The analysis of extensive textual data is currently undergoing substantial advancements, as evidenced by the exponential rise in the number of publications employing such methodologies across diverse domains within the social sciences. These domains include but are not limited to history, management, economics, sociology, and political science. The objective of this paper is to critically examine the state-of-the-art methods, their progressions, and their applications. Additionally, we aim to discuss the opportunities and challenges presented by the paradigm shift brought about by large language models, focusing specifically on their implications for document annotation tasks, such as Zero Shot Classification, and the exploration of vast corpora, exemplified by methods like RAG.