This repo contains the code for the first mid-term project of the course 2021 Language Understanding System.
You may want to setup a virtual environment or use conda. For instance we can run:
python3 -m venv venv
then we activate the virtual environment with
source venv/bin/activate
This step is not required because the scripts are only using the core python3 libraries.
We then need to install by compiling from source the following libraries:
- openfst
- opengrm
More information can be found here.
You also need to have perl installed, because I used the original script to perform the evaluation.
To run the script on the already parsed dataset you can launch the following terminal (assuming that you already source the virtual enviroment, if you are using it):
python main.py <task> <ngrams> <smooth> <only_evaluate: default false>Specify the parameter task with the name of the dataset you want to perform the run. It could be a value of:
- baseline
- atis
- AskUbuntuCorpus
- ChatbotCorpus
- WebApplicationsCorpus
Other parameters can be specify to perform a specify evaluation with a certain configuration.