KG^2: Learning to Reason Science Exam Questions with Contextual Knowledge Graph Embeddings

Abstract

The AI2 Reasoning Challenge (ARC), a new benchmark dataset for question answering (QA) has been recently released. ARC only contains natural science questions authored for human exams, which are hard to answer and require advanced logic reasoning. On the ARC Challenge Set, existing state-of-the-art QA systems fail to significantly outperform random baseline, reflecting the difficult nature of this task. In this paper, we propose a novel framework for answering science exam questions, which mimics human solving process in an open-book exam. To address the reasoning challenge, we construct contextual knowledge graphs respectively for the question itself and supporting sentences. Our model learns to reason with neural embeddings of both knowledge graphs. Experiments on the ARC Challenge Set show that our model outperforms the previous state-of-the-art QA systems.

References in corpus (11)

• Neural Message Passing for Quantum Chemistry
• Bidirectional Attention Flow for Machine Comprehension
• SQuAD: 100,000+ Questions for Machine Comprehension of Text
• Discriminative Embeddings of Latent Variable Models for Structured Data
• Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge
• A large annotated corpus for learning natural language inference
• Question Answering via Integer Programming over Semi-Structured Knowledge
• A Decomposable Attention Model for Natural Language Inference
• Learning Symmetric Collaborative Dialogue Agents with Dynamic Knowledge Graph Embeddings
• Answering Complex Questions Using Open Information Extraction
• Markov Logic Networks for Natural Language Question Answering

Cited by in corpus (11)

• Multi-domain Dialogue State Tracking as Dynamic Knowledge Graph Enhanced Question Answering
• Improving Question Answering with External Knowledge
• Scalable Multi-Hop Relational Reasoning for Knowledge-Aware Question Answering
• Improving Question Answering by Commonsense-Based Pre-Training
• Answering Science Exam Questions Using Query Rewriting with Background Knowledge
• ExplanationLP: Abductive Reasoning for Explainable Science Question Answering
• Learning Contextualized Knowledge Structures for Commonsense Reasoning
• Case-Based Abductive Natural Language Inference
• Encoding Explanatory Knowledge for Zero-shot Science Question Answering
• Careful Selection of Knowledge to solve Open Book Question Answering
• Answering questions by learning to rank -- Learning to rank by answering questions