LaSTUS lab is delighted to present two recent contributions to the research community, demonstrating the wide-reaching research done at TALN.

Dr. Ahmed AbuRa'ed's paper uses abstractive and extractive methods of text summarisation for information extraction from scientific papers. These generate 'work reports', containing the key information from a study which can then be used to generate effecitve citation sentences. Performance of these extractive methods was well above a baseline in terms of evaluation by both human and automatic metrics.

Pablo Accuosto's work tackles the lack of annotated data for use in text mining for scientific texts. It provides annotated abstracts in the diverse domains of computational lingustics and biomedicine, and is put to use in fine-tuning transformer-based models for argument mining.