Evolution of Topics in the Psychology Domain
Latent Dirichlet Allocation, a particular method for topic modeling is a generative probabilistic model that models texts as a mixture of underlying topics that could be used for further psychology domain specific research.
Abstract
Topic modeling is a set of statistical methods for modeling collections of discrete data such as text corpora. It is used as a text-mining tool to discover the hidden semantic structures in a text body. Latent Dirichlet Allocation, a particular method for topic modeling is a generative probabilistic model that models texts as a mixture of underlying topics. In this thesis, Latent Dirichlet Allocation is used on a large corpus of texts from the domain of psychology. A model with 100 topics is generated, and the resulting topics are labeled. The occurrence of the topics is analysed over a time span of 40 years. The resulting model could be used for further psychology domain specific research.