This repository contains the source code for binary sentiment analysis of tweets in Russian using combination of tf-idf and logistic regression.
The corpus of short texts in Russian based on Twitter messages is available at http://study.mokoron.com and described in the atricle "Y. Rubtsova, 'Constructing a Corpus for Sentiment Classification Training', Software & Systems, vol. 109, no. 1, pp. 72-78, 2015").
Experiments showed F-measure score up to 90.1%.