2,598 German-language documents annotated with fine-grained geo-entities and 15 traffic- and industry-related n-ary relations. The collection includes newswire articles, Twitter messages, and official traffic reports from police and railway sources.
Use Cases
- Train named entity recognition models to identify fine-grained geo-entities like streets and stops in German text
- Develop n-ary relation extraction systems to detect complex event structures such as Accidents or Strikes
- Benchmark cross-domain NLP performance using the mix of formal newswire and informal Twitter messages
Strengths
- 2,598 German-language documents sourced from newswire, Twitter, and official traffic reports
- Fine-grained geo-entity labels including specific categories for streets, stops, and routes
- 15 distinct n-ary relation types covering events like Accidents, Traffic jams, Acquisitions, and Strikes