Hugging Face's logo

Spaces:
mginoben
/
tagalog-profanity-classification
Sleeping

App Files Community
1
mginoben commited on
Commit
bce6532
•
1 Parent(s): 3172d47

Seperate text variables on external files

Browse files
Files changed (4) hide show
  1. contractions.json +33 -0
  2. lookup_words.txt +153 -0
  3. obj_pronouns.txt +10 -0
  4. profanities.json +16 -0
contractions.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "di": "hindi",
3
+ "to": "ito",
4
+ "no": "ano",
5
+ "kundi": "kung hindi",
6
+ "nya": "niya",
7
+ "nyo": "ninyo",
8
+ "niyo": "ninyo",
9
+ "pano": "paano",
10
+ "sainyo": "sa inyo",
11
+ "sayo": "sa iyo",
12
+ "pag": "kapag",
13
+ "kesa": "kaysa",
14
+ "dun": "doon",
15
+ "ganto": "ganito",
16
+ "nandun": "nandoon",
17
+ "saka": "tsaka",
18
+ "ung": "yung",
19
+ "wag": "huwag",
20
+ "sya": "siya",
21
+ "bat": "bakit",
22
+ "yon": "iyon",
23
+ "yun": "iyon",
24
+ "dyan": "diyan",
25
+ "jan": "diyan",
26
+ "andito": "nandito",
27
+ "san": "saan",
28
+ "ganun": "ganoon",
29
+ "kelan": "kailan",
30
+ "raw": "daw",
31
+ "itong": "ito ang",
32
+ "lng": "lang"
33
+ }
lookup_words.txt ADDED
@@ -0,0 +1,153 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ bobo
2
+ b0b0
3
+ bubog
4
+ bubon
5
+ bote
6
+ boboto
7
+ baboy
8
+ bwiset
9
+ bisikleta
10
+ bwisan
11
+ bwispikol
12
+ bwisak
13
+ bwisangka
14
+ bwisibida
15
+ bwisisi
16
+ gago
17
+ gagamba
18
+ gagapang
19
+ gagarin
20
+ gagawin
21
+ gagaya
22
+ gagayak
23
+ gagayat
24
+ gagbayaw
25
+ gagibang
26
+ gagilagid
27
+ gaging
28
+ gaginang
29
+ gagiring
30
+ kupal
31
+ kapal
32
+ kapak
33
+ kapalit
34
+ kupkup
35
+ kapuluan
36
+ kupad
37
+ kupas
38
+ kupat
39
+ kupa
40
+ kupak
41
+ kupali
42
+ kupas-kupas
43
+ kupi
44
+ kupig
45
+ kupit
46
+ kupo
47
+ kupog
48
+ kupong-kupong
49
+ kupra
50
+ kupya
51
+ pakshet
52
+ peksman
53
+ pekspeks
54
+ pekto
55
+ pekpek
56
+ pakpak
57
+ pakbet
58
+ paksain
59
+ pakikibaka
60
+ pakikipag
61
+ pakipot
62
+ pakitang
63
+ pakiusap
64
+ pakikiramay
65
+ pakinabang
66
+ pakinig
67
+ pakonti
68
+ paksiw
69
+ pakwan
70
+ pakyu
71
+ pakyaw
72
+ pakyas
73
+ pakay
74
+ pucha
75
+ pusila
76
+ pusod
77
+ pusong
78
+ puspos
79
+ pusakal
80
+ puchero
81
+ punyeta
82
+ punta
83
+ punla
84
+ punso
85
+ punas
86
+ punyal
87
+ punong
88
+ punay
89
+ punas
90
+ punlang
91
+ punugan
92
+ puno
93
+ punay
94
+ punyagi
95
+ puta
96
+ pota
97
+ putahe
98
+ putakti
99
+ puti
100
+ putik
101
+ puto
102
+ putok
103
+ putangina
104
+ tanga
105
+ tangna
106
+ tantra
107
+ tangay
108
+ tanghal
109
+ tanglaw
110
+ tangi
111
+ tanggal
112
+ tanguan
113
+ tanguay
114
+ tangway
115
+ tangis
116
+ tanguin
117
+ tangan
118
+ tanggalin
119
+ tanggol
120
+ tangina
121
+ tangal
122
+ tangan
123
+ tanganos
124
+ tangas
125
+ tangga
126
+ tanghal
127
+ tango
128
+ tangol
129
+ tangos
130
+ tarantado
131
+ taray
132
+ tarik
133
+ taripa
134
+ taro
135
+ ulol
136
+ bungol
137
+ bulol
138
+ bulag
139
+ bulakbol
140
+ bulalakaw
141
+ bilog
142
+ bilugan
143
+ bubog
144
+ burak
145
+ burot
146
+ balot
147
+ balut
148
+ baluktot
149
+ baluktotin
150
+ bulilit
151
+ kang
152
+ bubuka
153
+ buka
obj_pronouns.txt ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ ko
2
+ mo
3
+ nya
4
+ natin
5
+ namin
6
+ ninyo
7
+ nila
8
+ ka
9
+ nyo
10
+ ng
profanities.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bobo": ["boboka"],
3
+ "bwiset": ["buwesit", "buwiset"],
4
+ "gago": ["gaga"],
5
+ "kupal": [],
6
+ "pakshet": [],
7
+ "pakyu": [],
8
+ "pucha": [],
9
+ "punyeta": [],
10
+ "puta": ["pota"],
11
+ "putangina": ["pukingina", "kinangina"],
12
+ "tanga": [],
13
+ "tangina": ["tangna"],
14
+ "tarantado": [],
15
+ "ulol": []
16
+ }