keyphrase 0.1.3 → 0.2.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/README.md +5 -4
- data/lib/keyphrase/stoplist/afr.rb +14 -0
- data/lib/keyphrase/stoplist/aka.rb +10 -0
- data/lib/keyphrase/stoplist/amh.rb +10 -0
- data/lib/keyphrase/stoplist/ara.rb +490 -0
- data/lib/keyphrase/stoplist/aze.rb +175 -0
- data/lib/keyphrase/stoplist/bel.rb +11 -0
- data/lib/keyphrase/stoplist/ben.rb +408 -0
- data/lib/keyphrase/stoplist/bul.rb +528 -0
- data/lib/keyphrase/stoplist/cat.rb +711 -0
- data/lib/keyphrase/stoplist/ces.rb +560 -0
- data/lib/keyphrase/stoplist/cmn.rb +1119 -0
- data/lib/keyphrase/stoplist/dan.rb +25 -0
- data/lib/keyphrase/stoplist/deu.rb +631 -0
- data/lib/keyphrase/stoplist/ell.rb +275 -0
- data/lib/keyphrase/stoplist/eng.rb +2 -589
- data/lib/keyphrase/stoplist/epo.rb +183 -0
- data/lib/keyphrase/stoplist/est.rb +13 -0
- data/lib/keyphrase/stoplist/fin.rb +857 -0
- data/lib/keyphrase/stoplist/fra.rb +699 -0
- data/lib/keyphrase/stoplist/guj.rb +234 -0
- data/lib/keyphrase/stoplist/heb.rb +204 -0
- data/lib/keyphrase/stoplist/hin.rb +235 -0
- data/lib/keyphrase/stoplist/hrv.rb +25 -0
- data/lib/keyphrase/stoplist/hun.rb +1195 -0
- data/lib/keyphrase/stoplist/hye.rb +55 -0
- data/lib/keyphrase/stoplist/ind.rb +768 -0
- data/lib/keyphrase/stoplist/ita.rb +670 -0
- data/lib/keyphrase/stoplist/jav.rb +10 -0
- data/lib/keyphrase/stoplist/jpn.rb +144 -0
- data/lib/keyphrase/stoplist/kan.rb +92 -0
- data/lib/keyphrase/stoplist/kat.rb +383 -0
- data/lib/keyphrase/stoplist/khm.rb +245 -0
- data/lib/keyphrase/stoplist/kor.rb +610 -0
- data/lib/keyphrase/stoplist/lat.rb +14 -0
- data/lib/keyphrase/stoplist/lav.rb +171 -0
- data/lib/keyphrase/stoplist/lit.rb +484 -0
- data/lib/keyphrase/stoplist/mal.rb +11 -0
- data/lib/keyphrase/stoplist/mar.rb +109 -0
- data/lib/keyphrase/stoplist/mkd.rb +11 -0
- data/lib/keyphrase/stoplist/mya.rb +285 -0
- data/lib/keyphrase/stoplist/nep.rb +265 -0
- data/lib/keyphrase/stoplist/nld.rb +423 -0
- data/lib/keyphrase/stoplist/nob.rb +186 -0
- data/lib/keyphrase/stoplist/ori.rb +11 -0
- data/lib/keyphrase/stoplist/pan.rb +473 -0
- data/lib/keyphrase/stoplist/pes.rb +801 -0
- data/lib/keyphrase/stoplist/pol.rb +338 -0
- data/lib/keyphrase/stoplist/por.rb +570 -0
