A super-basic and work-in-progress implementation of Latent Dirichlet Allocation in Go. Ported from a Python implementation by Aaron Cohen, with a few minor tweaks. Takes advantage of goroutines for parallelization of repeated model-fitting runs.
##TODO:
- Do whatever one does with a Go program to allow it to be imported directly (e.g.,
import http://github.com/stevenbedrick/golda). - Complete the implementation (allow serialization of models, etc.)
- Implement hyperparameter tuning?
If you think you've found a bug, you're probably right. Email Steven Bedrick or open an issue.
Licensed under Matt Might's CRAPL for now.