---
license: cc-by-4.0
language:
- zh
pretty_name: PerCN
size_categories:
- 10K<n<100K
task_categories:
- text-classification
tags:
- mbti
- personality
- chinese
- social-media
---

# PerCN Dataset

## Overview
PerCN is a Chinese dataset for MBTI personality type prediction. Each sample contains multiple short posts from the same user, and labels are a 4-d binary vector corresponding to the four MBTI dimensions. The texts include typical Chinese social media expressions, emojis, and colloquial phrasing.

## Data Format
The dataset is provided in `JSONL` format with three splits: `train.jsonl`, `eval.jsonl`, and `test.jsonl`.

Each line is a JSON object with the following fields:
- `texts`: `List[str]`, multiple texts from the same user (variable length).
- `labels`: `List[int]`, a length-4 binary vector (0/1).

Example (truncated):
```json
{
  "texts": ["...", "...", "..."],
  "labels": [0, 1, 0, 1]
}
```

## Usage
Load from local files with `datasets`:
```python
from datasets import load_dataset

# Load from local paths
dataset = load_dataset(
    "json",
    data_files={
        "train": "train.jsonl",
        "validation": "eval.jsonl",
        "test": "test.jsonl",
    },
)

print(dataset["train"][0]["texts"][:3])
print(dataset["train"][0]["labels"])  # 4-d binary vector
```

## Ethics & Use
- The data contains user-generated content and may include personal or sensitive information. Do not attempt de-anonymization or re-identification.
- Recommended for research and educational use only. Comply with applicable laws and platform policies.