- data/lib/keyphrase/stoplist/ron.rb +444 -0
- data/lib/keyphrase/stoplist/rus.rb +569 -0
- data/lib/keyphrase/stoplist/sin.rb +10 -0
- data/lib/keyphrase/stoplist/slk.rb +428 -0
- data/lib/keyphrase/stoplist/slv.rb +456 -0
- data/lib/keyphrase/stoplist/sna.rb +11 -0
- data/lib/keyphrase/stoplist/spa.rb +731 -0
- data/lib/keyphrase/stoplist/srp.rb +11 -0
- data/lib/keyphrase/stoplist/swe.rb +428 -0
- data/lib/keyphrase/stoplist/tam.rb +135 -0
- data/lib/keyphrase/stoplist/tel.rb +10 -0
- data/lib/keyphrase/stoplist/tgl.rb +157 -0
- data/lib/keyphrase/stoplist/tha.rb +125 -0
- data/lib/keyphrase/stoplist/tuk.rb +11 -0
- data/lib/keyphrase/stoplist/tur.rb +514 -0
- data/lib/keyphrase/stoplist/ukr.rb +38 -0
- data/lib/keyphrase/stoplist/urd.rb +527 -0
- data/lib/keyphrase/stoplist/uzb.rb +10 -0
- data/lib/keyphrase/stoplist/vie.rb +655 -0
- data/lib/keyphrase/stoplist/yid.rb +204 -0
- data/lib/keyphrase/stoplist/zul.rb +39 -0
- data/lib/keyphrase/stoplist.rb +13 -10
- data/lib/keyphrase/version.rb +1 -1
- data/lib/keyphrase.rb +20 -12
- metadata +71 -3
|
@@ -0,0 +1,109 @@
|
|
|
1
|
+
class Keyphrase
|
|
2
|
+
module Stoplist
|
|
3
|
+
class Mar
|
|
4
|
+
def self.stopwords
|
|
5
|
+
@@stopwords ||= [
|
|
6
|
+
"अधिक",
|
|
7
|
+
"अनेक",
|
|
8
|
+
"अशी",
|
|
9
|
+
"असलयाचे",
|
|
10
|
+
"असलेल्या",
|
|
11
|
+
"असा",
|
|
12
|
+
"असून",
|
|
13
|
+
"असे",
|
|
14
|
+
"आज",
|
|
15
|
+
"आणि",
|
|
16
|
+
"आता",
|
|
17
|
+
"आपल्या",
|
|
18
|
+
"आला",
|
|
19
|
+
"आली",
|
|
20
|
+
"आले",
|
|
21
|
+
"आहे",
|
|
22
|
+
"आहेत",
|
|
23
|
+
"एक",
|
|
24
|
+
"एका",
|
|
25
|
+
"कमी",
|
|
26
|
+
"करणयात",
|
|
27
|
+
"करून",
|
|
28
|
+
"का",
|
|
29
|
+
"काम",
|
|
30
|
+
"काय",
|
|
31
|
+
"काही",
|
|
32
|
+
"किवा",
|
|
33
|
+
"की",
|
|
34
|
+
"केला",
|
|
35
|
+
"केली",
|
|
36
|
+
"केले",
|
|
37
|
+
"कोटी",
|
|
38
|
+
"गेल्या",
|
|
39
|
+
"घेऊन",
|
|
40
|
+
"जात",
|
|
41
|
+
"झाला",
|
|
42
|
+
"झाली",
|
|
43
|
+
"झाले",
|
|
44
|
+
"झालेल्या",
|
|
45
|
+
"टा",
|
|
46
|
+
"डॉ",
|
|
47
|
+
"तर",
|
|
48
|
+
"तरी",
|
|
49
|
+
"तसेच",
|
|
50
|
+
"ता",
|
|
51
|
+
"ती",
|
|
52
|
+
"तीन",
|
|
53
|
+
"ते",
|
|
54
|
+
"तो",
|
|
55
|
+
"त्या",
|
|
56
|
+
"त्याचा",
|
|
57
|
+
"त्याची",
|
|
58
|
+
"त्याच्या",
