The thesis evaluates the method of topic modeling. It investigates how topic modeling could be applied on a material consisting of articles from publications within library and information science, testing and evaluating the method and its application on a diachronically filtered corpus. Based on the Latent Direchlet allocation model, 20 topics are extracted from each period of 5 years between 1990 and 2014. The results shows the high potential of applying topic modeling in historical research, and reveals several potential problems connected to material collection in a way that makes the results representative of the field one is attempting to map.