Lecture by Keli Du (Trier) in the lecture series “Praxis der Digital Humanities”

The lecture series “Praxis der Digital Humanities” will take place within the framework of the Master 'Digital Humanities'. It is open to all interested parties, also from other courses or from outside the university, and will be designed as a video conference with lecture and discussion possibility.

As a quantitative text analytic method, Latent Dirichlet Allocation (LDA), often referred to as Topic Modeling, has been widely used in Digital Humanities in recent years to explore numerous unstructured textual data. However, Topic Modeling itself has not yet been conclusively explored due to its complexity in terms of, for example, the algorithms or parameter settings used. It is also often observed when using Topic Modeling that this method is sensitive to the setting of the respective parameters. Therefore, LDA Topic Modeling is also often strongly criticized. In this talk, the systematic evaluation of LDA Topic Modeling will be presented. The goal of the evaluation is to understand the influence of the crucial factors on Topic Modeling from two perspectives, namely Topic Modeling-based document classification and Topic Coherence.

A special feature of this year's program is that some of the talks will be inserted into the series program not only digitally, but virtually, so to speak. These talks take place online in other event contexts. Instead of repeating them in our program, we get to be guests at the other lecture series.We thank our colleagues in Berlin (Digital History, Prof. Dr. Torsten Hiltmann) and Dresden (Applied Linguistics, Prof. Dr. Simon Meier-Vieracker) for their cooperation.