|
|
59
|
+
"त्याना",
|
|
60
|
+
"त्यानी",
|
|
61
|
+
"त्यामुळे",
|
|
62
|
+
"त्री",
|
|
63
|
+
"दिली",
|
|
64
|
+
"दोन",
|
|
65
|
+
"न",
|
|
66
|
+
"नाही",
|
|
67
|
+
"निर्ण्य",
|
|
68
|
+
"पण",
|
|
69
|
+
"पम",
|
|
70
|
+
"परयतन",
|
|
71
|
+
"पाटील",
|
|
72
|
+
"म",
|
|
73
|
+
"मात्र",
|
|
74
|
+
"माहिती",
|
|
75
|
+
"मी",
|
|
76
|
+
"मुबी",
|
|
77
|
+
"म्हणजे",
|
|
78
|
+
"म्हणाले",
|
|
79
|
+
"म्हणून",
|
|
80
|
+
"या",
|
|
81
|
+
"याचा",
|
|
82
|
+
"याची",
|
|
83
|
+
"याच्या",
|
|
84
|
+
"याना",
|
|
85
|
+
"यानी",
|
|
86
|
+
"येणार",
|
|
87
|
+
"येत",
|
|
88
|
+
"येथील",
|
|
89
|
+
"येथे",
|
|
90
|
+
"लाख",
|
|
91
|
+
"व",
|
|
92
|
+
"व्यकत",
|
|
93
|
+
"सर्व",
|
|
94
|
+
"सागित्ले",
|
|
95
|
+
"सुरू",
|
|
96
|
+
"हजार",
|
|
97
|
+
"हा",
|
|
98
|
+
"ही",
|
|
99
|
+
"हे",
|
|
100
|
+
"होणार",
|
|
101
|
+
"होत",
|
|
102
|
+
"होता",
|
|
103
|
+
"होती",
|
|
104
|
+
"होते",
|
|
105
|
+
]
|
|
106
|
+
end
|
|
107
|
+
end
|
|
108
|
+
end
|
|
109
|
+
end
|
|
@@ -0,0 +1,285 @@
|
|
|
1
|
+
class Keyphrase
|
|
2
|
+
module Stoplist
|
|
3
|
+
class Mya
|
|
4
|
+
def self.stopwords
|
|
5
|
+
@@stopwords ||= [
|
|
6
|
+
"အပေါ်",
|
|
7
|
+
"အနက်",
|
|
8
|
+
"အမြဲတမ်း",
|
|
9
|
+
"အတွင်းတွင်",
|
|
10
|
+
"မကြာမီ",
|
|
11
|
+
"မတိုင်မီ",
|
|
12
|
+
"ဒါ့အပြင်",
|
|
13
|
+
"အောက်မှာ",
|
|
14
|
+
"အထဲမှာ",
|
|
15
|
+
"ဘယ်တော့မျှ",
|
|
16
|
+
"မကြာခဏ",
|
|
17
|
+
"တော်တော်လေး",
|
|
18
|
+
"စဉ်တွင်",
|
|
19
|
+
"နှင့်အတူ",
|
|
20
|
+
"နှင့်",
|
|
21
|
+
"နှင့်တကွ",
|
|
22
|
+
"ကျွန်တော်",
|
|
23
|
+
"ကျွန်မ",
|
|
24
|
+
"ငါ",
|
|
25
|
+
"ကျုပ်",
|
|
26
|
+
"ကျွနု်ပ်",
|
|
27
|
+
"ကျနော်",
|
|
28
|
+
"ကျမ",
|
|
29
|
+
"သူ",
|
|
30
|
+
"သူမ",
|
|
31
|
+
"ထိုဟာ",
|
|
32
|
+
"ထိုအရာ",
|
|
33
|
+
"ဤအရာ",
|
|
34
|
+
"ထို",
|
|
35
|
+
"၄င်း",
|
|
36
|
+
"ကျွန်တော်တို့",
|
|
37
|
+
"ကျွန်မတို့",
|
|
38
|
+
"ငါတို့",
|
|
39
|
+
"ကျုပ်တို့",
|
|
40
|
+
"ကျွနု်ပ်တို့",
|
|
41
|
+
"ကျနော်တို့",
|
|
42
|
+
"ကျမတို့",
|
|
43
|
+
"သင်",
|
|
44
|
+
"သင်တို့",
|
|
45
|
+
"နင်တို့",
|
|
46
|
+
"မင်း",
|
|
47
|
+
"မင်းတို့",
|
|
48
|
+
"သူတို့",
|
|
49
|
+
"ကျွန်တော်အား",
|
|
50
|
+
"ကျွန်တော်ကို",
|
|
51
|
+
"ကျွန်မကို",
|
|
52
|
+
"ငါကို",
|
|
53
|
+
"ကျုပ်ကို",
|
|
54
|
+
"ကျွနု်ပ်ကို",
|
|
55
|
+
"သူ့ကို",
|
|
56
|
+
