RomanticFragility-BrandShift: A dataset for studying romantic relationship fragility and online cognitive shift toward marriage-related brands

The stability of marital and romantic relationships depends not only on the maintenance of intimate ties at the individual level, but also on the regional institutional environment of marriage, patterns of emotional expression, and consumer cultural practices. With social media becoming a key arena for public discussions of marriage, romance, and brand meanings, online discourse on marriage-related brands and high-frequency searches for keywords such as “marriage” and “divorce” provide an empirical basis for tracing regional marital vulnerability and its temporal dynamics.To facilitate data reuse and methodological replication, this paper compiles and releases a multi-source integrated dataset for measuring marital vulnerability. The dataset includes three components: (1) annual statistics on marriage and divorce registrations for 31 provincial-level regions in mainland China from 2022 to 2024; (2) provincial-level daily search frequencies for the keywords “marriage” and “divorce” from June 2022 to December 2024, totaling 58,590 observations; and (3) social media texts related to the DR brand collected over the same period. After deduplication, rule-based cleaning, removal of low-information texts, and thematic relevance screening, the text data were organized into 35,654 structured records labeled by region, month, gender, theme ID, and narrative domain.Using marriage registration statistics, search frequency data, and brand-related texts, this paper constructs two core indicators: marital vulnerability and the brand perception shift index. The final dataset is provided in two Excel files that retain the original inputs and intermediate processing results while also offering analytical outputs for province-month matching. The dataset supports descriptive analysis, temporal distribution analysis, and indicator replication, and may be used in marriage and family research, regional comparative studies, digital marketing, computational communication, and methodological training in multi-source panel data construction, text classification, and theme identification.