Using POTATO for interpretable information extraction

About

This article is an introduction to the POTATO library. POTATO is a language independent human-in-the-loop XAI (explainable AI) framework for extracting and evaluating interpretable graph features for any classification problem in Natural Language Processing (NLP).

The article includes:

• A short introduction to rule-based methods for text classification
• Introduction to defining graph patterns in POTATO
• Learning patterns automatically
• The human-in-the-loop (HITL) framework

Introduction

Currently, text processing tasks (as many other domains) are dominated by machine learning models. …

computational-linguistics explainable ai information information extraction interpretability machine learning nlp