pere commited on
Commit
df907da
β€’
1 Parent(s): aac2c7b
Files changed (33) hide show
  1. langid/langid_datafilesfin_dev.tsv β†’ correct/correct_datafiles/correct_dev.tsv +2 -2
  2. langid/langid_datafilesdan_dev.tsv β†’ correct/correct_datafiles/correct_test.tsv +2 -2
  3. langid/langid_datafileseng_dev.tsv β†’ correct/correct_datafiles/correct_train.tsv +2 -2
  4. correct/create_correct_dataset.py +140 -0
  5. create_dataset_from_tsv.py β†’ create_training_files.py +1 -1
  6. langid/create_langid_trainset.py +1 -1
  7. langid/langid_datafilesfao_dev.tsv +0 -3
  8. langid/langid_datafilesisl_dev.tsv +0 -3
  9. langid/langid_datafilesnno_dev.tsv +0 -3
  10. langid/langid_datafilesnob_dev.tsv +0 -3
  11. langid/langid_datafilessma_dev.tsv +0 -3
  12. langid/langid_datafilessme_dev.tsv +0 -3
  13. langid/langid_datafilessmj_dev.tsv +0 -3
  14. langid/langid_datafilessmn_dev.tsv +0 -3
  15. langid/langid_datafilessms_dev.tsv +0 -3
  16. langid/langid_datafilesswe_dev.tsv +0 -3
  17. {tsv_source_target β†’ translate_dataset}/finnob.tsv +0 -0
  18. {tsv_source_target β†’ translate_dataset}/finsma.tsv +0 -0
  19. {tsv_source_target β†’ translate_dataset}/finsme.tsv +0 -0
  20. {tsv_source_target β†’ translate_dataset}/finsmn.tsv +0 -0
  21. {tsv_source_target β†’ translate_dataset}/finsms.tsv +0 -0
  22. {tsv_source_target β†’ translate_dataset}/nobnyn.tsv +0 -0
  23. {tsv_source_target β†’ translate_dataset}/nobsma.tsv +0 -0
  24. {tsv_source_target β†’ translate_dataset}/nobsme.tsv +0 -0
  25. {tsv_source_target β†’ translate_dataset}/nobsmj.tsv +0 -0
  26. {tsv_source_target β†’ translate_dataset}/smanob.tsv +0 -0
  27. {tsv_source_target β†’ translate_dataset}/smenob.tsv +0 -0
  28. {tsv_source_target β†’ translate_dataset}/smesma.tsv +0 -0
  29. {tsv_source_target β†’ translate_dataset}/smesmj.tsv +0 -0
  30. {tsv_source_target β†’ translate_dataset}/smesmn.tsv +0 -0
  31. tsv_all_source_target/dev.tsv +0 -3
  32. tsv_all_source_target/test.tsv +0 -3
  33. tsv_all_source_target/train.tsv +0 -3
langid/langid_datafilesfin_dev.tsv β†’ correct/correct_datafiles/correct_dev.tsv RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:73207d81fc90e7b4933bfd6cf70fbf75dfdecf4fc97eebfd1eb8288bfa440239
3
- size 111790
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fec0f2575d5925b993aefdb9715ace66f3a4948258c8cfa1600fd9b264fd8963
3
+ size 2320597
langid/langid_datafilesdan_dev.tsv β†’ correct/correct_datafiles/correct_test.tsv RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2e9c2fa5c5adb1092ac3c48b87753a6a63977d2a92878c7747f0612b801717f9
3
- size 117358
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5e7595989509c671e3bc088129a080ab4292ba472ef21bc8c92d470ebc4f1196
3
+ size 2339422
