Search for K: Assessing Five Topic-Modeling Approaches to 120,000 Canadian Articles

Analysing Longitudinal Social Science Questionnaires: Topic modelling with BERT-based Embeddings

“…PMI is the probability of a pair of topic words with logarithmic calculation. Equation ( 2) denotes the NPMI calculation, where P (x, y) represents the probability of token x and token y occuring in the documents, is a smoothing constant and a "higher γ gives a higher NPMI more weight" [30]. The positive value of NPMI coherence means that the topic is relevant to the document.…”

Agreeing to Disagree: Choosing Among Eight Topic-Modeling Methods

“…Yet, computational social science has not been well informed by an explosion in methods and algorithms for topic modeling in the past two decades [1], [5], [6]. As suggested by DiMaggio [7], most topic-modeling techniques require various methodological decisions that many social scientists are unfamiliar with, have not considered, or lack experience with.…”