"Analysis of Recent Topics in Social Networks Using Natural Language Processing"

The purpose of our work was to collect Russian-language dataset for Stance Detection task based on posts by users of social network Twitter, also we made some experiments on our data. We called our dataset TwiRuS (Twitter Russian Stances).

Collected dataset: TwiRuS

In final version of our dataset we have 5 targets (topics): cancel culture, feminism, LGBTQ+, ageism and lookism. Additionally we collected sentiment labels for each tweet.

Total count of tweets for each topic

Topic	Count of tweets
cancel culture	1567
feminism	4375
LGBTQ+	12332
ageism	1620
lookism	1290

Total count of labels for stances and sentiments

Stance Label	Count of tweets	Sentiment Label	Count of tweets
Favor	5916	Positive	2972
Against	6476	Negative	9514
Neutral	8792	Neutral	8698

Distributions of stance and sentiment labels for each topic

cancel culture

feminism

LGBTQ+

ageism

lookism