Úloha odpovídání na otázky v češtině pomocí strojového překladu a mezijazykového přenosu znalostí

Macková, Kateřina

Reading comprehension and question answering are computer science disciplines in the field of natural language processing and information retrieval. Reading comprehension is the ability of the model to read text, process it and understand its meaning. One of its applications is in question answering tasks, which is concerned with building a system that can automatically find an answer in the text to a certain question relied on the content of the text. It is a well-studied task, with huge training datasets in English. However, there are no Czech datasets and models for this task. This work focuses on building reading comprehension and question answering systems for Czech, without requiring any manually annotated Czech training data. Our main focus is to create Czech training and development datasets, create the models for the Czech question answering system using Czech data, and create the models for the Czech question answering system using English data and cross-lingual transfer and compare the results and select the best model. First of all, we translated freely available English question answering datasets SQuAD 1.1 and SQuAD 2.0 to Czech to create training and development datasets. We then trained and evaluated several BERT and XLM-RoBERTa baseline models used for the question answering task in...

guest :: login Digital Repository
		Search		Submit		Help		About