This paper describes a series of experiments aimed at testing the hybrid morphological analyzer NLTK4RUSSIAN on the datasets of the Dialogue Evaluation 2017. We provide a detailed explanation of the principles and main operation modes of the system. Experiments on training the system were based on RNC and OpenCorpora subcorpora. Certain problems dealing with the data format were solved in course of research: we performed conversion of Universal Dependencies markup format to PyMorphy2 format. The analyzer was tested on social network subcorpus, news subcorpus and fiction subcorpus. Test results for the NLTK4RUSSIAN morphological analyzer correspond to the gold standard of morphological annotation for the Russian language.