用混合方法构建跨学科性别研究数据集，揭示其发展脉络与影响力。

The multidisciplinary and socially anchored nature of Feminist Studies presents unique challenges for bibliometric analysis, as this research area transcends traditional disciplinary boundaries and reflects discussions from feminist and LGBTQIA+ social movements. This paper proposes a novel approach for identifying gender/sex related publications scattered across diverse scientific disciplines. Using the Dimensions database, we employ bibliometric techniques, natural language processing (NLP) and manual curation to compile a dataset of scientific publications that allows for the analysis of Gender Studies and its influence across different disciplines. This is achieved through a methodology that combines a core of specialized journals with a comprehensive keyword search over titles. These keywords are obtained by applying Topic Modeling (BERTopic) to the corpus of titles and abstracts from the core. This methodological strategy, divided into two stages, reflects the dynamic interaction between Gender Studies and its dialogue with different disciplines. This hybrid system surpasses basic keyword search by mitigating potential biases introduced through manual keyword enumeration. The resulting dataset comprises over 1.9 million scientific documents published between 1668 and 2023, spanning four languages. This dataset enables a characterization of Gender Studies in terms of addressed topics, citation and collaboration dynamics, and institutional and regional participation. By addressing the methodological challenges of studying "more-than-disciplinary" research areas, this approach could also be adapted to delineate other conversations where disciplinary boundaries are difficult to disentangle.