Finding Scientific Topics in Continuously Growing Text Corpora

This is a brief presentation of our framework and app (www.psychtopics.org) for continuous topic detection in scientific publications. Using psychology as an example, the presented framework makes use of RollingLDA and has a high degree of automation once the initial model is created. The model integrates new publications while keeping time series of topic trends consistent. This can help various stakeholders like researchers or policy-makers to evaluate how research fields evolve over time. unknown unknown