Generate Your Counterfactuals: Towards Controlled Counterfactual Generation for Text

TL;DR

This paper introduces GYC, a framework for generating controlled, diverse, and goal-oriented counterfactual text samples to evaluate and improve the fairness and robustness of NLP systems.

Contribution

GYC is a novel framework that produces plausible, diverse, and goal-directed counterfactual texts for testing NLP models and debiasing methods.

Findings

GYC generates counterfactuals with high plausibility and diversity.

Counterfactuals effectively target specific conditions like sentiment or named entities.

GYC improves model evaluation and debiasing processes.

Abstract

Machine Learning has seen tremendous growth recently, which has led to larger adoption of ML systems for educational assessments, credit risk, healthcare, employment, criminal justice, to name a few. The trustworthiness of ML and NLP systems is a crucial aspect and requires a guarantee that the decisions they make are fair and robust. Aligned with this, we propose a framework GYC, to generate a set of counterfactual text samples, which are crucial for testing these ML systems. Our main contributions include a) We introduce GYC, a framework to generate counterfactual samples such that the generation is plausible, diverse, goal-oriented, and effective, b) We generate counterfactual samples, that can direct the generation towards a corresponding condition such as named-entity tag, semantic role label, or sentiment. Our experimental results on various domains show that GYC generates counterfactual text samples exhibiting the above four properties. GYC generates counterfactuals that can act as test cases to evaluate a model and any text debiasing algorithm.