# Alexandra Instituttet NLP & DaCY

Production-ready Danish language pipelines, named entity recognition (DaNE), and benchmarking suites (ScandEval).

## Access profile

| Field | Value |
| --- | --- |
| Category | API |
| Sector | Other |
| Access type | Direct |
| Auth required | Public |
| Reliability | Stable |
| Official | Yes |
| Status | active |
| Agent can | read, search |

## Links

- Website: https://alexandra.dk
- Docs: https://github.com/centre-for-humanities-computing/DaCY
- GitHub: https://github.com/alexandrainst
- Directory entry: https://agentaccess.dk/alexandra-instituttet-nlp

## About

Alexandra Instituttet develops and maintains core open-source NLP toolkits and benchmark infrastructure for the Danish language. Key assets include DaCY (a state-of-the-art Danish NLP framework built on spaCy), DaNE (the standard Danish Named Entity dataset covering persons, locations, and organizations), and ScandEval (the standard Scandinavian LLM evaluation leaderboard).

## Agent use cases

- Extract Danish named entities (personer, organisationer, lokationer) from unstructured text
- Benchmark and evaluate LLM performance on standardized Danish comprehension tasks via ScandEval
- Tokenize, POS-tag, and parse Danish sentence dependencies in Python

## Notes

Fully open-source Python packages distributed via PyPI and Hugging Face. Excellent documentation and research reproducibility.

---
Source: [AgentAccess](https://agentaccess.dk/) — directory of agent interfaces for Denmark.