langid/langid_datafileseng_dev.tsv β†’ correct/correct_datafiles/correct_train.tsv RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0308930eea2afe03b4bb8cdce1c362baff25aebd5575d3277d3165bc3aca6e19
3
- size 88862
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d42cbf75385b1918f7dd9fa479e65519ecbe9bfcf1996132c4236ea695cd9fdf
3
+ size 18565347
correct/create_correct_dataset.py ADDED
@@ -0,0 +1,140 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ from datasets import load_dataset
2
+ import random
3
+ import argparse
4
+ import csv
5
+ import glob
6
+ import pandas as pd
7
+ from sklearn.model_selection import train_test_split
8
+
9
+
10
+ def main(args):
11
+ trans_chars = ",.:;!"
12
+
13
+ filelist = glob.glob('../langid/sentences/*.txt')
14
+
15
+ data = pd.DataFrame()
16
+
17
+ for tsvfile in filelist:
18
+ print(f"Processing {tsvfile}")
19
+ tmp = pd.read_csv(tsvfile, sep='\t',on_bad_lines='skip',engine='python',encoding='utf8')
20
+ if len(tmp.columns)==1:
21
+ tmp.insert(0,'id','')
22
+
23
+ tmp.columns=['id','source']
24
+ data=pd.concat([data,tmp])
25
+
26
+
27
+ # Trim
28
+ data['source'] = data['source'].str.strip()
29
+
30
+ # Drop rows that does not end with punctation
31
+ data = data[data['source'].str[-1:].isin([".",",",":",";","!"])]
32
+
33
+ # For not creating chaos later with . . . Just remove examples with elipsis
34
+ data = data[~data['source'].str.contains("...", regex=False)]
35
+ data = data[~data['source'].str.contains(". . .", regex=False)]
36
+
37
+ #Drop the id
38
+ data = data.drop(['id'],axis=1)
39
+
40
+ # Duplicate column
41
+ data['target'] = data['source']
42
+
43
+ # Apply each method to 10% of the corpus
44
+ # set a stop
45
+ stop =int(len(data)/10)
46
+ stop_small = int(stop/2)
47
+
48
+ #Main shuffling
49
+ data = data.sample(frac=1).reset_index(drop=True)
50
+
51
+ # Lowercase in 10% of the cases
52
+ data.loc[:stop,'source'] = data['source'].str.lower()
53
+ print(f"Lower casing 10% - '{data.loc[0]['target']}' -> '{data.loc[0]['source']}'")
54
+ data = data.sample(frac=1).reset_index(drop=True)
55
+
56
+ # Uppercase in 5% of the cases
57
+ data.loc[:stop_small,'source'] = data['source'].str.upper()
58
+ print(f"Upper casing 5% - '{data.loc[0]['target']}' -> '{data.loc[0]['source']}'")
59
+ data = data.sample(frac=1).reset_index(drop=True)
60
+
61
+
62
+ # Remove all spaces in 10% of the cases
63
+ data.loc[:stop,'source'] = data['source'].str.replace(" ","")
64
+ print(f"Removing space 10% - '{data.loc[0]['target']}' -> '{data.loc[0]['source']}'")
65
+ data = data.sample(frac=1).reset_index(drop=True)
66
+
67
+ # Remove both spaces and do lowercasing in 10%
68
+ data.loc[:stop,'source'] = data['source'].str.replace(" ","")
69
+ data.loc[:stop,'source'] = data['source'].str.lower()
70
