OpenAI and Python Driven Tripple D - Developing a Database from Qualitative Data in Political Science: A Case Study of Migrant Motivations

The advent of computational social science represents a paradigm shift in the study of the social sciences and humanities. The recent wave of artificial intelligence (AI) and subsequent technological transformations in knowledge production mechanisms have led political scientists to revisit traditional data collection and analysis approaches. The field of migration studies faces two challenges: one common to all subjects, the other specific to it. In the former, there is a massive and growing secondary literature; mastering it requires a huge time investment. In the latter, there is a shortage of reliable data on migration and refugee flows. To address both weaknesses, we have designed an ‘algorithm’ using a Large Language Model (GPT-4) along with various Python libraries to extract data from the literature on the topic of irregular or undocumented migrants and refugees coming to Europe. The algorithm is operationalized in six phases: data extraction from literature, building a database, editing /modification/visualization of variables, and thematic clustering of details. The algorithm can be applied to any topic; for demonstration purposes, we have explored, through 100 peer-reviewed articles, book chapters, and policy reports, the motivations of migrants from the global south in traveling to Europe.