Julian Risch, Timo Möller, Julian Gutsch, Malte Pietsch
Semantic Answer Similarity for Evaluating Question Answering Models
A paper that presents a metric to compare ground-truth annotations of question answering datasets with model predictions with regard to semantic similarity rather than lexical similarity of answers.
13 August 2021
Bogdan Kostić, Julian Risch, Timo Möller
Multi-modal Retrieval of Tables and Texts Using Tri-encoder Models
A paper that presents a new multi-modal dataset based on text and table datasets from related work and compares retrieval performance of bi-encoder and tri-encoder models.
09 August 2021
Timo Möller, Julian Risch, Malte Pietsch
GermanQuAD and German DPR: Improving Non-English Question Answering and Passage Retrieval
A paper that presents GermanQuAD, a dataset of extractive question/answer pairs, and its adaption to Germany DPR, a training dataset for dense passage retrieval.
26 April 2021
Timo Möller, Malte Pietsch, G. Anthony Reina, Raghavan Jayakumar
COVID-QA: A Question Answering Dataset for COVID-19
A SQuAD style Question Answering dataset with 2.019 QA pairs annotated by medical expert.
18 June 2020
Branden Chan, Timo Möller, Stefan Schweter
German's Next Language Model
This paper present the experiments which lead to the creation of our BERT and ELECTRA based German language models, GBERT and GELECTRA.
03 December 2020