Learning from the machine: A topic modelling and network analysis of the CJE
The Cambridge Journal of Economics (CJE) was founded in 1977 with the aim of creating a forum for different traditions within heterodox economics. Although the mission statement hints at the main topics of interest to the journal, it may not necessarily offer an accurate description of the main areas of research explored by papers published there. What would the journal’s “aims and scope” look like if it were the output of a machine learning algorithm used for recovering the latent structure of the CJE archive? We apply the topic-modelling Latent Dirichlet Allocation (LDA) technique to the full texts of (1608) articles published in the CJE between 1977 and 2013 and build a network where the strengths of the links are given by semantic similarities obtained via LDA. We then use the Louvain method to identify research communities within the network. The resulting science map sheds light on the variety of research programs publishing papers in the CJE, on their nature, as well as on the relationships between research topics and communities gathered under the umbrella of heterodoxy.
Area:
Keywords: Cambridge Journal of Economics, economics journals, research communities, heterodox economics, topic modelling, network analysis
Please Login in order to download this file