"သူမကို",
|
|
57
|
+
"ထိုအရာကို",
|
|
58
|
+
"သင့်ကို",
|
|
59
|
+
"သင်တို့ကို",
|
|
60
|
+
"နင်တို့ကို",
|
|
61
|
+
"မင်းကို",
|
|
62
|
+
"မင်းတို့ကို",
|
|
63
|
+
"ငါတို့ကို",
|
|
64
|
+
"ကျုပ်တို့ကို",
|
|
65
|
+
"ကျွနု်ပ်တို့ကို",
|
|
66
|
+
"မိမိကိုယ်တိုင်",
|
|
67
|
+
"မိမိဘာသာ",
|
|
68
|
+
"မင်းကိုယ်တိုင်",
|
|
69
|
+
"မင်းဘာသာ",
|
|
70
|
+
"မင်းတို့ကိုယ်တိုင်",
|
|
71
|
+
"မင်းတို့ဘာသာ",
|
|
72
|
+
"သူကိုယ်တိုင်",
|
|
73
|
+
"ကိုယ်တိုင်",
|
|
74
|
+
"သူမကိုယ်တိုင်",
|
|
75
|
+
"သူ့ဘာသာ",
|
|
76
|
+
"သူ့ကိုယ်ကို",
|
|
77
|
+
"ကိုယ့်ကိုယ်ကို",
|
|
78
|
+
"မိမိကိုယ်ကို",
|
|
79
|
+
"၄င်းပင်",
|
|
80
|
+
"ထိုအရာပင်",
|
|
81
|
+
"သည့်",
|
|
82
|
+
"မည့်",
|
|
83
|
+
"တဲ့",
|
|
84
|
+
"ကျွနု်ပ်၏",
|
|
85
|
+
"ကျွန်တော်၏",
|
|
86
|
+
"ကျွန်မ၏",
|
|
87
|
+
"ကျနော်၏",
|
|
88
|
+
"ကျမ၏",
|
|
89
|
+
"သူ၏",
|
|
90
|
+
"သူမ၏",
|
|
91
|
+
"ထိုအရာ၏",
|
|
92
|
+
"ထိုဟာ၏",
|
|
93
|
+
"ကျွနု်ပ်တို့၏",
|
|
94
|
+
"ငါတို့၏",
|
|
95
|
+
"ကျွန်တော်တို့၏",
|
|
96
|
+
"ကျွန်မတို့၏",
|
|
97
|
+
"ကျနော်တို့၏",
|
|
98
|
+
"ကျမတို့၏",
|
|
99
|
+
"သင်၏",
|
|
100
|
+
"သင်တို့၏",
|
|
101
|
+
"မင်း၏",
|
|
102
|
+
"မင်းတို့၏",
|
|
103
|
+
"သူတို့၏",
|
|
104
|
+
"ကျွန်တော့်ဟာ",
|
|
105
|
+
"ကျွန်မဟာ",
|
|
106
|
+
"ကျနော်၏ဟာ",
|
|
107
|
+
"ကျမ၏ဟာ",
|
|
108
|
+
"ကျမဟာ",
|
|
109
|
+
"ကျနော်ဟာ",
|
|
110
|
+
"သူဟာ",
|
|
111
|
+
"သူမဟာ",
|
|
112
|
+
"သူ့ဟာ",
|
|
113
|
+
"ကျွနု်ပ်တို့ဟာ",
|
|
114
|
+
"ကျွန်တော်တို့ဟာ",
|
|
115
|
+
"ကျွန်မတို့ဟာ",
|
|
116
|
+
"သင်တို့ဟာ",
|
|
117
|
+
"မင်းတို့ဟာ",
|
|
118
|
+
"သူတို့ဟာ",
|
|
119
|
+
"သူမတို့ဟာ",
|
|
120
|
+
"ဤအရာ",
|
|
121
|
+
"ဟောဒါ",
|
|
122
|
+
"ဟောဒီ",
|
|
123
|
+
"ဟောဒီဟာ",
|
|
124
|
+
"ဒီဟာ",
|
|
125
|
+
"ဒါ",
|
|
126
|
+
"ထိုအရာ",
|
|
127
|
+
"၄င်းအရာ",
|
|
128
|
+
"ယင်းအရာ",
|
|
129
|
+
"အဲဒါ",
|
|
130
|
+
"ဟိုဟာ",
|
|
131
|
+
"အချို့",
|
|
132
|
+
"တစ်ခုခု",
|
|
133
|
+
"အဘယ်မဆို",
|
|
134
|
+
"ဘယ်အရာမဆို",
|
|
135
|
+
"အဘယ်မည်သော",
|
|
136
|
+
"အကြင်",
|
|
137
|
+
"အရာရာတိုင်း",
|
|
138
|
+
"စိုးစဉ်မျှ",
|
|
139
|
+
"စိုးစဉ်းမျှ",
|
|
140
|
+
"ဘယ်လောက်မဆို",
|
|
141
|
+
"တစ်စုံတစ်ရာ",
|
|
142
|
+
"တစုံတရာ",
|
|
143
|
+
"အလျဉ်းမဟုတ်",
|
|
144
|
+
"မည်သည့်နည်းနှင့်မျှမဟုတ်",
|
|
145
|
+
"အလျဉ်းမရှိသော",
|
|
146
|
+
"အခြားဖြစ်သော",
|
|
147
|
+
"အခြားသော",
|
|
148
