docs: use HF paper link

README.md

@@ -259,7 +259,7 @@ There a many use cases, where this dataset can help:
 * LM: If you search for high-quality data to train LMs or LLMs for German: go for it!
 * Summarization: Look at the first entry of the `content` array of each newspaper article. A potential `<strong>` tag is used to decorate the short description of an article. Given that information, it is possible to construct a summarization dataset.
 * Topic Classification: The `type` information can be useful to train various kind of classification models. Also the `ressourt` information can be quite useful.
-* In detail analysis: Very recently a [paper](https://www.arxiv.org/abs/2506.05388) about analysing German newspapers (for Gender Bias and Discrimination) was published. However, the dataset is not freely available, so why not use this dataset for similar analysis?
+* In detail analysis: Very recently a [paper](https://huggingface.co/papers/2506.05388) about analysing German newspapers (for Gender Bias and Discrimination) was published. However, the dataset is not freely available, so why not use this dataset for similar analysis?
 
 Overall, the dataset comes in the rawest form (no preprocessing is done, no information were removed) possible, which allows great research on German NLP.