GOLD : geometry problem solver with natural language description

Zhang, Jiaxin and Moshfeghi, Yashar (2024) GOLD : geometry problem solver with natural language description. In: 2024 Annual Conference of the North American Chapter of the Association for Computational Linguistics, 2024-06-16 - 2024-06-21.

[thumbnail of Zhang-Moshfeghi-NAACL-2024-GOLD-Geometry-Problem-Solver-with-Natural] Text. Filename: Zhang-Moshfeghi-NAACL-2024-GOLD-Geometry-Problem-Solver-with-Natural.pdf
Accepted Author Manuscript
Restricted to Repository staff only until 21 June 2025.
License: Strathprints license 1.0
Download (1MB) | Request a copy

Abstract

Addressing the challenge of automated geometry math problem-solving in artificial intelligence (AI) involves understanding multimodal 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.

ORCID iDs

Zhang, Jiaxin ORCID logoORCID: https://orcid.org/0000-0001-7355-7975 and Moshfeghi, Yashar ORCID logoORCID: https://orcid.org/0000-0003-4186-1088;
CORE (COnnecting REpositories)