kargaranamir
commited on
Commit
β’
461ed17
1
Parent(s):
483cabe
add batch of langs.
Browse files- languages/bar_Latn.json +1 -1
- languages/brh_Arab.json +19 -0
- languages/glk_Arab.json +30 -0
- languages/hac_Arab.json +19 -0
- languages/khb_Talu.json +19 -0
- languages/mon_Mong.json +3 -3
- languages/sdh_Arab.json +18 -0
languages/bar_Latn.json
CHANGED
@@ -16,7 +16,7 @@
|
|
16 |
},
|
17 |
{
|
18 |
"Site Name": "twitter.com/bayernuhr",
|
19 |
-
"Site URL": "https://twitter.com/bayernuhr",
|
20 |
"Category": "social media",
|
21 |
"Confidence": "π©",
|
22 |
"Info": "glotlid search on http://indigenoustweets.com/bar/",
|
|
|
16 |
},
|
17 |
{
|
18 |
"Site Name": "twitter.com/bayernuhr",
|
19 |
+
"Site URL": "https://twitter.com/bayernuhr/",
|
20 |
"Category": "social media",
|
21 |
"Confidence": "π©",
|
22 |
"Info": "glotlid search on http://indigenoustweets.com/bar/",
|
languages/brh_Arab.json
ADDED
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"Language Name": "Brahui",
|
3 |
+
"Family": "Dravidian",
|
4 |
+
"Subgrouping": "",
|
5 |
+
"Number of Speakers": "2_800_000",
|
6 |
+
"Supported by allenai/MADLAD-400 or facebook/flores": "π₯",
|
7 |
+
"Sites": [
|
8 |
+
{
|
9 |
+
"Site Name": "talarbrahui",
|
10 |
+
"Site URL": "https://talarbrahui.com/",
|
11 |
+
"Category": "news",
|
12 |
+
"Confidence": "π©",
|
13 |
+
"Info": "confirmed by webpage metadata",
|
14 |
+
"Possible Parallel Languages": "",
|
15 |
+
"Links": []
|
16 |
+
}
|
17 |
+
]
|
18 |
+
}
|
19 |
+
|
languages/glk_Arab.json
ADDED
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"Language Name": "Gilaki",
|
3 |
+
"Family": "Indo-European",
|
4 |
+
"Subgrouping": "Iranian",
|
5 |
+
"Number of Speakers": "3_000_000",
|
6 |
+
"Supported by allenai/MADLAD-400 or facebook/flores": "π₯",
|
7 |
+
"Sites": [
|
8 |
+
{
|
9 |
+
"Site Name": "t.me/gilaki_twitter",
|
10 |
+
"Site URL": "https://t.me/gilaki_twitter/",
|
11 |
+
"Category": "social media",
|
12 |
+
"Confidence": "π©",
|
13 |
+
"Info": "collection of gilaki tweets by native speaker",
|
14 |
+
"Possible Parallel Languages": "",
|
15 |
+
"Links": []
|
16 |
+
},
|
17 |
+
{
|
18 |
+
"Site Name": "glk.wikipedia.org",
|
19 |
+
"Site URL": "https://glk.wikipedia.org/wiki/",
|
20 |
+
"Category": "article",
|
21 |
+
"Confidence": "π©",
|
22 |
+
"Info": "confirmed by webpage metadata",
|
23 |
+
"Possible Parallel Languages": "many",
|
24 |
+
"Links": []
|
25 |
+
}
|
26 |
+
]
|
27 |
+
}
|
28 |
+
|
29 |
+
|
30 |
+
|
languages/hac_Arab.json
ADDED
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"Language Name": "Gorani",
|
3 |
+
"Family": "Indo-European",
|
4 |
+
"Subgrouping": "Iranian",
|
5 |
+
"Number of Speakers": "300_000",
|
6 |
+
"Supported by allenai/MADLAD-400 or facebook/flores": "π₯",
|
7 |
+
"Sites": [
|
8 |
+
{
|
9 |
+
"Site Name": "anfsorani.com/ΩΫΨ±Ψ§Ω
Ϋ",
|
10 |
+
"Site URL": "https://anfsorani.com/ΩΫΨ±Ψ§Ω
Ϋ",
|
11 |
+
"Category": "news",
|
12 |
+
"Confidence": "π©",
|
13 |
+
"Info": "confirmed by native speaker as main dialect of Gorani (Hawrami).",
|
14 |
