?
Политические категории в обсуждениях соседских онлайн-сообществ: методика эмпирического исследования на примере г. Волгограда
In the context of digitalization of society, neighborhood communication is increasingly moving online. The communication of neighbors on public social media platforms creates extensive archives of data that can be used for political science research. These data open up new opportunities for the study of grassroots politics, group identities, political activism, etc. Nevertheless, there is currently a gap in research related to the lack of comprehensive methods that allow analyzing neighborhood online communities on a city scale, not limited to individual cases. The purpose of this work is to eliminate the existing gap by proposing a methods for identifying neighborhood online communities on a city scale using data from the VKontakte social network. Based on this methodology, an attempt is being made to identify in the discussions of the neighboring communities of the city of Volgograd an array of publications in which political categories occur. The paper describes the process of searching for neighborhood online communities, and also describes the preprocessing of text data. A mechanism is proposed for cleaning the corpus of texts from advertisements, which represented a wide proportion of publications, but acted as noise that hindered the main task. Publications mentioning political categories are identified by automatically encoding the cleared corpus into groups of keywords, followed by rechecking. The resulting archive, covering 17 years of neighborhood interaction, made it possible to determine, through the use of thematic modeling tools, the main periods of development of neighborhood communities. Starting from short non-political messages in groups of neighboring youth, ending with an established political repertoire: the work of homeowners associations; interaction with representatives of authorities, mainly executive ones, to solve local problems; organization and participation in local patriotic events. The most active types of neighborhood online-communities were identified: territorial public self-governments, communities of shareholders and microdistrict communities.