Master Thesis - Eleni Metheniti

Linguistically inspired morphological inflection with a sequence to sequence model

Originally submitted and presented in September 2019 at Saarland University/DFKI Saarbrücken, published as a preprint in 2020.

Prerequisites

• Python3
• Pytorch
• Wikinflection Corpus

Experiments 1 & 2

• make dataset for train: python dataset.py --ĺang
• train: python main.py --lang (see other parameters in configure.py)

Languages: names from Wikinflection (ISO 693-3)

Experiment 3

• make small dataset for train: python dataset_Exp3.py --ĺang
• train: python main_Exp3.py --lang (see other parameters in configure.py)

Cite

Linguistically inspired morphological inflection with a sequence to sequence model E. Metheniti, G. Neumann, J. van Genabith - arXiv preprint arXiv:2009.02073, 2020