Welcome to Quoll’s documentation!

Quoll is the name of carnivorous marsupials living in Australia, New Guinea and Tazmania. It is also a python library for running NLP pipelines, based on the LuigiNLP workflow system. Quoll takes care of the sequence of tasks that are common to basic Machine Learning experiments with textual input: preprocessing, feature extraction, vectorizing, classification and evaluation. Provided that you prepare instances and labels, all these tasks can be ran as a pipeline, in one go. Quoll is built on top of several applications in the LaMachine Software distribution (Ucto, Frog, Colibri-core, PyNLPl), as well as popular python packages (Numpy, Scipy and Scikit-learn).

Quoll has the following advantages:

  • Can run full supervised machine learning pipeline with one command.
  • Stores intermediate output of the pipeline.
  • Maintains a full log of your experiments.
  • Offers various options at each stage of the pipeline.
  • If part of the pipeline is already completed, will continue from that point.
  • Experiments with different settings can be distinguished based on filenames.

Indices and tables