Decide on which DTM model to use, if any

juanrloaiza / latinamerican-philosophy-mining

Text mining philosophy journals in Latin America.

0 stars 2 forks source link

Decide on which DTM model to use, if any #19

Closed miguelgondu closed 1 year ago

miguelgondu commented 1 year ago

We are not happy with the new results from either Blei's binary, Gensim, or Tomotopy's. Which one should we use? There might be other tools in other languages (R, Cpp...)

One example: https://github.com/Arnie0426/FastDTM

miguelgondu commented 1 year ago

[ ] Let the Tomotopy train for looooong.
[ ] Train a 90 topic in the Blei binary.
[ ] Take a look at FastDTM
[ ] Move to another compiled model/language.
[ ] Drop DTMs.

juanrloaiza commented 1 year ago

We finally decided to use Blei's binary.