
'僵屍',
'元神',
'八仙',
'冤鬼',
'冤魂',
'凶神',
'十八羅漢',
'厲鬼',
'后土',
'吸血鬼',
'哼哈二將',
'回祿',
'土地',
'土地公',
'地祇',
'基督',
'夜叉',
'大仙',
'大羅神仙',
'天',
'天主',
'天人',
'天仙',
'天使',
'天公',
'天后',
'天帝',
'天父',
'天王',
'天神',
'天龍',
'太陽神',
'女神',
'好兄弟',
'妖',
'妖兒',
'妖女',
'妖怪',
'妖精',
'妖魔',
'妖魔鬼怪',
'媽祖',
'媽祖婆',
'孤魂',
'孤魂野鬼',
'守護神',
'安琪兒',
'山君',
'山神',
'巨靈',
'帝君',
'年獸',
'幸運之神',
'幽靈',
'幽魂',
'床母',
'庶神',
'怨鬼',
'怪物',
'怪獸',
'恩主',
'惡煞',
'惡神',
'惡鬼',
'惡魔',
'愛神',
'戰神',
'撒但',
'撒旦',
'散仙',
'春神',
'月之女神',
'月神',
'林默娘',
'樹神',
'死神',
'殭',
'殭尸',
'殭屍',
'水怪',
'水神',
'水鬼',
'水魔',
'河神',
'泥菩薩',
'海神',
'海龍王',
'游魂',
'火神',
'灶君',
'灶王爺',
'灶神',
'煙客',
'煞',
'煞星',
'煞氣',
'煞神',
'牛頭馬面',
'牛鬼蛇神',
'牟尼',
'牟尼佛',
'狐仙',
'獰鬼',
'玉女',
'玉帝',
'玉皇',
'玉皇上帝',
'玉皇大帝',
'王母娘娘',
'異形',
'異物',
'疫鬼',
'病魔',
'瘟神',
'白衣大士',
'皇天',
'皇天后土',
'真主',
'真神',
'矮靈',
'碟仙',
'祇',
'祝融',
'神',
'神仙',
'神女',
'神怪',
'神明',
'神格',
'神祇',
'神衹',
'神道',
'神靈',
'神鬼',
'神魂',
'神魔',
'福德正神',
'福神',
'稷',
'穹天',
'精怪',
'精靈',
'羅剎',
'羅漢',
'美人魚',
'老天',
'老天爺',
'老妖',
'耶和華',
'耶穌',
'耶穌基督',
'聖母',
'聖父',
'聖靈',
'花神',
'芳魂',
'菩薩',
'蜮',
'衹',
'觀世音',
'觀世音大士',
'觀世音菩薩',
'觀音',
'觀音佛祖',
'觀音大士',
'觀音菩薩',
'護法',
'財神',
'財神爺',
'農神',
'造物主',
'造物者',
'遊魂',
'過路財神',
'邪靈',
'邪鬼',
'邪魔',
'酒神',
'釋尊',
'釋迦',
'釋迦牟尼',
'釋迦牟尼佛',
'金剛',
'金童',
'金鋼',
'門神',
'閰王',
'閻王',
'閻王爺',
'閻羅王',
'阿拉',
'陰童',
'雨神',
'靈',
'靈魂',
'風姨',
'風神',
'飛天',
'饕餮',
'鬼',
'鬼卒',
'鬼影',
'鬼怪',
'鬼物',
'鬼王',
'鬼神',
'鬼蜮',
'鬼魂',
'鬼魅',
'魂',
'魂魄',
'魃',
'魅',
'魅影',
'魍',
'魍魅',
'魍魍兒',
'魍魎',
'魑',
'魑魅',
'魔',
'魔手',
'魔邪',
'魔障',
'魔鬼',
'龍王',
'龍王爺',
}

0 comments on commit 83cb7eb

Please sign in to comment.