A self-explanatory contrastive logical knowledge learning method for sentiment analysis

Yulin Chen, Bo Yuan, Beishui Liao, Dov M. Gabbay

Abstract: Highlights•A new approach for incorporating label semantic knowledge and contrastive learning into self-explanatory methods.•A novel contrastive logical knowledge learning (CLK) method to provide model interpretability for sentiment analysis tasks.•A new explanation verification technique that combines counterfactual reasoning and deep logic rule learning.•Competitive performance and model interpretability for fine-grained sentiment analysis tasks.