Clusters (articles, news, atc...) by topic via the Latent Dirichlet allocation (LDA) model.
You can check lda.html for visualize the clustering done.
You can apply this notebook to any other dataset. Make sure to tune the parameters for the model to match your requirements.
For more info about preprocessing done (Tokenizing, lemmatizing, stopwords removal), Check this
For more info about the model used for clustering, check this