Publication
Modeling Quality of Experience in German Automatic Text Summarization and Machine Translation
Dinh Nam Pham; Vivien Macketanz; Shushen Manakhimova; Sebastian Möller
In: Christian Wartena; Ulrich Heid (Hrsg.). Proceedings of the 21st Conference on Natural Language Processing (KONVENS 2025): Workshops. Conference on Natural Language Processing (KONVENS-2025), September 10-12, Hannover, Germany, Pages 169-175, HsH Applied Academics, 2025.
Abstract
We present a model for predicting the Quality of Experience (QoE) of German machinegenerated text from Automatic Text Summarization (ATS) and Machine Translation (MT). Based on previously established quality dimensions, we fine-tuned BERT for ATS and ELECTRA for MT, which performed best per task. Adding linguistic features further improved accuracy. For ATS, BERT excelled as a multitarget regressor; for MT, separate ELECTRA models performed best. Our results show that combining linguistic features with language models enables robust QoE prediction.