GOLD: Geometry Problem Solver with Natural Language Description

TL;DR

GOLD is a novel AI model that improves automated geometry problem-solving by better interpreting diagrams and converting geometric relations into natural language, leading to significant accuracy gains over previous methods.

Contribution

The paper introduces GOLD, a new model that enhances geometric relation extraction and natural language conversion, outperforming existing models on multiple geometry problem datasets.

Findings

GOLD outperforms Geoformer by 12.7% and 42.1% in accuracy on key subsets.

GOLD surpasses PGPSNet by 1.8% and 3.2% on respective datasets.

The approach effectively combines diagram interpretation with large language models.

Abstract

Addressing the challenge of automated geometry math problem-solving in artificial intelligence (AI) involves understanding multi-modal information and mathematics. Current methods struggle with accurately interpreting geometry diagrams, which hinders effective problem-solving. To tackle this issue, we present the Geometry problem sOlver with natural Language Description (GOLD) model. GOLD enhances the extraction of geometric relations by separately processing symbols and geometric primitives within the diagram. Subsequently, it converts the extracted relations into natural language descriptions, efficiently utilizing large language models to solve geometry math problems. Experiments show that the GOLD model outperforms the Geoformer model, the previous best method on the UniGeo dataset, by achieving accuracy improvements of 12.7% and 42.1% in calculation and proving subsets. Additionally, it surpasses the former best model on the PGPS9K and Geometry3K datasets, PGPSNet, by obtaining accuracy enhancements of 1.8% and 3.2%, respectively.