+ print(f"Removing space and doing lovercase 10% - '{data.loc[0]['target']}' -> '{data.loc[0]['source']}'")
71
+ data = data.sample(frac=1).reset_index(drop=True)
72
+
73
+ # Remove a random number of spaces in 10% of the cases
74
+ for index, row in data[0:stop].iterrows():
75
+ source = row['source']
76
+ #Find the spaces
77
+ spacepos = [pos for pos, char in enumerate(source) if char == " "]
78
+ random.shuffle(spacepos)
79
+
80
+ #Reduce to a random number
81
+ spacepos = spacepos[0:random.randint(0,len(spacepos))]
82
+
83
+ ##Sort in reverse order
84
+ spacepos.sort(reverse=True)
85
+
86
+ ##Loop and replace
87
+ for s in spacepos:
88
+ source = source[:s] + source[s+1:]
89
+
90
+ data.loc[index,'source'] = source
91
+
92
+ print(f"Removing a random number of spaces 10% - '{data.loc[0]['target']}' -> '{data.loc[0]['source']}'")
93
+ data = data.sample(frac=1).reset_index(drop=True)
94
+
95
+ # Remove all punctation in 10% of the cases
96
+ trans_table = source.maketrans("", "", trans_chars)
97
+ data.loc[:stop,'source'] = data['source'].str.translate(trans_table)
98
+ print(f"Removing punctation 10% - '{data.loc[0]['target']}' -> '{data.loc[0]['source']}'")
99
+ data = data.sample(frac=1).reset_index(drop=True)
100
+
101
+ # Remove a random number of commas in 10% of the cases
102
+ for index, row in data[0:stop].iterrows():
103
+ source = row['source']
104
+ #Find the spaces
105
+ spacepos = [pos for pos, char in enumerate(source) if char == ", "]
106
+ random.shuffle(spacepos)
107
+
108
+ #Reduce to a random number
109
+ spacepos = spacepos[0:random.randint(0,len(spacepos))]
110
+
111
+ ##Sort in reverse order
112
+ spacepos.sort(reverse=True)
113
+
114
+ ##Loop and replace
115
+ for s in spacepos:
116
+ source = source[:s] + " " +source[s+1:]
117
+
118
+ data.loc[index,'source'] = source
119
+
120
+ print(f"Removing a random number of commas 10% - '{data.loc[0]['target']}' -> '{data.loc[0]['source']}'")
121
+ data = data.sample(frac=1).reset_index(drop=True)
122
+
123
+ # Train - test - dev
124
+ train, test = train_test_split(data, test_size=0.2)
125
+ test, dev = train_test_split(test, test_size=0.5)
126
+
127
+ # Write the datasets to disk
128
+ train.to_csv('correct_datafiles/correct_train.tsv', index=False, header=False, sep='\t')
129
+ test.to_csv('correct_datafiles/correct_test.tsv', index=False, header=False, sep='\t')
130
+ dev.to_csv('correct_datafiles/correct_dev.tsv', index=False, header=False, sep='\t')
131
+
132
+ def parse_args():
133
+ # Parse commandline
134
+ parser = argparse.ArgumentParser()
135
+ args = parser.parse_args()
136
+ return args
137
+
138
+ if __name__ == "__main__":
139
+ args = parse_args()
140
+ main(args)
create_dataset_from_tsv.py β†’ create_training_files.py RENAMED
@@ -5,7 +5,7 @@ import glob
5
  import sys
6
  import os
7
 
