Back COLING 2022 - Two Workshop papers from LaSTUS lab

COLING 2022 - Two Workshop papers from LaSTUS lab

LaSTUS lab members Euan McGill and Santiago Egea, and visiting members Silvia Casola, Ishmael Obonyo and Luis Chiruzzo are presenting papers at COLING workshops this year.
14.10.2022

Imatge inicial

Recent projects involving LaSTUS lab are being presented to the Computational Linguistics community this month, both with the involvement of lab leader Prof. Horacio Saggion. COLING 2022, or The 29th International Conference on Computational Linguistics, showcases the state of the art in NLP research and applied computational methods in linguistics while addressing seven "grand challenges" at the forefront of research in the field. The event, along with 20 associated workshops, is taking place in the Republic of Korea for the first time and researchers from all over the world and from diverse backgrounds are attending both in person and virtually.

 

In the The Fifth Workshop on Technologies for Machine Translation of Low-Resource Languages (LoResMT 2022), lab members Euan McGill and Santiago Egea, and Luis Chiruzzo visiting from the Universidad de la República, Montevideo, Uruguay are presenting their paper "Translating Spanish into Spanish Sign Language: Combining Rules and Data-driven Approaches". This study of data augmentation methods with regards to aiding neural machine translation between signed and spoken languages. In this case, the researchers found that pre-training with rule-based augmented data before fine-tuning on real parallel Spanish/Spanish Sign Language data was beneficial for translation performance in a neural machine translation system.

 

In the Third Workshop on Scholarly Document Processing, lab visitors Silvia Casola from the Università degli Studi di Padova, Italy and Ishmael Obonyo from the University of Nairobi, Kenya are presenting their work "Exploring the limits of a base BART for multi-document summarization in the medical domain". Their work takes on the "Multi-document Summarization for Literature Review" shared task, and experiments with using limited input data and fewer computational resources when automatically reviewing scientific results.

Multimedia

Categories:

SDG - Sustainable Development Goals:

Els ODS a la UPF

Contact