|
+
"အခြားတစ်ခု",
|
|
149
|
+
"အခြားတစ်ယောက်",
|
|
150
|
+
"အားလုံး",
|
|
151
|
+
"အရာရာတိုင်း",
|
|
152
|
+
"အကုန်လုံး",
|
|
153
|
+
"အလုံးစုံ",
|
|
154
|
+
"အရာခပ်သိမ်း",
|
|
155
|
+
"တစ်ခုစီ",
|
|
156
|
+
"အသီးသီး",
|
|
157
|
+
"တစ်ဦးဦး",
|
|
158
|
+
"တစ်ခုခု",
|
|
159
|
+
"ကိုယ်စီကိုယ်ငှ",
|
|
160
|
+
"ကိုယ်စီ",
|
|
161
|
+
"တစ်ဦးစီ",
|
|
162
|
+
"တစ်ယောက်စီ",
|
|
163
|
+
"တစ်ခုစီ",
|
|
164
|
+
"အကုန်",
|
|
165
|
+
"အပြည့်အစုံ",
|
|
166
|
+
"လုံးလုံး",
|
|
167
|
+
"နှစ်ခုလုံး",
|
|
168
|
+
"နှစ်ယောက်လုံး",
|
|
169
|
+
"နှစ်ဘက်လုံး",
|
|
170
|
+
"တစ်စုံတစ်ရာ",
|
|
171
|
+
"တစ်စုံတစ်ခု",
|
|
172
|
+
"တစုံတခု",
|
|
173
|
+
"တစ်စုံတစ်ယောက်",
|
|
174
|
+
"တစုံတယောက်",
|
|
175
|
+
"တစ်ယောက်ယောက်",
|
|
176
|
+
"မည်သူမဆို",
|
|
177
|
+
"ဘာမျှမရှိ",
|
|
178
|
+
"ဘာမှမရှိ",
|
|
179
|
+
"အဘယ်အရာမျှမရှိ",
|
|
180
|
+
"လူတိုင်း",
|
|
181
|
+
"လူတကာ",
|
|
182
|
+
"နှင့်",
|
|
183
|
+
"ပြီးလျှင်",
|
|
184
|
+
"၄င်းနောက်",
|
|
185
|
+
"သို့မဟုတ်",
|
|
186
|
+
"သို့တည်းမဟုတ်",
|
|
187
|
+
"သို့မဟုတ်လျှင်",
|
|
188
|
+
"ဒါမှမဟုတ်",
|
|
189
|
+
"ဖြစ်စေ",
|
|
190
|
+
"သို့စေကာမူ",
|
|
191
|
+
"ဒါပေမယ့်",
|
|
192
|
+
"ဒါပေမဲ့",
|
|
193
|
+
"မှတစ်ပါး",
|
|
194
|
+
"မှလွဲလျှင်",
|
|
195
|
+
"အဘယ်ကြောင့်ဆိုသော်",
|
|
196
|
+
"သောကြောင့်",
|
|
197
|
+
"သဖြင့်",
|
|
198
|
+
"၍",
|
|
199
|
+
"သည့်အတွက်ကြောင့်",
|
|
200
|
+
"လျှင်",
|
|
201
|
+
"ပါက",
|
|
202
|
+
"အကယ်၍",
|
|
203
|
+
"သော်ငြားလည်း",
|
|
204
|
+
"စေကာမူ",
|
|
205
|
+
"နည်းတူ",
|
|
206
|
+
"ပေမယ့်",
|
|
207
|
+
"ပေမဲ့",
|
|
208
|
+
"ထိုနည်းတူစွာ",
|
|
209
|
+
"ထိုနည်းတူ",
|
|
210
|
+
"ကဲ့သို့",
|
|
211
|
+
"သကဲ့သို့",
|
|
212
|
+
"ယင်းကဲ့သို့",
|
|
213
|
+
"ထိုကဲ့သို့",
|
|
214
|
+
"နှင့်စပ်လျဉ်း၍",
|
|
215
|
+
"ဤမျှ",
|
|
216
|
+
"ဤမျှလောက်",
|
|
217
|
+
"ဤကဲ့သို့",
|
|
218
|
+
"အခုလောက်ထိ",
|
|
219
|
+
"ဒါကတော့",
|
|
220
|
+
"အဘယ်ကဲ့သလို့",
|
|
221
|
+
"မည်ကဲ့သို့",
|
|
222
|
+
"မည်သည့်နည်းနှင့်",
|
|
223
|
+
"မည်သည့်နည်းဖြင့်",
|
|
224
|
+
"မည်သည့်နည့်နှင့်မဆို",
|
|
225
|
+
"မည်သည့်နည်းဖြင့်မဆို",
|
|
226
|
+
"မည်သို့",
|
|
227
|
+
"ဘယ်လိုလဲ",
|
|
228
|
+
"သို့ပေတည့်",
|
|
229
|
+
"သို့ပေမည့်",
|
|
230
|
+
"ဘယ်နည်းနှင့်",
|
|
231
|
+
"မည်ရွေ့မည်မျှ",
|
|
232
|
+
"အဘယ်မျှလောက်",
|
|
233
|
+
"ဘယ်လောက်",
|
|
234
|
+
"မည်သူ",
|
|
235
|
+
"ဘယ်သူ",
