kargaranamir commited on
Commit
a89c8a9
β€’
1 Parent(s): f0f8e62

add sma, sme, smj.

Browse files
languages/sma_Latn.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Language Name": "South Saami",
3
+ "Family": "Uralic",
4
+ "Subgrouping": "Saami",
5
+ "Number of Speakers": "600",
6
+ "Supported by allenai/MADLAD-400": 0,
7
+ "Supported by facebook/flores": 0,
8
+ "Supported by cis-lmu/Glot500": 0,
9
+ "Sites": [
10
+ {
11
+ "Site Name": "ovttas.no",
12
+ "Site URL": "https://ovttas.no/sma/",
13
+ "Category": "education",
14
+ "Confidence": "🟩",
15
+ "Info": "confirmed by webpage metadata and glotlid",
16
+ "Possible Parallel Languages": "smj_Latn, sme_Latn, nob_Latn",
17
+ "Links": []
18
+ },
19
+ {
20
+ "Site Name": "sametinget.se",
21
+ "Site URL": "https://www.sametinget.se/lang/sma",
22
+ "Category": "parliament news",
23
+ "Confidence": "🟩",
24
+ "Info": "confirmed by webpage metadata and glotlid",
25
+ "Possible Parallel Languages": "sme_Latn, smj_Latn, swe_Latn, eng_Latn",
26
+ "Links": []
27
+ }
28
+ ]
29
+ }
30
+
languages/sme_Latn.json ADDED
@@ -0,0 +1,48 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Language Name": "North Saami",
3
+ "Family": "Uralic",
4
+ "Subgrouping": "Saami",
5
+ "Number of Speakers": "25_000",
6
+ "Supported by allenai/MADLAD-400": 0,
7
+ "Supported by facebook/flores": 0,
8
+ "Supported by cis-lmu/Glot500": 1,
9
+ "Sites": [
10
+ {
11
+ "Site Name": "ovttas.no",
12
+ "Site URL": "https://ovttas.no/",
13
+ "Category": "education",
14
+ "Confidence": "🟩",
15
+ "Info": "confirmed by webpage metadata and glotlid",
16
+ "Possible Parallel Languages": "sma_Latn, smj_Latn, nob_Latn",
17
+ "Links": []
18
+ },
19
+ {
20
+ "Site Name": "regjeringen.no",
21
+ "Site URL": "https://regjeringen.no/se",
22
+ "Category": "government",
23
+ "Confidence": "🟩",
24
+ "Info": "confirmed by webpage metadata and glotlid",
25
+ "Possible Parallel Languages": "eng_Latn, nno_Latn",
26
+ "Links": []
27
+ },
28
+ {
29
+ "Site Name": "sametinget.se",
30
+ "Site URL": "https://www.sametinget.se/lang/sme",
31
+ "Category": "parliament news",
32
+ "Confidence": "🟩",
33
+ "Info": "confirmed by webpage metadata and glotlid",
34
+ "Possible Parallel Languages": "sma_Latn, smj_Latn, swe_Latn, eng_Latn",
35
+ "Links": []
36
+ },
37
+ {
38
+ "Site Name": "yle.fi",
39
+ "Site URL": "https://yle.fi/sapmi",
40
+ "Category": "news",
41
+ "Confidence": "🟩",
42
+ "Info": "confirmed by webpage metadata and glotlid",
43
+ "Possible Parallel Languages": "fin_Latn, sms_Latn, smn_Latn",
44
+ "Links": []
45
+ }
46
+ ]
47
+ }
48
+
languages/smj_Latn.json ADDED
@@ -0,0 +1,39 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Language Name": "Lule Sami",
3
+ "Family": "Uralic",
4
+ "Subgrouping": "Saami",
5
+ "Number of Speakers": "650",
6
+ "Supported by allenai/MADLAD-400": 0,
7
+ "Supported by facebook/flores": 0,
8
+ "Supported by cis-lmu/Glot500": 0,
9
+ "Sites": [
10
+ {
11
+ "Site Name": "ovttas.no",
12
+ "Site URL": "https://ovttas.no/smj/",
13
+ "Category": "education",
14
+ "Confidence": "🟩",
15
+ "Info": "confirmed by webpage metadata and glotlid",
16
+ "Possible Parallel Languages": "sma_Latn, sme_Latn, nob_Latn",
17
+ "Links": []
18
+ },
19
+ {
20
+ "Site Name": "sametinget.se",
21
+ "Site URL": "https://www.sametinget.se/lang/smj",
22
+ "Category": "parliament news",
23
+ "Confidence": "🟩",
24
+ "Info": "confirmed by webpage metadata and glotlid",
25
+ "Possible Parallel Languages": "sme_Latn, sma_Latn, swe_Latn, eng_Latn",
26
+ "Links": []
27
+ },
28
+ {
29
+ "Site Name": "misc",
30
+ "Site URL": "misc",
31
+ "Category": "misc",
32
+ "Confidence": "🟩",
33
+ "Info": "confirmed by glotlid and webpage metadata",
34
+ "Possible Parallel Languages": "",
35
+ "Links": ["https://www.skolverket.se/download/18.17b5ae11882bf5c56c368/1685367311902/pdf11570.pdf"]
36
+ }
37
+ ]
38
+ }
39
+
languages/smn_Latn.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Language Name": "Inari Saami",
3
+ "Family": "Uralic",
4
+ "Subgrouping": "Saami",
5
+ "Number of Speakers": "400",
6
+ "Supported by allenai/MADLAD-400": 0,
7
+ "Supported by facebook/flores": 0,
8
+ "Supported by cis-lmu/Glot500": 0,
9
+ "Sites": [
10
+ {
11
+ "Site Name": "yle.fi",
12
+ "Site URL": "https://yle.fi/sapmi/saami",
13
+ "Category": "news",
14
+ "Confidence": "🟩",
15
+ "Info": "confirmed by webpage metadata and glotlid",
16
+ "Possible Parallel Languages": "fin_Latn, sms_Latn, sme_Latn",
17
+ "Links": []
18
+ }
19
+ ]
20
+ }
21
+
languages/sms_Latn.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "Language Name": "Skolt Saami",
3
+ "Family": "Uralic",
4
+ "Subgrouping": "Saami",
5
+ "Number of Speakers": "330",
6
+ "Supported by allenai/MADLAD-400": 0,
7
+ "Supported by facebook/flores": 0,
8
+ "Supported by cis-lmu/Glot500": 0,
9
+ "Sites": [
10
+ {
11
+ "Site Name": "yle.fi",
12
+ "Site URL": "https://yle.fi/sapmi/saamm",
13
+ "Category": "news",
14
+ "Confidence": "🟩",
15
+ "Info": "confirmed by webpage metadata and glotlid",
16
+ "Possible Parallel Languages": "fin_Latn, smn_Latn, sme_Latn",
17
+ "Links": []
18
+ }
19
+ ]
20
+ }
21
+