Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -1,5 +1,112 @@
|
|
1 |
-
---
|
2 |
-
license: other
|
3 |
-
license_name: youtube-terms-of-service
|
4 |
-
license_link: https://www.youtube.com/t/terms
|
5 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: other
|
3 |
+
license_name: youtube-terms-of-service
|
4 |
+
license_link: https://www.youtube.com/t/terms
|
5 |
+
task_categories:
|
6 |
+
- text-classification
|
7 |
+
language:
|
8 |
+
- ko
|
9 |
+
tags:
|
10 |
+
- youtube
|
11 |
+
- korean
|
12 |
+
- sentiment-analysis
|
13 |
+
- social-media
|
14 |
+
- user-generated-content
|
15 |
+
pretty_name: Korean YouTube Comment Sentiment Dataset
|
16 |
+
size_categories:
|
17 |
+
- 1K<n<10K
|
18 |
+
---
|
19 |
+
# Korean YouTube Comment Sentiment Dataset
|
20 |
+
|
21 |
+
## Data Overview
|
22 |
+
### Summary
|
23 |
+
|
24 |
+
본 데이터셋은 유튜브에서 수집된 **한국어 댓글 5,482개**와 이에 대응하는 **감정 레이블**(긍정, 부정, 중립, 불명확)로 구성된 감정 분류용 데이터셋입니다.
|
25 |
+
|
26 |
+
- 주요 레이블: `긍정`, `부정`, `중립`, `불명확`
|
27 |
+
|
28 |
+
### Features
|
29 |
+
- **수집 대상**: 요리, 뷰티, 게임, 여행, 쇼핑 등 분야의 **10만 명 이상 구독자를 보유한 유튜브 채널**
|
30 |
+
- **형식**: JSON (`id`, `text`, `label`)
|
31 |
+
- **검수**: 한국인 검수자에 의한 **수작업 라벨링 및 교차 검토**
|
32 |
+
|
33 |
+
본 데이터셋은 구어체, 이모지, 줄임말 등 실제 사용자 표현이 반영되어 있습니다.
|
34 |
+
|
35 |
+
## Dataset Structure
|
36 |
+
|
37 |
+
### Dataset Fields
|
38 |
+
|
39 |
+
| Field | Type | Description |
|
40 |
+
|------------|--------|-----------------------------------------------------------------------------|
|
41 |
+
| `id` | string | 각 댓글의 고유 ID|
|
42 |
+
| `text` | string | 유튜브 댓글 본문 텍스트 |
|
43 |
+
| `label` | string | 감정 레이블. [`긍정`, `부정`, `중립`, `불명확`]|
|
44 |
+
| `..` | .. | ..|
|
45 |
+
|
46 |
+
|
47 |
+
In our repository, the dataset is divided into the following splits:
|
48 |
+
- `train`: 3,421개
|
49 |
+
- `validation`: 868개
|
50 |
+
- `test`: 1,193개
|
51 |
+
|
52 |
+
|
53 |
+
## Dataset Details
|
54 |
+
|
55 |
+
### Source Data
|
56 |
+
|
57 |
+
- 2024년 상반기, **구독자 10만 명 이상**을 보유한 한국 유튜버의 영상 댓글 수집
|
58 |
+
- 수집 분야: 요리, 뷰티, 게임, 여행, 쇼핑 등
|
59 |
+
|
60 |
+
### Annotations
|
61 |
+
|
62 |
+
- 콘텐츠 맥락을 이해하는 **한국인 검수자 5인**이 직접 수작업 라벨링
|
63 |
+
- 해석이 모호한 사례는 검토 후 `uncertain`으로 분류
|
64 |
+
|
65 |
+
### Privacy
|
66 |
+
|
67 |
+
모든 댓글에서 사용자 계정명, 영상 URL 등의 개인정보는 제거되었으며,
|
68 |
+
실제 사용자 식별이 불가능하도록 가명 처리 또는 제거 작업을 수행했습니다.
|
69 |
+
다만, 댓글 내용 중 일부에 유추 가능한 채널명, 인물명, 지명 등이 **잔존할 가능성이 있습니다**.
|
70 |
+
|
71 |
+
이 데이터는 연구 목적으로만 제공되며, 원본 유튜브 사용자와 직접적인 연계가 되지 않도록 유의 바랍니다.
|
72 |
+
|
73 |
+
## Additional Information
|
74 |
+
|
75 |
+
### Licensing Information
|
76 |
+
|
77 |
+
이 데이터셋은 YouTube에서 수집된 댓글을 기반으로 하며, YouTube의 [이용약관](https://www.youtube.com/t/terms)을 준수합니다.
|
78 |
+
모든 사용자 이름, ID, 영상 링크 등 개인정보는 제거하였으나, 일부 표현에 사용자명 또는 채널명을 암시하는 정보가 **잔존할 수 있습니다**.
|
79 |
+
|
80 |
+
이 데이터셋은 **비상업적 연구 목적**으로만 사용 가능하며, 재배포는 금지됩니다.
|
81 |
+
YouTube 또는 원 작성자의 권리를 침해하지 않도록 주의해 주세요.
|
82 |
+
|
83 |
+
## Disclaimer
|
84 |
+
|
85 |
+
This dataset contains user-generated content originally posted on YouTube.
|
86 |
+
All data has been anonymized and is presented for academic research purposes only.
|
87 |
+
Redistribution or use for commercial purposes is not permitted.
|
88 |
+
|
89 |
+
If you are a rights holder and believe your content has been used improperly, please contact us at [email protected].
|
90 |
+
|
91 |
+
### Citation Information
|
92 |
+
|
93 |
+
```
|
94 |
+
@misc{korean_youtube_sentiment_2025,
|
95 |
+
title={Korean YouTube Comment Sentiment Dataset},
|
96 |
+
author={Hanjun Jung, Jinsoo Kim, Junhyeok Choi, Suil Lee, Seongjae Kang},
|
97 |
+
year={2025},
|
98 |
+
publisher={Hugging Face},
|
99 |
+
}
|
100 |
+
```
|
101 |
+
|
102 |
+
### Contributions
|
103 |
+
**Repository:** [링크]
|
104 |
+
|
105 |
+
**Email**: [email protected]
|
106 |
+
|
107 |
+
**Team**
|
108 |
+
- Hanjun Jung
|
109 |
+
- Jinsoo Kim
|
110 |
+
- Junhyeok Choi
|
111 |
+
- Suil Lee
|
112 |
+
- Seongjae Kang
|