Temporal analysis of computational economics: a topic modeling approach

This study offers a comprehensive investigation into the thematic evolution within computational economics over the past two decades, leveraging advanced topic modeling techniques. Utilizing latent semantic analysis, latent dirichlet allocation (LDA), and BERTopic models, we discerned the distinctiveness, relevance, and coherence of topics generated. BERTopic emerged superior, capturing more distinct and relevant themes in the field. It is noteworthy that our work offers a broader analysis than previous studies which examined a smaller subset of research with only the LDA algorithm. Through this methodology, we identified and labeled 14 key research themes and visualized their interrelationships and prevalence over time. Notably, general domains like market analysis, equilibrium analysis, and financial modeling maintained substantial representation, while specific areas such as game theory and allocation mechanisms observed increased attention. In contrast, domains like asset pricing and fair division saw a decline in interest. This study provides a systematic organization of the literature, captures shifts in trends, and offers recommendations for future research in computational economics.