QubitPi commited on
Commit
c23c545
1 Parent(s): 197443a

Support Sanskrit

Browse files
Files changed (3) hide show
  1. README.md +21 -2
  2. ontology.png +0 -0
  3. wiktionary/wiktextract/extract.py +6 -1
README.md CHANGED
@@ -9,6 +9,7 @@ language:
9
  - ko
10
  - peo
11
  - akk
 
12
  configs:
13
  - config_name: Languages
14
  data_files:
@@ -24,22 +25,27 @@ configs:
24
  path: old-persian-wiktextract-data.jsonl
25
  - split: Akkadian
26
  path: akkadian-wiktextract-data.jsonl
 
 
27
  - config_name: Graph
28
  data_files:
29
  - split: AllLanguage
30
  path: word-definition-graph-data.jsonl
31
  tags:
 
 
32
  - Wiktionary
 
33
  - German
34
  - Latin
35
  - Ancient Greek
36
  - Korean
37
  - Old Persian
38
  - Akkadian
39
- - Vocabulary
40
  - Knowledge Graph
41
  size_categories:
42
- - 1M<n<10M
43
  ---
44
 
45
  Wiktionary Data on Hugging Face Datasets
@@ -61,6 +67,7 @@ supports the following languages:
61
  - __한국어__ - Korean
62
  - __𐎠𐎼𐎹__ - [Old Persian](https://en.wikipedia.org/wiki/Old_Persian_cuneiform)
63
  - __𒀝𒅗𒁺𒌑(𒌝)__ - [Akkadian](https://en.wikipedia.org/wiki/Akkadian_language)
 
64
 
65
  [wiktionary-data]() was originally a sub-module of [wilhelm-graphdb](https://github.com/QubitPi/wilhelm-graphdb). While
66
  the dataset it's getting bigger, I noticed a wave of more exciting potentials this dataset can bring about that
@@ -84,11 +91,23 @@ There are __two__ data subsets:
84
  - `Korean`
85
  - `OldPersian`
86
  - `Akkadian`
 
87
 
88
  2. __Graph__ subset that is useful for constructing knowledge graphs:
89
 
90
  - `AllLanguage`: all the languages in a giant graph
91
 
 
 
 
 
 
 
 
 
 
 
 
92
  Development
93
  -----------
94
 
 
9
  - ko
10
  - peo
11
  - akk
12
+ - sa
13
  configs:
14
  - config_name: Languages
15
  data_files:
 
25
  path: old-persian-wiktextract-data.jsonl
26
  - split: Akkadian
27
  path: akkadian-wiktextract-data.jsonl
28
+ - split: Sanskrit
29
+ path: sanskrit-wiktextract-data.jsonl
30
  - config_name: Graph
31
  data_files:
32
  - split: AllLanguage
33
  path: word-definition-graph-data.jsonl
34
  tags:
35
+ - Natural Language Processing
36
+ - NLP
37
  - Wiktionary
38
+ - Vocabulary
39
  - German
40
  - Latin
41
  - Ancient Greek
42
  - Korean
43
  - Old Persian
44
  - Akkadian
45
+ - Sanskrit
46
  - Knowledge Graph
47
  size_categories:
48
+ - 100M<n<1B
49
  ---
50
 
51
  Wiktionary Data on Hugging Face Datasets
 
67
  - __한국어__ - Korean
68
  - __𐎠𐎼𐎹__ - [Old Persian](https://en.wikipedia.org/wiki/Old_Persian_cuneiform)
69
  - __𒀝𒅗𒁺𒌑(𒌝)__ - [Akkadian](https://en.wikipedia.org/wiki/Akkadian_language)
70
+ - __संस्कृतम्__ - Sanskrit, or Classical Sanskrit
71
 
72
  [wiktionary-data]() was originally a sub-module of [wilhelm-graphdb](https://github.com/QubitPi/wilhelm-graphdb). While
73
  the dataset it's getting bigger, I noticed a wave of more exciting potentials this dataset can bring about that
 
91
  - `Korean`
92
  - `OldPersian`
93
  - `Akkadian`
94
+ - `Sanskrit`
95
 
96
  2. __Graph__ subset that is useful for constructing knowledge graphs:
97
 
98
  - `AllLanguage`: all the languages in a giant graph
99
 
100
+ The _Graph_ data ontology is the following:
101
+
102
+ <div align="center">
103
+ <img src="ontology.png" size="50%" alt="Error loading ontology.png"/>
104
+ </div>
105
+
106
+ > [!TIP]
107
+ >
108
+ > Two words are structurally similar if and only if the two shares the same\
109
+ > [stem](https://en.wikipedia.org/wiki/Word_stem)
110
+
111
  Development
112
  -----------
113
 
ontology.png ADDED
wiktionary/wiktextract/extract.py CHANGED
@@ -44,7 +44,8 @@ def extract_data(wiktextract_data_path: str):
44
  open("ancient-greek-wiktextract-data.jsonl", "w") as ancient_greek,
45
  open("korean-wiktextract-data.jsonl", "w") as korean,
46
  open("old-persian-wiktextract-data.jsonl", "w") as old_persian,
47
- open("akkadian-wiktextract-data.jsonl", "w") as akkadian
 
48
  ):
49
  for line in data:
50
  vocabulary = json.loads(line)
@@ -81,6 +82,10 @@ def extract_data(wiktextract_data_path: str):
81
  if vocabulary["lang"] == "Akkadian":
82
  akkadian.write(json.dumps({"term": term, "part of speech": pos, "definitions": definitions, "audios": audios}))
83
  akkadian.write("\n")
 
 
 
 
84
 
85
  def extract_graph(wiktextract_data_path: str):
86
  import json
 
44
  open("ancient-greek-wiktextract-data.jsonl", "w") as ancient_greek,
45
  open("korean-wiktextract-data.jsonl", "w") as korean,
46
  open("old-persian-wiktextract-data.jsonl", "w") as old_persian,
47
+ open("akkadian-wiktextract-data.jsonl", "w") as akkadian,
48
+ open("sanskrit-wiktextract-data.jsonl", "w") as sanskrit
49
  ):
50
  for line in data:
51
  vocabulary = json.loads(line)
 
82
  if vocabulary["lang"] == "Akkadian":
83
  akkadian.write(json.dumps({"term": term, "part of speech": pos, "definitions": definitions, "audios": audios}))
84
  akkadian.write("\n")
85
+ if vocabulary["lang"] == "Sanskrit":
86
+ sanskrit.write(json.dumps({"term": term, "part of speech": pos, "definitions": definitions, "audios": audios}))
87
+ sanskrit.write("\n")
88
+
89
 
90
  def extract_graph(wiktextract_data_path: str):
91
  import json