README.md

---
license: cc-by-4.0
language:
- de
size_categories:
- 1K<n<10K
---

# DFKI MobIE: Tokenized Sentences

This datasets hosts a sentence-tokenized version of the [DFKI MobIE](https://github.com/dfki-nlp/mobie) dataset.

## Creation

The following script can be used to reproduce the creation of the dataset:

```python
import json

from flair.datasets import NER_GERMAN_MOBIE

corpus = NER_GERMAN_MOBIE()

with open("./train.jsonl", "wt") as f_out:
    for sentence in corpus.train:
        current_example = {
            "text": sentence.to_tokenized_string()
        }
        f_out.write(json.dumps(current_example) + "\n")
```

The extracted dataset has 6,900 sentences.