DeepText is a deep learning based biomedical event extraction system, written in Scala and Lua (using Torch). A detailed description of the approach implemented in this system may be found in a research paper titled DeepText: Biomedical event extraction via deep learning and recursive projection model.
The system extracts the biomedical events in an end-to-end fashion with minimal preprocessing and feature engineering, thanks to the recursive projection model and deep neural networks.
In this machine learning system, I try to embrace the scala cake-pattern to keep the project extendable and maintainable. Every case of the experiment is designed as a scala trait and an application object implementing such a trait. You will also see that the system is modular, which is implemented as a trait as well. So, we can mix those traits easily to come up with a new higher level component and these components are reusable in a sense that we mix them with others. This philosophy allows us to test different setups in our system easily (just mixing lower level trait components to try new setup).