Conversational Question Generation in Russian

In this paper, we discuss possibilities of automatic generation of conversational questions in Russian. We are exploring the possibility of using “A Conversational Question Answering Challenge” (CoQA) dataset translated into Russian for training an encoder-decoder model. We review several techniques for improving the quality of questions generated in the Russian language. The results are evaluated manually. Combining a neural network-based approach with a rules-based approach, we develop a system for automatic examination of university students.