The Technological Trajectory of Semantic Analysis: A Historical-Methodological Review of NLP in Social Sciences

Abstract

This paper examines the evolution and application of quantitative semantic analysis tools in social sciences, tracking their development from early statistical methods to contemporary large language models. The analysis demonstrates how computational advances have transformed qualitative research capabilities, enabling the systematic analysis of vast textual datasets while maintaining interpretative depth. The study presents a comprehensive review of key methodological approaches, including statistical analysis, topic modeling, semantic networks, and dimensionality reduction techniques, while examining their practical applications in social science research. Special attention is given to recent developments in natural language processing, particularly the emergence of transformer-based models and their impact on research methodologies. The paper provides a detailed typology of cases for applying machine learning strategies in social sciences, covering applications from sentiment analysis to cross-cultural studies. The research concludes by addressing methodological considerations and ethical implications for future research, emphasizing the importance of balancing technological innovation with research integrity and social responsibility.