PathReasoner: Modeling Reasoning Path with Equivalent Extension for Logical Question Answering

TL;DR

PathReasoner is a novel architecture that models logical reasoning as paths, using an atom extension strategy and specialized transformer modules, achieving strong performance and generalization on logical reasoning benchmarks.

Contribution

It introduces PathReasoner, a new model that transforms logical samples into reasoning paths and employs an atom extension strategy supported by equivalent formulas.

Findings

Achieves competitive results on logical reasoning benchmarks.

Demonstrates strong generalization abilities.

Effectively models logical structure and consistency.

Abstract

Logical reasoning task has attracted great interest since it was proposed. Faced with such a task, current competitive models, even large language models (e.g., ChatGPT and PaLM 2), still perform badly. Previous promising LMs struggle in logical consistency modeling and logical structure perception. To this end, we model the logical reasoning task by transforming each logical sample into reasoning paths and propose an architecture \textbf{PathReasoner}. It addresses the task from the views of both data and model. To expand the diversity of the logical samples, we propose an atom extension strategy supported by equivalent logical formulas, to form new reasoning paths. From the model perspective, we design a stack of transformer-style blocks. In particular, we propose a path-attention module to joint model in-atom and cross-atom relations with the high-order diffusion strategy. Experiments show that PathReasoner achieves competitive performances on two logical reasoning benchmarks and great generalization abilities.