The problem of intimate partner violence (IPV), which has become more acute under quarantine during the COVID-19 pandemic, affects millions of people and families around the world. It is extremely difficult for victims to disclose their experience of IPV and seek help from the relevant services. They prefer to search for help from people of their kind, on the Internet, in particular. Nevertheless, the forms, types of violence, people’s stance and the sentiment of opinions regarding violence are particularly diverse. The paper proposes a method for creating an annotated corpus of texts through crowdsourcing. Using the methodology of the crowdsourcing study by M. Sabou et al., the authors expand and supplement it with important points specific to burning social problems, such as compiling search queries, describing data processing, as well as evaluating the relevance of the data obtained. The key result of the study is the first Russian annotated IPV text corpus. The main type of crowdsourcing presented in the paper is volunteer crowdsourcing, based on the annotators’ loyalty to the problem under study. Despite the fact that the possibilities of its application are increasingly expanding, the scientific community still lacks a set of guidelines similar to traditional and expert methods for creating annotated corpora.