Topic modelling for qualitative studies

Sergey I. Nikolenko, Sergei Koltcov, Olessia Koltsova. Topic modelling for qualitative studies. Journal of Information Science, 2015.


This is a companion page for the paper above; here we show LDA and ISLDA results, datasets, and raw data for the experimental results used in the paper (where applicable).


LDA (Latent Dirichlet Allocation) results, 400 topics, three months dataset (see below):

 LDA result 1 (ZIP, 918 Кб)

 LDA result 2 (ZIP, 455 Кб)

 LDA result 3 (ZIP, 915 Кб)

 LDA result 4 (ZIP, 916 Кб)

 LDA result 5 (ZIP, 483 Кб)

ISLDA (Interval Semi-Supervised LDA) results, 400 topics, three months dataset (see below):

 ISLDA result 1 (ZIP, 916 Кб)

 ISLDA result 2 (ZIP, 907 Кб)

 ISLDA result 3 (ZIP, 913 Кб)

 ISLDA result 4 (ZIP, 908 Кб)

 ISLDA result 5 (ZIP, 904 Кб)

Datasets from the Russian language LiveJournal top bloggers. A dataset archive contains the word-document matrix in Mallet format, vocabulary, raw lemmatized text, and metadata (links to blog posts corresponding to documents).

 March 2012 dataset (57K documents, 80K words)

 April 2012 dataset (62K documents, 90K words)

 September 2012 dataset (54K documents, 83K words)

 March 2013 dataset (103K documents, 110K words)

 three months dataset (235K documents, 193K words)

 

Experimental data:

 human interpretation experiments (ZIP, 157 Кб)