Social sciences and AI joining forces: towards new approaches for computational social sciences

Abstract
While computational social science is seen as an interdisciplinary field of data science and social science, it is often overlooked that qualitative social research plays an important role in the various steps of the research process. In this chapter, we argue that structurally integrating qualitative research methods can fill blind spots and reduce bias in computational social science. To provide a roadmap for the systematic integration of qualitative social scientific research in computational social science, we provide an example case of fine-tuning and prompting AI-based large language models with manually annotated data for the task of analyzing social solidarity on social media and political debates. We describe the five-step research process–concept definition and operationalization (step 1), data collection and sampling (step 2), annotation (step 3), computing (step 4) and model evaluation (step 5)–highlighting the complementary contributions of each methodological perspective and skill set. We argue that methods from qualitative and quantitative social research, together with computer science unfold much greater potential than disciplinary approaches for the application of AI methods in the social sciences.