E-GPS: Explainable Geometry Problem Solving Via Top-Down Solver and Bottom-Up Generator

Wenjun Wu,Lingling Zhang,Jun Liu,Xi Tang,Yaxian Wang,Shaowei Wang,QianYing Wang

DOI: https://doi.org/10.1109/cvpr52733.2024.01312

2024-01-01

Abstract:Geometry Problem Solving has drawn growing attention recently due to its application prospects in intelligent ed-ucation field. However, existing methods are still inade-quate to meet the needs of practical application, suffering from the following limitations: 1) explainability is not en-sured which is essential in real teaching scenarios; 2) the small scale and incomplete annotation of existing datasets make it hard for model to comprehend geometric knowl-edge. To tackle the above problems, we propose a novel method called Explainable Geometry Problem Solving (E-GPS). E-GPS first parses the geometric diagram and prob-lem text into unified formal language representations. Then, the answer and explainable reasoning and solving steps are obtained by a Top-Down Problem Solver (TD-PS), which innovatively solves the problem from the target and focuses on what is needed. To alleviate the data issues, a Bottom-Up Problem Generator (BU-PG) is devised to augment the data set with various well-annotated constructed geome-try problems. It enables us to train an enhanced theorem predictor with a better grasp of theorem knowledge, which further improves the efficiency ofTD-PS. Extensive experi-ments demonstrate that E-GPS maintains comparable solving performances with fewer steps and provides outstanding explainability.