CrisiLangs Dataset

Dataset Description

CrisiLangs is a multilingual dataset for the generation of warning messages in crisis scenarios, in English, Italian, Bulgarian, German and Greek. It is derived from CrisiText: 1,000 events were sampled, reviewed and post-edited in English by two experts, and then translated into the other four languages.

---

Dataset Summary

• Name: CrisiLangs
• Domain: Crisis and emergency situations
• Size: 1,000 events (411 scenarios) in 5 parallel languages; train 800 / validation 100 / test 100 events
• Languages: English, Italian, Bulgarian, German, Greek
• Task: Text generation (warning messages from crisis events)

---

Data Fields

• scenario_id (string): ID of the crisis scenario.
• crisitext_scenario_id (string): ID of the same scenario in CrisiText.
• split (string): Dataset split.
• guidelines (list[string]): Crisis type(s).
• events (list[dict]): For each event:
  • event_id (string): Event identifier.
  • crisitext_event_id (string): ID of the same event in CrisiText.
  • event (dict): Crisis event description, per language.
  • message (dict): Warning message, per language.

guidelines.json contains the emergency guidelines for each crisis type, in all five languages.

---

Usage

You can load the dataset directly with the 🤗 datasets library:

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("LanD-FBK/CrisiLangs")

# Look at one example
print(dataset["train"][0])