|
|
236
|
+
"မည်သည့်အကြောင်းကြောင့်",
|
|
237
|
+
"ဘာအတွက်ကြောင့်",
|
|
238
|
+
"အဘယ်ကြောင့်",
|
|
239
|
+
"မည်သည့်အတွက်ကြောင့်",
|
|
240
|
+
"ဘာကြောင့်",
|
|
241
|
+
"ဘာအတွက်နဲ့လဲ",
|
|
242
|
+
"မည်သည်",
|
|
243
|
+
"ဘာလဲ",
|
|
244
|
+
"အဘယ်အရာနည်း",
|
|
245
|
+
"မည်သည့်အရပ်မှာ",
|
|
246
|
+
"ဘယ်နေရာတွင်",
|
|
247
|
+
"မည်သည့်နေရာတွင်",
|
|
248
|
+
"မည်သည့်နေရာသို့",
|
|
249
|
+
"ဘယ်နေရာသို့",
|
|
250
|
+
"ဘယ်နေရာမှာ",
|
|
251
|
+
"ဘယ်သူ၏",
|
|
252
|
+
"မည်သည့်အရာ၏",
|
|
253
|
+
"မည်သည့်အခါ",
|
|
254
|
+
"ဘယ်အချိန်",
|
|
255
|
+
"ဘယ်အခါ",
|
|
256
|
+
"မည်သည့်အချိန်",
|
|
257
|
+
"ဘယ်တော့",
|
|
258
|
+
"မည်သူကို",
|
|
259
|
+
"မည်သူက",
|
|
260
|
+
"ဘယ်သူ့ကို",
|
|
261
|
+
"မည်သူမည်ဝါ",
|
|
262
|
+
"မည်သည့်အရာ",
|
|
263
|
+
"ဘယ်အရာ",
|
|
264
|
+
"မည်သို့ပင်ဖြစ်စေ",
|
|
265
|
+
"ဘယ်လိုပဲဖြစ်ဖြစ်",
|
|
266
|
+
"မည်ရွေ့မည်မျှဖြစ်စေ",
|
|
267
|
+
"မည်သည့်နည်းနှင့်မဆို",
|
|
268
|
+
"ဘယ်နည်းနဲ့ဖြစ်ဖြစ်",
|
|
269
|
+
"မည်သူမဆို",
|
|
270
|
+
"ဘယ်သူမဆို",
|
|
271
|
+
"အဘယ်သူမဆို",
|
|
272
|
+
"မည်သည့်အရာမဆို",
|
|
273
|
+
"ဘာဖြစ်ဖြစ်",
|
|
274
|
+
"မည်သည့်အရာဖြစ်ဖြစ်",
|
|
275
|
+
"မည်သည့်အရပ်၌မဆို",
|
|
276
|
+
"မည်သည့်နေရာမဆို",
|
|
277
|
+
"ဘယ်အခါမဆို",
|
|
278
|
+
"ဘယ်အချိန်မဆို",
|
|
279
|
+
"ဘယ်အခါဖြစ်ဖြစ်",
|
|
280
|
+
"အချိန်အခါမရွေး",
|
|
281
|
+
]
|
|
282
|
+
end
|
|
283
|
+
end
|
|
284
|
+
end
|
|
285
|
+
end
|
|
@@ -0,0 +1,265 @@
|
|
|
1
|
+
class Keyphrase
|
|
2
|
+
module Stoplist
|
|
3
|
+
class Nep
|
|
4
|
+
def self.stopwords
|
|
5
|
+
@@stopwords ||= [
|
|
6
|
+
"छ",
|
|
7
|
+
"र",
|
|
8
|
+
"पनि",
|
|
9
|
+
"छन्",
|
|
10
|
+
"लागि",
|
|
11
|
+
"भएको",
|
|
12
|
+
"गरेको",
|
|
13
|
+
"भने",
|
|
14
|
+
"गर्न",
|
|
15
|
+
"गर्ने",
|
|
16
|
+
"हो",
|
|
17
|
+
"तथा",
|
|
18
|
+
"यो",
|
|
19
|
+
"रहेको",
|
|
20
|
+
"उनले",
|
|
21
|
+
"थियो",
|
|
22
|
+
"हुने",
|
|
23
|
+
"गरेका",
|
|
24
|
+
"थिए",
|
|
25
|
+
"गर्दै",
|
|
26
|
+
"तर",
|
|
27
|
+
"नै",
|
|
28
|
+
"को",
|
|
29
|
+
"मा",
|
|
30
|
+
"हुन्",
|
|
31
|
+
"भन्ने",
|
|
32
|
+
"हुन",
|
|
33
|
+
"गरी",
|
|
34
|
+
"त",
|
|
35
|
+
"हुन्छ",
|
|
36
|
+
"अब",
|
|
37
|
+
"के",
|
|
38
|
+
"रहेका",
|
|
39
|
+
"गरेर",
|
|
40
|
+
"छैन",
|
|
41
|
+
"दिए",
|
|
42
|
+
"भए",
|
|
43
|
+
"यस",
|
|
44
|
+
"ले",
|
|
45
|
+
"गर्नु",
|
|
46
|
