A simple implementation of Gibbs-Lda algo written in CoffeeScript running on node.js. == Routine == Please compile all the .coffee files for the js modules, also including the files under langcode/ $ coffee -c *.coffee Orgnize all the data files under 'data/corpus', then run $ node token_main.js which will tokenize all the documents under data/corpus for each file as a single document, and create the docuemnts array and vocabulary file. then, $ node lda_main.js to run lda process, you can tune the parameters in lda_main.coffee the performance depends on #document * #kTopic, #voc * #kTopic once it's done, the model's phi and theta matrix will be under data/ dir then, $ node render_topic.js to create the topics/words/docs json file for data-server $ node data_server.js data server is simple browsing server for the data, but is very coupled to my own business, so make your own, :) for any information, you can contact me at winters.mi(at)gmail.com