File size: 4,067 Bytes
2cddd11 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 |
import sys, os
# from normalizers.english import EnglishTextNormalizer
error_words_freqs = {}
infile = sys.argv[1]
# setname = sys.argv[2]
insert_error = 0
insert_rare = 0
freqlist_test = {}
# eng_norm = EnglishTextNormalizer()
freqlist = {}
with open("./blist/word_freq.txt") as fin:
for line in fin:
word, freq = line.split()
freqlist[word.upper()] = int(freq)
with open("./blist/all_rare_words.txt") as fin:
rareset = set()
for line in fin:
rareset.add(line.strip().upper())
project_set = set()
with open(infile) as fin:
lines = fin.readlines()
for i, line in enumerate(lines):
if line.startswith('id:'):
project = line.strip(')\n').split('-')[-3:]
project = '-'.join(project)
if "REF:" in line:
nextline = lines[i+1].split()
for j, word in enumerate(line.split()):
if '*' in word:
insert_error += 1
if nextline[j].upper() in rareset:
insert_rare += 1
line = line.replace('*', '')
line.replace('%BCACK', '')
for word in line.split()[1:]:
if not word.startswith('('):
if word.upper() not in freqlist_test:
freqlist_test[word.upper()] = 1
else:
freqlist_test[word.upper()] += 1
if word != word.lower() and word.upper() in error_words_freqs:
error_words_freqs[word.upper()] += 1
elif word != word.lower() and word.upper() not in error_words_freqs:
error_words_freqs[word.upper()] = 1
elif word == word.lower() and word.upper() not in error_words_freqs:
error_words_freqs[word.upper()] = 0
print(len(error_words_freqs.keys()))
print(insert_rare)
commonwords = []
rarewords = []
oovwords = []
common_freq = 0
rare_freq = 0
oov_freq = 0
common_error = 0
rare_error = 0
oov_error = 0
partial_error = 0
partial_freq = 0
very_common_error = 0
very_common_words = 0
words_error_freq = {}
words_total_freq = {}
for word, error in error_words_freqs.items():
if word in rareset:
rarewords.append(word)
rare_freq += freqlist_test[word]
rare_error += error
elif word not in freqlist:
oovwords.append(word)
oov_freq += freqlist_test[word] if word in freqlist_test else 1
oov_error += error
else:
if freqlist[word] <= 10 and freqlist[word] >= 3:
if freqlist[word] not in words_error_freq:
words_error_freq[freqlist[word]] = error
words_total_freq[freqlist[word]] = freqlist_test[word]
else:
words_error_freq[freqlist[word]] += error
words_total_freq[freqlist[word]] += freqlist_test[word]
if freqlist[word] <= 10 and freqlist[word] >= 3:
very_common_error += error
very_common_words += freqlist_test[word]
commonwords.append(word)
common_freq += freqlist_test[word]
common_error += error
total_words = common_freq + rare_freq + oov_freq
total_errors = common_error+rare_error+oov_error + insert_error
WER = total_errors / total_words
print('='*89)
print('Common words error freq: {} / {} = {}'.format(common_error, common_freq, common_error/common_freq))
print('Rare words error freq: {} / {} = {}'.format(rare_error+insert_rare, rare_freq, (rare_error + insert_rare)/rare_freq))
print('OOV words error freq: {} / {} = {}'.format(oov_error, oov_freq, oov_error/max(oov_freq, 1)))
print('WER estimate: {} / {} = {}'.format(total_errors, total_words, WER))
# print('Partial word count: {} / {}'.format(partial_error, partial_freq))
print('Insert error: {} / {} = {}'.format(insert_error - insert_rare, total_words, (insert_error - insert_rare)/total_words))
print('Insertion + OOV error {}'.format((insert_error + oov_error - insert_rare) / total_words))
# print('Very common words error freq: {} / {} = {}'.format(very_common_error, very_common_words, very_common_error/very_common_words))
print('='*89) |