+
"Possible Parallel Languages": "tur_Latn, kmr_Latn, eng_Latn, ckb_Arab, kiu_Latn, arb_Arab, fas_Arab, spa_Latn, deu_Latn, rus_Cyrl",
|
15 |
+
"Links": []
|
16 |
+
}
|
17 |
+
]
|
18 |
+
}
|
19 |
+
|
languages/khb_Talu.json
ADDED
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"Language Name": "Tai Lue",
|
3 |
+
"Family": "Tai-Kadai",
|
4 |
+
"Subgrouping": "Daic",
|
5 |
+
"Number of Speakers": "550_000",
|
6 |
+
"Supported by allenai/MADLAD-400 or facebook/flores": "π₯",
|
7 |
+
"Sites": [
|
8 |
+
{
|
9 |
+
"Site Name": "dw12.com",
|
10 |
+
"Site URL": "https://www.dw12.com/",
|
11 |
+
"Category": "news",
|
12 |
+
"Confidence": "π©",
|
13 |
+
"Info": "confirmed by webpage metadata, glotlid, and glotscript. access with IP address from china.",
|
14 |
+
"Possible Parallel Languages": "",
|
15 |
+
"Links": []
|
16 |
+
}
|
17 |
+
]
|
18 |
+
}
|
19 |
+
|
languages/mon_Mong.json
CHANGED
@@ -7,7 +7,7 @@
|
|
7 |
"Sites": [
|
8 |
{
|
9 |
"Site Name": "wikibilig.mn",
|
10 |
-
"Site URL": "https://wikibilig.mn",
|
11 |
"Category": "article",
|
12 |
"Confidence": "π©",
|
13 |
"Info": "confirmed by webpage metadata, glotlid, and glotscript",
|
@@ -25,7 +25,7 @@
|
|
25 |
},
|
26 |
{
|
27 |
"Site Name": "mongolian.news.cn",
|
28 |
-
"Site URL": "http://mongolian.news.cn",
|
29 |
"Category": "news",
|
30 |
"Confidence": "π©",
|
31 |
"Info": "confirmed by webpage metadata, glotlid, and glotscript",
|
@@ -34,7 +34,7 @@
|
|
34 |
},
|
35 |
{
|
36 |
"Site Name": "mongol.world",
|
37 |
-
"Site URL": "https://mongol.world",
|
38 |
"Category": "news",
|
39 |
"Confidence": "π©",
|
40 |
"Info": "confirmed by webpage metadata, glotlid, and glotscript",
|
|
|
7 |
"Sites": [
|
8 |
{
|
9 |
"Site Name": "wikibilig.mn",
|
10 |
+
"Site URL": "https://wikibilig.mn/",
|
11 |
"Category": "article",
|
12 |
"Confidence": "π©",
|
13 |
"Info": "confirmed by webpage metadata, glotlid, and glotscript",
|
|
|
25 |
},
|
26 |
{
|
27 |
"Site Name": "mongolian.news.cn",
|
28 |
+
"Site URL": "http://mongolian.news.cn/",
|
29 |
"Category": "news",
|
30 |
"Confidence": "π©",
|
31 |
"Info": "confirmed by webpage metadata, glotlid, and glotscript",
|
|
|
34 |
},
|
35 |
{
|
36 |
"Site Name": "mongol.world",
|
37 |
+
"Site URL": "https://mongol.world/",
|
38 |
"Category": "news",
|
39 |
"Confidence": "π©",
|
40 |
"Info": "confirmed by webpage metadata, glotlid, and glotscript",
|
languages/sdh_Arab.json
ADDED
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"Language Name": "Southern-Kurdish (Feyli)",
|
3 |
+
"Family": "Indo-European",
|
4 |
+
"Subgrouping": "Iranian",
|
5 |
+
"Number of Speakers": "1_500_000",
|
6 |
+
"Supported by allenai/MADLAD-400 or facebook/flores": "π₯",
|
7 |
+
"Sites": [
|
8 |
+
{
|
9 |
+
"Site Name": "shafaq.com/ku",
|
10 |
+
"Site URL": "https://shafaq.com/ku/",
|
11 |
+
"Category": "news",
|
12 |
+
"Confidence": "π©",
|
13 |
+
"Info": "confirmed by native speaker as Feyli dialect",
|
14 |
+
"Possible Parallel Languages": "arb_Arab, eng_Latn",
|
15 |
+
"Links": []
|
16 |
+
}
|
17 |
+
]
|
18 |
+
}
|