8
- filelist = glob.glob('tsv_source_target/*.tsv')
9
 
10
  data = pd.DataFrame()
11
 
 
5
  import sys
6
  import os
7
 
8
+ filelist = glob.glob('translate_dataset/*.tsv')
9
 
10
  data = pd.DataFrame()
11
 
langid/create_langid_trainset.py CHANGED
@@ -32,7 +32,7 @@ for tsvfile in filelist:
32
  # Write the datasets to disk
33
  train.to_csv('langid_datafiles/'+lang+'_train.tsv', index=False, header=False, sep='\t')
34
  test.to_csv('langid_datafiles/'+lang+'_test.tsv', index=False, header=False, sep='\t')
35
- dev.to_csv('langid_datafiles'+lang+'_dev.tsv', index=False, header=False, sep='\t')
36
 
37
 
38
  print("Finished")
 
32
  # Write the datasets to disk
33
  train.to_csv('langid_datafiles/'+lang+'_train.tsv', index=False, header=False, sep='\t')
34
  test.to_csv('langid_datafiles/'+lang+'_test.tsv', index=False, header=False, sep='\t')
35
+ dev.to_csv('langid_datafiles/'+lang+'_dev.tsv', index=False, header=False, sep='\t')
36
 
37
 
38
  print("Finished")
langid/langid_datafilesfao_dev.tsv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:9e9862ecdd770235ee1e6e1389cd13c941c1fc67e347206addb5ceef8d8303ce
3
- size 104101
 
 
 
 
langid/langid_datafilesisl_dev.tsv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:431256015e7ecd8c7db47eb24de3030a9f0dfafcb6e473ad62f961ecb1ab02b2
3
- size 116070
 
 
 
 
langid/langid_datafilesnno_dev.tsv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:b98781d8bfcf20efab1fc3ad723cc72a2edcb825b77ac87ac0ba55d71b395ea8
3
- size 105195
 
 
 
 
langid/langid_datafilesnob_dev.tsv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:40ec96b3a9056efe07930ffaa869289c443d7181878e2b30fb40bf635bb0da88
3
- size 123710
 
 
 
 
langid/langid_datafilessma_dev.tsv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:d401a7e7325abde28f0616f5da413a776cbf0cf71acd472d0da429dc6dd54868
3
- size 113845
 
 
 
 
langid/langid_datafilessme_dev.tsv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:978a1bb983d7def98d552fdf0b2456b70cae30b350ecbcbbeb0711b95c0cb322
3
- size 97202
 
 
 
 
langid/langid_datafilessmj_dev.tsv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:726f6ba67de111833253522028cd84c2a1d725ed627b62e35827875aa5889b38
3
- size 107928
 
 
 
 
langid/langid_datafilessmn_dev.tsv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:fd8f7b55bf174a1a442f00e6e7ea5ba6af5378cd6bf01209ab195acc6a06a21f
3
- size 111870
 
 
 
 
langid/langid_datafilessms_dev.tsv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:19960f2c00c8058a2b4eb96d8639ee9fe08a1dbe99ae41ab6a6809aeafdc9e39
3
- size 70369
 
 
 
 
langid/langid_datafilesswe_dev.tsv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:80dea5798eef38ae4e84010f6c21581ae0a229a03cd476647e5497473e44a750
3
- size 100161
 
 
 
 
{tsv_source_target β†’ translate_dataset}/finnob.tsv RENAMED
File without changes
{tsv_source_target β†’ translate_dataset}/finsma.tsv RENAMED
File without changes
{tsv_source_target β†’ translate_dataset}/finsme.tsv RENAMED
File without changes
{tsv_source_target β†’ translate_dataset}/finsmn.tsv RENAMED
File without changes
{tsv_source_target β†’ translate_dataset}/finsms.tsv RENAMED
File without changes
{tsv_source_target β†’ translate_dataset}/nobnyn.tsv RENAMED
File without changes
{tsv_source_target β†’ translate_dataset}/nobsma.tsv RENAMED
File without changes
{tsv_source_target β†’ translate_dataset}/nobsme.tsv RENAMED
File without changes
{tsv_source_target β†’ translate_dataset}/nobsmj.tsv RENAMED
File without changes
{tsv_source_target β†’ translate_dataset}/smanob.tsv RENAMED
File without changes
{tsv_source_target β†’ translate_dataset}/smenob.tsv RENAMED
File without changes
{tsv_source_target β†’ translate_dataset}/smesma.tsv RENAMED
File without changes
{tsv_source_target β†’ translate_dataset}/smesmj.tsv RENAMED
File without changes
{tsv_source_target β†’ translate_dataset}/smesmn.tsv RENAMED
File without changes
tsv_all_source_target/dev.tsv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:f5de7e7fc05d9697ff9863f4846364878492e847d236c33d4e55495c66b3c2ae
3
- size 25212526
 
 
 
 
tsv_all_source_target/test.tsv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:53096012b9e502826ee0028367316aa624d6a1b847eb255f173b09d9475af4ce
3
- size 25054929
 
 
 
 
tsv_all_source_target/train.tsv DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:1f28b52818b79814f12c32e2ddf11f434e52439497a11bb8f251a89d682ac91e
3
- size 201104421