+
"औं",
|
|
47
|
+
"सो",
|
|
48
|
+
"त्यो",
|
|
49
|
+
"कि",
|
|
50
|
+
"जुन",
|
|
51
|
+
"यी",
|
|
52
|
+
"का",
|
|
53
|
+
"गरि",
|
|
54
|
+
"ती",
|
|
55
|
+
"न",
|
|
56
|
+
"छु",
|
|
57
|
+
"छौं",
|
|
58
|
+
"लाई",
|
|
59
|
+
"नि",
|
|
60
|
+
"उप",
|
|
61
|
+
"अक्सर",
|
|
62
|
+
"आदि",
|
|
63
|
+
"कसरी",
|
|
64
|
+
"क्रमशः",
|
|
65
|
+
"चाले",
|
|
66
|
+
"अगाडी",
|
|
67
|
+
"अझै",
|
|
68
|
+
"अनुसार",
|
|
69
|
+
"अन्तर्गत",
|
|
70
|
+
"अन्य",
|
|
71
|
+
"अन्यत्र",
|
|
72
|
+
"अन्यथा",
|
|
73
|
+
"अरु",
|
|
74
|
+
"अरुलाई",
|
|
75
|
+
"अर्को",
|
|
76
|
+
"अर्थात",
|
|
77
|
+
"अर्थात्",
|
|
78
|
+
"अलग",
|
|
79
|
+
"आए",
|
|
80
|
+
"आजको",
|
|
81
|
+
"ओठ",
|
|
82
|
+
"आत्म",
|
|
83
|
+
"आफू",
|
|
84
|
+
"आफूलाई",
|
|
85
|
+
"आफ्नै",
|
|
86
|
+
"आफ्नो",
|
|
87
|
+
"आयो",
|
|
88
|
+
"उदाहरण",
|
|
89
|
+
"उनको",
|
|
90
|
+
"उहालाई",
|
|
91
|
+
"एउटै",
|
|
92
|
+
"एक",
|
|
93
|
+
"एकदम",
|
|
94
|
+
"कतै",
|
|
95
|
+
"कम से कम",
|
|
96
|
+
"कसै",
|
|
97
|
+
"कसैले",
|
|
98
|
+
"कहाँबाट",
|
|
99
|
+
"कहिलेकाहीं",
|
|
100
|
+
"का",
|
|
101
|
+
"किन",
|
|
102
|
+
"किनभने",
|
|
103
|
+
"कुनै",
|
|
104
|
+
"कुरा",
|
|
105
|
+
"कृपया",
|
|
106
|
+
"केही",
|
|
107
|
+
"कोही",
|
|
108
|
+
"गए",
|
|
109
|
+
"गरौं",
|
|
110
|
+
"गर्छ",
|
|
111
|
+
"गर्छु",
|
|
112
|
+
"गर्नुपर्छ",
|
|
113
|
+
"गयौ",
|
|
114
|
+
"गैर",
|
|
115
|
+
"चार",
|
|
116
|
+
"चाहनुहुन्छ",
|
|
117
|
+
"चाहन्छु",
|
|
118
|
+
"चाहिए",
|
|
119
|
+
"छू",
|
|
120
|
+
"जताततै",
|
|
121
|
+
"जब",
|
|
122
|
+
"जबकि",
|
|
123
|
+
"जसको",
|
|
124
|
+
"जसबाट",
|
|
125
|
+
"जसमा",
|
|
126
|
+
"जसलाई",
|
|
127
|
+
"जसले",
|
|
128
|
+
"जस्तै",
|
|
129
|
+
"जस्तो",
|
|
130
|
+
"जस्तोसुकै",
|
|
131
|
+
"जहाँ",
|
|
132
|
+
"जान",
|
|
133
|
+
"जाहिर",
|
|
134
|
+
"जे",
|
|
135
|
+
"जो",
|
|
136
|
+
"ठीक",
|
|
137
|
+
"तत्काल",
|
|
138
|
+
"तदनुसार",
|
|
139
|
+
"तपाईको",
|
|
140
|
+
"तपाई",
|
|
141
|
+
"पर्याप्त",
|
|
142
|
+
"पहिले",
|
|
143
|
+
"पहिलो",
|
|
144
|
+
"पहिल्यै",
|
|
145
|
+
"पाँच",
|
|
146
|
+
"पाँचौं",
|
|
147
|
+
"तल",
|
|
148
|
+
"तापनी",
|
|
149
|
+
"तिनी",
|
|
150
|
+
"तिनीहरू",
|
|
151
|
+
"तिनीहरुको",
|
|
152
|
+
"तिनिहरुलाई",
|
|
153
|
+
"तिमी",
|
|
154
|
+
"तिर",
|
|
155
|
+
"तीन",
|
|
156
|
+
"तुरुन्तै",
|
|
157
|
+
