hominluo/SocSci210

# SocSci210 Dataset

This dataset contains survey response data from social science research, presented in the paper ["Finetuning LLMs for Human Behavior Prediction in Social Science Experiments"](https://arxiv.org/abs/2509.05830), with three different experimental split mappings.

## Mapping Files

The mapping files are located in the `metadata/` folder:
- `metadata/participant_mapping.json`: Study-level seen/unseen splits by participant
- `metadata/task_mapping.json`: Sample-level train/eval splits (75/25) by task
- `metadata/condition_mapping.json`: Sample-level train/eval splits (75/25) by condition

## Loading the Dataset and Mappings

```python
from datasets import load_dataset
from huggingface_hub import hf_hub_download
import json

# Load the dataset
dataset = load_dataset("socratesft/SocSci210")

# Load a mapping file
mapping_file = hf_hub_download(
repo_id="socratesft/SocSci210",
filename="metadata/participant_mapping.json",
repo_type="dataset"
)

with open(mapping_file, 'r') as f:
participant_mapping = json.load(f)
```