KIDS-DMM uses an open-source Java package to implement the algorithm.

• Java （Version=1.8）

We procided the following three short text datasets for evaluation, SearchSnippets, GoogleNews, and Biomedical. All of corpus files and the corresponding label files have been prepared in the path ./datasets according to the survey, Short Text Topic Modeling Techniques, Applications, and Performance: A Survey.

Taking SearchSnippets as an example, the dataset file path is as follows.

datasets

SearchSnippets

word_wiki

SearchSnippets.txt

SearchSnippets_label.txt

SearchSnippets_vocab.txt

SearchSnippets_Word2VecSim.txt

For the corresponding word_wiki and the word2VecSim, you can download from this following this paper.

bash run.sh