SBI-RAG: Enhancing Math Word Problem Solving for Students through Schema-Based Instruction and Retrieval-Augmented Generation
Prakhar Dixit, Tim Oates
2024-10-18
Summary
This paper discusses SBI-RAG, a new framework that helps students improve their skills in solving math word problems by using schema-based instruction combined with a large language model.
What's the problem?
Many students find math word problems challenging because they struggle to identify important information and decide which math operations to use. This can lead to confusion and incorrect answers, making it difficult for them to succeed in math.
What's the solution?
To help students overcome these challenges, the authors developed the SBI-RAG framework, which combines schema-based instruction (SBI) with a large language model (LLM). SBI teaches students to categorize math problems based on their structure, which helps them understand how to approach different types of problems. The LLM assists in generating solutions step-by-step, guiding students through the reasoning process. The authors tested this framework on the GSM8K dataset and found that it improved students' ability to solve problems accurately and clearly.
Why it matters?
This research is important because it provides a new way to teach math problem-solving that could benefit many students, especially those who struggle with traditional methods. By combining effective teaching strategies with advanced technology, SBI-RAG has the potential to enhance learning outcomes in mathematics and make it easier for students to grasp complex concepts.
Abstract
Many students struggle with math word problems (MWPs), often finding it difficult to identify key information and select the appropriate mathematical operations.Schema-based instruction (SBI) is an evidence-based strategy that helps students categorize problems based on their structure, improving problem-solving accuracy. Building on this, we propose a Schema-Based Instruction Retrieval-Augmented Generation (SBI-RAG) framework that incorporates a large language model (LLM).Our approach emphasizes step-by-step reasoning by leveraging schemas to guide solution generation. We evaluate its performance on the GSM8K dataset, comparing it with GPT-4 and GPT-3.5 Turbo, and introduce a "reasoning score" metric to assess solution quality. Our findings suggest that SBI-RAG enhances reasoning clarity and problem-solving accuracy, potentially providing educational benefits for students