File size: 4,518 Bytes
cdf0345
5fe24c6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
cdf0345
 
 
 
 
 
 
 
 
5fe24c6
 
 
cdf0345
5fe24c6
cdf0345
 
5fe24c6
cdf0345
5fe24c6
cdf0345
5fe24c6
cdf0345
5fe24c6
cdf0345
5fe24c6
 
cdf0345
5fe24c6
cdf0345
5fe24c6
cdf0345
5fe24c6
cdf0345
5fe24c6
cdf0345
5fe24c6
 
 
cdf0345
 
5fe24c6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
cdf0345
 
 
 
 
5fe24c6
cdf0345
 
 
 
5fe24c6
cdf0345
5fe24c6
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
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
---
dataset_info:
  features:
    - name: thread_no
      dtype: int64
    - name: archived_on
      dtype: int64
    - name: semantic_url
      dtype: string
    - name: "no"
      dtype: int64
    - name: resto
      dtype: int64
    - name: time
      dtype: int64
    - name: now
      dtype: string
    - name: name
      dtype: string
    - name: trip
      dtype: string
    - name: sub
      dtype: string
    - name: com
      dtype: string
    - name: country
      dtype: string
    - name: country_name
      dtype: string
    - name: filename
      dtype: string
    - name: ext
      dtype: string
    - name: fsize
      dtype: int64
    - name: md5
      dtype: string
    - name: w
      dtype: int32
    - name: h
      dtype: int32
    - name: tn_w
      dtype: int32
    - name: tn_h
      dtype: int32
    - name: tim
      dtype: int64
    - name: replies
      dtype: int32
    - name: images
      dtype: int32
    - name: bumplimit
      dtype: int32
    - name: imagelimit
      dtype: int32
    - name: archived
      dtype: int32
    - name: closed
      dtype: int32
    - name: toxicity
      dtype: float32
    - name: severe_toxicity
      dtype: float32
    - name: inflammatory
      dtype: float32
    - name: profanity
      dtype: float32
    - name: insult
      dtype: float32
    - name: obscene
      dtype: float32
    - name: spam
      dtype: float32
    - name: entities
      dtype: string
  splits:
    - name: train
      num_bytes: 19440000000
      num_examples: 134529233
  download_size: 19440000000
  dataset_size: 134529233
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*-of-00270.parquet
license: cc-by-4.0
task_categories:
  - text-classification
language:
  - en
tags:
  - 4chan
  - toxicity
  - perspective-api
  - named-entities
  - political
pretty_name: "/pol/ 4chan Augmented (Jun 2016 - Nov 2019)"
size_categories:
  - 100M<n<1B
---

# /pol/ 4chan Augmented Dataset

**134.5M posts** from 3.4M threads on 4chan's /pol/ board (June 2016 - November 2019), augmented with Perspective API toxicity scores and named entity recognition.

## Dataset Description

This dataset contains posts from 4chan's Politically Incorrect (/pol/) board, collected between June 2016 and November 2019. Each post has been augmented with:

- **Perspective API toxicity scores** (7 dimensions): toxicity, severe_toxicity, inflammatory, profanity, insult, obscene, spam
- **Named Entity Recognition**: extracted entities stored as JSON arrays

### Source

Original data from [Papasavva et al. (2020)](https://zenodo.org/records/3606810) — "Raiders of the Lost Kek: 3.5 Years of Augmented 4chan Posts from the Politically Incorrect Board".

### Format

270 parquet shards with zstd compression (~70MB each). The dataset is directly loadable with HuggingFace `datasets`:

```python
from datasets import load_dataset
ds = load_dataset("ylelauta/pol-4chan-augmented")
```

### Schema

| Field | Type | Description |
|-------|------|-------------|
| `thread_no` | int64 | Thread number (from OP) |
| `no` | int64 | Post number |
| `resto` | int64 | 0 = OP, >0 = reply-to thread number |
| `time` | int64 | Unix timestamp |
| `com` | string | Comment HTML |
| `country` / `country_name` | string | Poster's country flag |
| `sub` | string | Subject (OP only) |
| `name` / `trip` | string | Poster identity |
| `filename` / `ext` / `fsize` / `md5` / `w` / `h` / `tim` | mixed | Image metadata |
| `replies` / `images` | int32 | Thread stats (OP only) |
| `toxicity` | float32 | Perspective API toxicity score (0-1) |
| `severe_toxicity` | float32 | Severe toxicity score |
| `inflammatory` | float32 | Inflammatory score |
| `profanity` | float32 | Profanity score |
| `insult` | float32 | Insult score |
| `obscene` | float32 | Obscene score |
| `spam` | float32 | Spam score |
| `entities` | string | JSON array of named entities |

### Statistics

- **134,529,233** posts
- **3,397,911** threads
- **270** parquet shards
- Date range: June 2016 - November 2019

### Citation

```bibtex
@inproceedings{papasavva2020raiders,
  title={Raiders of the Lost Kek: 3.5 Years of Augmented 4chan Posts from the Politically Incorrect Board},
  author={Papasavva, Antonis and Zannettou, Savvas and De Cristofaro, Emiliano and Stringhini, Gianluca and Blackburn, Jeremy},
  booktitle={Proceedings of the International AAAI Conference on Web and Social Media},
  year={2020}
}
```

### License

CC-BY-4.0 (following the original dataset license)