Bi-directional Cognitive Thinking Network for Machine Reading Comprehension

TL;DR

This paper introduces a bi-directional cognitive framework for machine reading comprehension that mimics human reverse and inertial thinking, improving answer accuracy by decoupling knowledge directions.

Contribution

It presents a novel bi-directional cognitive knowledge framework and network that enhance machine reading comprehension by simulating human-like thinking processes.

Findings

Improved performance on DuReader dataset.

Effective decoupling of bi-directional knowledge.

Enhanced reasoning capabilities in QA tasks.

Abstract

We propose a novel Bi-directional Cognitive Knowledge Framework (BCKF) for reading comprehension from the perspective of complementary learning systems theory. It aims to simulate two ways of thinking in the brain to answer questions, including reverse thinking and inertial thinking. To validate the effectiveness of our framework, we design a corresponding Bi-directional Cognitive Thinking Network (BCTN) to encode the passage and generate a question (answer) given an answer (question) and decouple the bi-directional knowledge. The model has the ability to reverse reasoning questions which can assist inertial thinking to generate more accurate answers. Competitive improvement is observed in DuReader dataset, confirming our hypothesis that bi-directional knowledge helps the QA task. The novel framework shows an interesting perspective on machine reading comprehension and cognitive science.