README.md

---
license: mit
task_categories:
  - text-classification
language:
  - en
tags:
  - movies
  - sensitivity
  - content-warnings
  - recommendation-systems
size_categories:
  - 10K<n<100K
configs:
  - config_name: trait_sensitivity
    default: true
    data_files:
      - split: train
        path: movie_trait_sensitivity.csv
  - config_name: ddd_raw
    data_files:
      - split: train
        path: imdb_sensitivity_table.csv
---

# IMDb Movie Sensitivity Dataset

This dataset provides sensitivity scores for movies based on 20 user traits, combining data from:
- **Does the Dog Die (DDD)**: Fine-grained content warnings from crowd-sourced votes
- **IMDb Parent Guide**: Coarse-grained severity ratings (Sex & Nudity, Violence & Gore, Profanity, Alcohol/Drugs, Frightening Scenes)

## Dataset Configurations

### `trait_sensitivity` (default)
Fused sensitivity scores for 61,424 movies across 20 traits.

```python
from datasets import load_dataset
ds = load_dataset("Dionysianspirit/imdb-sensitivity-table")
```

### `ddd_raw`
Raw DDD voting data for 32,604 movies.

```python
from datasets import load_dataset
ds = load_dataset("Dionysianspirit/imdb-sensitivity-table", "ddd_raw")
```

## Schema (trait_sensitivity)

| Column | Description |
|--------|-------------|
| `imdbId` | 7-digit IMDb ID (string) |
| `{trait}_final` | Final sensitivity score (0-1) |
| `{trait}_unsafe` | Boolean: score >= 0.66 threshold |
| `{trait}_source` | Source: "DDD", "ParentGuide", "Both", or "None" |

## 20 Traits

1. Animal lover (avoid animal harm/death)
2. Arachnophobia / reptile phobia
3. Anti-gore / squeamish
4. Avoid torture & extreme violence
5. Sexual violence sensitive
6. Kid-safety / child harm sensitive
7. Domestic abuse / stalking sensitive
8. Substance recovery / avoid drugs & alcohol
9. Self-harm & suicide sensitive
10. Mental health portrayal sensitive
11. Medical/health trauma avoider
12. Needle/medical procedure phobia
13. Claustrophobia / breathing distress sensitive
14. Photosensitivity & motion sickness
15. Hate speech / slur-sensitive
16. Gender/LGBTQ respect sensitive
17. Pregnancy/infant-loss sensitive
18. Disaster/accident avoider
19. Happy-ending preference
20. Horror avoider (avoids scares & supernatural)

## Fusion Formula

```
final_score = max(ddd_trigger, pg_risk)
unsafe = final_score >= 0.66

where:
  ddd_trigger = 1 if any DDD warning is triggered, else 0
  pg_risk = (weights · severity) / (3 * sum(weights))
```

## Citation

If you use this dataset, please cite the original data sources:
- Does the Dog Die: https://www.doesthedogdie.com/
- IMDb Parent Guide: https://www.kaggle.com/datasets/barryhaworth/imdb-parental-guide