"तेस्रो",
|
|
158
|
+
"तेस्कारण",
|
|
159
|
+
"पूर्व",
|
|
160
|
+
"प्रति",
|
|
161
|
+
"प्रतेक",
|
|
162
|
+
"प्लस",
|
|
163
|
+
"फेरी",
|
|
164
|
+
"बने",
|
|
165
|
+
"त्सपछि",
|
|
166
|
+
"त्सैले",
|
|
167
|
+
"त्यहाँ",
|
|
168
|
+
"थिएन",
|
|
169
|
+
"दिनुभएको",
|
|
170
|
+
"दिनुहुन्छ",
|
|
171
|
+
"दुई",
|
|
172
|
+
"देखि",
|
|
173
|
+
"बरु",
|
|
174
|
+
"बारे",
|
|
175
|
+
"बाहिर",
|
|
176
|
+
"देखिन्छ",
|
|
177
|
+
"देखियो",
|
|
178
|
+
"देखे",
|
|
179
|
+
"देखेको",
|
|
180
|
+
"देखेर",
|
|
181
|
+
"दोस्रो",
|
|
182
|
+
"धेरै",
|
|
183
|
+
"नजिकै",
|
|
184
|
+
"नत्र",
|
|
185
|
+
"नयाँ",
|
|
186
|
+
"निम्ति",
|
|
187
|
+
"बाहेक",
|
|
188
|
+
"बीच",
|
|
189
|
+
"बीचमा",
|
|
190
|
+
"भन",
|
|
191
|
+
"निम्न",
|
|
192
|
+
"निम्नानुसार",
|
|
193
|
+
"निर्दिष्ट",
|
|
194
|
+
"नौ",
|
|
195
|
+
"पक्का",
|
|
196
|
+
"पक्कै",
|
|
197
|
+
"पछि",
|
|
198
|
+
"पछिल्लो",
|
|
199
|
+
"पटक",
|
|
200
|
+
"पर्छ",
|
|
201
|
+
"पर्थ्यो",
|
|
202
|
+
"भन्छन्",
|
|
203
|
+
"भन्",
|
|
204
|
+
"भन्छु",
|
|
205
|
+
"भन्दा",
|
|
206
|
+
"भन्नुभयो",
|
|
207
|
+
"भर",
|
|
208
|
+
"भित्र",
|
|
209
|
+
"भित्री",
|
|
210
|
+
"म",
|
|
211
|
+
"मलाई",
|
|
212
|
+
"मात्र",
|
|
213
|
+
"माथि",
|
|
214
|
+
"मुख्य",
|
|
215
|
+
"मेरो",
|
|
216
|
+
"यति",
|
|
217
|
+
"यथोचित",
|
|
218
|
+
"यदि",
|
|
219
|
+
"यद्यपि",
|
|
220
|
+
"यसको",
|
|
221
|
+
"यसपछि",
|
|
222
|
+
"यसबाहेक",
|
|
223
|
+
"यसरी",
|
|
224
|
+
"यसो",
|
|
225
|
+
"यस्तो",
|
|
226
|
+
"यहाँ",
|
|
227
|
+
"यहाँसम्म",
|
|
228
|
+
"या",
|
|
229
|
+
"रही",
|
|
230
|
+
"राखे",
|
|
231
|
+
"राख्छ",
|
|
232
|
+
"राम्रो",
|
|
233
|
+
"रूप",
|
|
234
|
+
"लगभग",
|
|
235
|
+
"वरीपरी",
|
|
236
|
+
"वास्तवमा",
|
|
237
|
+
"बिरुद्ध",
|
|
238
|
+
"बिशेष",
|
|
239
|
+
"सायद",
|
|
240
|
+
"शायद",
|
|
241
|
+
"संग",
|
|
242
|
+
"संगै",
|
|
243
|
+
"सक्छ",
|
|
244
|
+
"सट्टा",
|
|
245
|
+
"सधै",
|
|
246
|
+
"सबै",
|
|
247
|
+
"सबैलाई",
|
|
248
|
+
"समय",
|
|
249
|
+
"सम्भव",
|
|
250
|
+
"सम्म",
|
|
251
|
+
"सही",
|
|
252
|
+
"साँच्चै",
|
|
253
|
+
"सात",
|
|
254
|
+
"साथ",
|
|
255
|
+
"साथै",
|
|
256
|
+
"सारा",
|
|
257
|
+
"सोही",
|
|
258
|
+
"स्पष्ट",
|
|
259
|
+
"हरे",
|
|
260
|
+
"हरेक",
|
|
261
|
+
]
|
|
262
|
+
end
|
|
263
|
+
end
|
|
264
|
+
end
|
|
265
|
+
end
|