A Bottom-Up DAG Structure Extraction Model for Math Word Problems
Published in AAAI, 2020
Recommended citation: Yixuan Cao, Feng Hong, Hongwei Li, and Ping Luo. A Bottom-Up DAG Structure Extraction Model for Math Word Problems. In AAAI, 2021. https://doi.org/10.1609/aaai.v35i1.16075
Automatically solving math word problem is a long studied problem in artificial intelligence research area.
We find that this problem can be viewed as a special case of nested relation extraction problem, as the math expresion and equation system is intrinsically a structure.
So, we borrow the bottom-up layer-by-layer extraction model from nested relation extraction, and it works well on both arithmetic and algebra problems. It achieves SOTA performance on MATH23K and DRAW1K (without external data).