Generate Your Counterfactuals: Towards Controlled Counterfactual Generation for Text

Abstract

Machine Learning has seen tremendous growth recently, which has led to a larger adaptation 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 novel framework GYC, to generate a set of exhaustive counterfactual text, 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 \texttt{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.

“FIJO”: a French Insurance Soft Skill Detection Dataset

“…In the long term, we would like to make our model more explainable both to help us understand the strengths and weaknesses of our model and explain results to recruiters and human resources staff to allow them to adapt their needs in recruitment. To do so, we plan to explore counterfactual generation [28,29].…”

Explaining Recommendation System Using Counterfactual Textual Explanations

“…The proposed models by Madaan et al (2021) and Wu et al (2021) both generated counterfactual explanations in a conditional manner using GPT2. Madaan et al (2021) defines named-entity tags, semantic role labels, or sentiments as conditions for the words, while Wu et al (2021) controls the types and locations of perturbations in the text. Both approaches allow more targeted and controlled generation of counterfactual explanations in text.…”

You reap what you sow: On the Challenges of Bias Evaluation Under Multilingual Settings

“…Another research direction is the use of causal inference for measuring biases in LLMs, for example to analyze if the generated text by an LLM is affected considerably by only changing the protected attributes or categories in the input (Huang et al, 2020;Madaan et al, 2021;Cheng et al, 2021). In line with this idea, Huang et al ( 2020) used a sentiment classifier to quantify and reduce the sentiment bias existent in LLMs.…”