From 15953b6d8cd684637845401cfa9fc8d1723c6826 Mon Sep 17 00:00:00 2001
From: Florian Borchert <fl.borchert@gmail.com>
Date: Fri, 12 Jul 2024 09:12:07 +0200
Subject: [PATCH] Update README.md

---
 README.md | 24 ++++++++++++++----------
 1 file changed, 14 insertions(+), 10 deletions(-)

diff --git a/README.md b/README.md
index 6d02971..8b940dd 100644
--- a/README.md
+++ b/README.md
@@ -40,17 +40,10 @@ dataset = load_dataset("distemist", "distemist_linking_bigbio_kb")
 
 To use xMEN with existing NER pipelines, you can also create a dataset at runtime.
 
-#### [spaCy](https://spacy.io/)
-
-```python
-from xmen.data import from_spacy
-docs = ... #  list of spaCy docs with entity spans
-dataset = from_spacy(docs)
-```
+#### Span-based Formats
 
-for an example, see: [examples/02_spaCy_German.ipynb](examples/02_spaCy_German.ipynb)
-
-#### [SpanMarker](https://github.com/tomaarsen/SpanMarkerNER)
+Any span-based annotation format (e.g., based on character offsets), can be converted to a xMEN-compatible dataset.
+For instance, using [SpanMarker](https://github.com/tomaarsen/SpanMarkerNER) predictions:
 
 ```python
 from span_marker import SpanMarkerModel
@@ -62,6 +55,17 @@ from xmen.data import from_spans
 dataset = from_spans(preds, sentences)
 ```
 
+#### [spaCy](https://spacy.io/)
+
+```python
+from xmen.data import from_spacy
+docs = ... #  list of spaCy docs with entity spans
+dataset = from_spacy(docs)
+```
+
+for an example, see: [examples/02_spaCy_German.ipynb](examples/02_spaCy_German.ipynb)
+
+
 ## 🔧 Configuration and CLI
 
 xMEN provides a convenient command line interface to prepare entity linking pipelines by creating target dictionaries and pre-computing indices to link to concepts in them.