A Differentiable Integer Linear Programming Solver for Explanation-Based Natural Language Inference
This addresses the challenge of making ILP-based NLI differentiable for better integration with deep learning, enabling more explainable and transparent NLI in complex domains, though it appears incremental as it builds on existing neuro-symbolic solvers.
The paper tackles the problem of integrating continuous language representations with Integer Linear Programming (ILP) for Natural Language Inference (NLI) by introducing Diff-Comb Explainer, a neuro-symbolic architecture that achieves superior performance compared to conventional ILP solvers, neuro-symbolic black-box solvers, and Transformer-based encoders, with significant improvements in precision, consistency, and faithfulness of explanations.
Integer Linear Programming (ILP) has been proposed as a formalism for encoding precise structural and semantic constraints for Natural Language Inference (NLI). However, traditional ILP frameworks are non-differentiable, posing critical challenges for the integration of continuous language representations based on deep learning. In this paper, we introduce a novel approach, named Diff-Comb Explainer, a neuro-symbolic architecture for explanation-based NLI based on Differentiable BlackBox Combinatorial Solvers (DBCS). Differently from existing neuro-symbolic solvers, Diff-Comb Explainer does not necessitate a continuous relaxation of the semantic constraints, enabling a direct, more precise, and efficient incorporation of neural representations into the ILP formulation. Our experiments demonstrate that Diff-Comb Explainer achieves superior performance when compared to conventional ILP solvers, neuro-symbolic black-box solvers, and Transformer-based encoders. Moreover, a deeper analysis reveals that Diff-Comb Explainer can significantly improve the precision, consistency, and faithfulness of the constructed explanations, opening new opportunities for research on neuro-symbolic architectures for explainable and transparent NLI in complex domains.