Update README.md

README.md

@@ -17,6 +17,9 @@ The model was trained on nearly 10 million hospital notes from the Amsterdam Uni
 
 ## Privacy
 By anonymizing the training data we made sure the model did not learn any representative associations linked to names. Apart from the training data, the model's vocabulary was also anonymized. This ensures that the model can not predict any names in the generative fill-mask task. 
+For more information on the model's anonymization see our publication: 
+
+Verkijk, S., & Vossen, P. (2022). Efficiently and thoroughly anonymizing a transformer language model for Dutch electronic health records: a two-step method. In Proceedings of the Thirteenth Language Resources and Evaluation Conference (pp. 1098-1103).
 
 ## Authors
 Stella Verkijk, Piek Vossen