TIGTEC : Token Importance Guided TExt Counterfactuals

TL;DR

TIGTEC is a modular method for generating sparse, plausible, and diverse textual counterfactual explanations by targeting high-contribution words using an innovative attention-based importance measure.

Contribution

It introduces a novel attention-based local feature importance and a beam search approach for efficient counterfactual generation in text data.

Findings

High success rate in generating counterfactuals

Produces sparse and diverse explanations

Applicable in both model-specific and model-agnostic settings

Abstract

Counterfactual examples explain a prediction by highlighting changes of instance that flip the outcome of a classifier. This paper proposes TIGTEC, an efficient and modular method for generating sparse, plausible and diverse counterfactual explanations for textual data. TIGTEC is a text editing heuristic that targets and modifies words with high contribution using local feature importance. A new attention-based local feature importance is proposed. Counterfactual candidates are generated and assessed with a cost function integrating semantic distance, while the solution space is efficiently explored in a beam search fashion. The conducted experiments show the relevance of TIGTEC in terms of success rate, sparsity, diversity and plausibility. This method can be used in both model-specific or model-agnostic way, which makes it very convenient for generating counterfactual explanations.