The paper presents the results of the study focused on the possibilities of analyzing the users’ discussions content, as well as their moods with reference to geolocation. The research data was collected from social network VKontakte. The data was distributed to city districts detecting citizens’ emotions in posts and reactions to them. As a result, the labeled dataset with two new features: the sentiment class and the percentage of attitude towards the class (used for further mapping) was created. The results of the work are useful for use in the process of urban management and planning the development of territories.