| --- |
| task_categories: |
| - text-generation |
| language: |
| - en |
| --- |
| # Train Dataset Topics |
| ## mixed output (480) |
| - emergency call pain location (body) ∼ 20 |
| - emergency call callback number of reporter ∼ 20 |
| - emergency call personal information (age, name, sex, ...) ∼ 20 |
| - fire department call person trapped in car after accident ∼ 20 |
| - emergency call geographical location description (country, area, city, street, house number, etc...) ∼ 10 |
| - vacation geographical location description ∼ 10 |
| - business transport geographical location description ∼ 10 |
| - shoplifting geographical location description ∼ 10 |
| - location description (attributes: city, street, house number, floor in house) ∼ 80 |
| - medical transport request ∼ 40 |
| - job interview ∼ 10 |
| - jobcenter call ∼ 10 |
| - job talk between friends ∼ 10 |
| - job talk about benefits ∼ 10 |
| - real estate transaction ∼ 20 |
| - real estate investment ∼ 20 |
| - police call report crime ∼ 40 |
| - environmental disaster / major loss event (flooding, volcanic eruption, earthquake, oil spill) ∼ 40 |
| - mental health crises ∼ 40 |
| - public disturbances ∼ 40 |
|
|
| ## empty output (50) |
| - emergency call scenario |
| - fire department call scenario |
| - police call scenario |
| - transport request |
|
|
| ## Eval Dataset |
| - Notitia Kinas cases 1-7 |
