Datasets:

Modalities:
Text
Formats:
json
ArXiv:
License:
File size: 3,842 Bytes
437e44a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
import os
from datasets import load_dataset
import matplotlib.pyplot as plt
from collections import Counter, defaultdict

# Load dataset
dataset = load_dataset(
    'json', 
    data_files='files/*.jsonl.gz', 
    split='train', 
    encoding='utf-8'
)

# Print dataset size
print(len(dataset)) # 5820634

# Create images directory if it doesn't exist
os.makedirs("images", exist_ok=True)

# Count languages
cnt = Counter()
for row in dataset:
    cnt[row["lang"]] += 1

# Plot overall language distribution
langs, freqs = zip(*cnt.items())

plt.figure(figsize=(10, 4))
plt.bar(langs, freqs)
plt.xticks(rotation=90)
plt.ylabel("# documents")
plt.title("Number of documents per language")
plt.tight_layout()
plt.savefig("images/nb_documents.png")
plt.close()

# Split into two groups
threshold = 10000
high = {k: v for k, v in cnt.items() if v > threshold}
low = {k: v for k, v in cnt.items() if v <= threshold}

# Create a figure with 2 subplots (vertical)
fig, axes = plt.subplots(nrows=2, ncols=1, figsize=(12, 8))

# High-frequency languages
if high:
    langs, freqs = zip(*high.items())
    axes[0].bar(langs, freqs)
    axes[0].set_xticklabels(langs, rotation=90)
    axes[0].set_ylabel("# documents")
    axes[0].set_title("Languages with more than 10,000 documents")

# Low-frequency languages
if low:
    langs, freqs = zip(*low.items())
    axes[1].bar(langs, freqs)
    axes[1].set_xticklabels(langs, rotation=90)
    axes[1].set_ylabel("# documents")
    axes[1].set_title("Languages with 10,000 or fewer documents")

plt.tight_layout()
plt.savefig("images/nb_documents_combined.png")
plt.close()

# Analyze text lengths for low-frequency languages
low_languages = set(low.keys())

text_lengths_low = defaultdict(list)

# Iterate through dataset efficiently
for row in dataset:
    lang = row["lang"]
    if lang in low_languages:
        text_len = len(row["text"]) if row["text"] else 0
        text_lengths_low[lang].append(text_len)

# Prepare data for boxplot
langs = list(text_lengths_low.keys())
data = [text_lengths_low[lang] for lang in langs]

# Plot
plt.figure(figsize=(12, 6))
plt.boxplot(data, labels=langs, showfliers=False)  # hide extreme outliers for clarity
plt.xticks(rotation=90)
plt.ylabel("Document length")
plt.title("Document length per language")
plt.tight_layout()
plt.savefig("images/boxplot_low.png")
plt.close()

# Analyze text lengths for low-frequency languages
high_languages = set(high.keys())

text_lengths_high = defaultdict(list)

# Iterate through dataset efficiently
for row in dataset:
    lang = row["lang"]
    if lang in high_languages:
        text_len = len(row["text"]) if row["text"] else 0
        text_lengths_high[lang].append(text_len)

# Prepare data for boxplot
langs = list(text_lengths_high.keys())
data = [text_lengths_high[lang] for lang in langs]

# Plot
plt.figure(figsize=(12, 6))
plt.boxplot(data, labels=langs, showfliers=False)  # hide extreme outliers for clarity
plt.xticks(rotation=90)
plt.ylabel("Document length")
plt.title("Document length per language")
plt.tight_layout()
plt.savefig("images/boxplot_high.png")
plt.close()

overall_max_high = max(max(lengths) for lengths in text_lengths_high.values())
overall_max_low = max(max(lengths) for lengths in text_lengths_low.values())
print("Maximum text length overall:", max(overall_max_high, overall_max_low )) # 99,996,139

# Collect all unique eurovoc concepts and ids
all_concepts = set()
all_concept_ids = set()

for row in dataset:
    # Add all eurovoc concepts and ids from this row
    concepts = row.get("eurovoc_concepts", [])
    concept_ids = row.get("eurovoc_concepts_ids", [])
    all_concepts.update(concepts)
    all_concept_ids.update(concept_ids)

print("Number of unique eurovoc_concepts:", len(all_concepts)) # 7097
print("Number of unique eurovoc_concepts_ids:", len(all_concept_ids)) # 7049