Support gensim4 LdaModel #73
Conversation
`LdaModel`, that was in `gensim.models.lda`(gensim3), is implemented by `gensim.models.ldamodel` now (gensim4). The proposed solution tries both and exits with an error if neither module can be imported.
Check wether version 3 or 4 of gensim is loaded and train the lda model accordingly.
Just added a missing space.
This error-handling may not comply with the general litstudy convention.
Convert the mayor version number to int as expected in the conditional.
Pass the num_topics parameter to LdaModel with gensim4.
Added the missing '.T'.
|
If you could run black and commit, that would fix the only failing test in the CI. |
…a main() function in the calling module.
|
Thanks @larsgrobe , is the pull request ready to be merged at this point, or are you still working on it? |
|
It should be ready, at least I am using the code as such. Note that I also added the "ensemble LDA model", which aims at getting more stable topics from the LDA algorithm. This would need documentation, the concept is decribed here: BRIGL, Tobias, 2019, Extracting Reliable Topics using Ensemble Latent Dirichlet Allocation [Bachelor Thesis]. Technische Hochschule Ingolstadt. Munich: Data Reply GmbH. Supervised by Alex Loosley. Available from: https://www.sezanzeb.de/machine_learning/ensemble_LDA/ Best, Lars. |
|
Great! Thanks for your contributions. I am very happy to see support for the new ensemble LDA. I would like to have more topic modeling algorithms in litstudy that have been developed in recent years. I have you have more ideas on topic modeling or are experimenting with other NLP processing techniques, do not hesitate to open a pull request, discussions, or issue to discuss your idea or your findings! |
Resolves #70
The name of the module providing LDA model support has changed between versions 3 and 4 of gensim. Besides that, the call to
LdaModel()had to be modified.