This repository implements Latent Dirichlet Allocation in Python 3, using strictly Python/NumPy routines.
Current version supports posterior inference via:
- Gibbs Sampling
I largely relied on the following two resources for implementing this algorithm:
A Theoretical and Practical ImplementationTutorial on Topic Modeling and Gibbs Sampling
Latent Dirichlet Allocation Using Gibbs Sampling
Potential future to-do's:
- add black-box inference
- add basic pre-processing pipeline (stemming, stopword removal, regex)
- speed up topic inference / topic sampling procedure