Dataset · turkish-nlp-suite
Turkish named entity recognition (NER) markup
turkish-nlp-suite/turkish-wikiNER
20,000 sentences from Turkish Wikipedia with re-annotated named entities for training NER models.
What's inside
The corpus consists of 20,000 sentences in Turkish taken from Wikipedia (a sample from the Kuzgunlar NER set) and re-annotated by the Co-one team. Texts are tokenized with named entity tags; this same set was used to create new Turkish language packages in the spaCy library.
What you can do with it
- Training and fine-tuning named entity recognition (NER) models for the Turkish language
- Testing and comparing token classification pipelines on Turkish texts
- Extracting names, dates, and organizations from Turkish news and encyclopedia texts
- Tasks
- Token classification
- Languages
- Turkish
- Size
- 10K–100K rows
- Downloads this month
- 512
- Likes
- 15
Models trained on it
- f1k/katip-mask-nanoToken classification 169
- f1k/katip-mask-baseToken classification 161
- f1k/katip-mask-miniToken classification 133
- furkanbekmezci/turkextract-turkish-information-extractionToken classification 81
- girayyagmur/bert-base-turkish-ner-casedToken classification 8
- zypchn/berturk-nerToken classification 6
- WildGenie/nerextended-turkish-crfToken classification 0
- altaidevorg/akana-pii-taggerToken classification 0
How to load it
from datasets import load_dataset
ds = load_dataset("turkish-nlp-suite/turkish-wikiNER")