Overview of the GermEval 2024 Shared Task on Statement Segmentation in German Easy Language (StaGE)

Abstract

Sentences in easy language should only contain a few statements to enhance their readability. However, there exists no common definition of statements or tools that could automatically extract them. With this shared task, we try to close this gap in German Easy Language. We drafted a set of annotation guidelines and created a dataset with manually extracted numbers of statements and statement spans. We attracted three participating teams that tried to automatize the annotation process to automatically annotate the statements. All submissions are based on a BERT classification system and could outperform our naive baselines. However, especially the statement span extraction is challenging and requires further research. Our shared task tries to facilitate and motivate further research in this direction.