Published on Sun May 30 2021

GeoQA: A Geometric Question Answering Benchmark Towards Multimodal Numerical Reasoning

Jiaqi Chen, Jianheng Tang, Jinghui Qin, Xiaodan Liang, Lingbo Liu, Eric P. Xing, Liang Lin
0
0
0
Abstract

Automatic math problem solving has recently attracted increasing attention as a long-standing AI benchmark. In this paper, we focus on solving geometric problems, which requires a comprehensive understanding of textual descriptions, visual diagrams, and theorem knowledge. However, the existing methods were highly dependent on handcraft rules and were merely evaluated on small-scale datasets. Therefore, we propose a Geometric Question Answering dataset GeoQA, containing 5,010 geometric problems with corresponding annotated programs, which illustrate the solving process of the given problems. Compared with another publicly available dataset GeoS, GeoQA is 25 times larger, in which the program annotations can provide a practical testbed for future research on explicit and explainable numerical reasoning. Moreover, we introduce a Neural Geometric Solver (NGS) to address geometric problems by comprehensively parsing multimodal information and generating interpretable programs. We further add multiple self-supervised auxiliary tasks on NGS to enhance cross-modal semantic representation. Extensive experiments on GeoQA validate the effectiveness of our proposed NGS and auxiliary tasks. However, the results are still significantly lower than human performance, which leaves large room for future research. Our benchmark and code are released at https://github.com/chen-judge/GeoQA .

Thu May 30 2019
NLP
MathQA: Towards Interpretable Math Word Problem Solving with Operation-Based Formalisms
A large-scale dataset of math word problems and an interpretable neural math problem solver that learns to map problems to operation programs. We introduce a new representation language to model precise operation programs corresponding to each math problem.
0
0
0
Wed Jun 02 2021
Artificial Intelligence
Sequence to General Tree: Knowledge-Guided Geometry Word Problem Solving
0
0
0
Thu Jan 28 2021
NLP
Weakly Supervised Neuro-Symbolic Module Networks for Numerical Reasoning
Neural Module Networks (NMNs) have been quite successful in incorporating explicit reasoning as learnable modules in various question answering tasks. We propose Weakly-Supervised Neuro-Symbolic Module Network (WNSMN) trained with answers as the sole supervision for numerical reasoning based MRC.
0
0
0
Sat Jul 03 2021
NLP
Neural-Symbolic Solver for Math Word Problems with Auxiliary Tasks
Neural-Symbolic Solver (NS-Solver) consists of a problem reader to encode problems, a programmer to generate symbolic equations, and a problem executor to obtain answers. The solver is also optimized via 4 new auxiliary objectives to enforce different symbolic reasoning.
0
0
0
Fri Mar 05 2021
Machine Learning
Measuring Mathematical Problem Solving With the MATH Dataset
MATH is a new dataset of 12,500 challenging mathematics problems. Each problem in MATH has a full step-by-step solution which can be used to teach models to generate answer derivations and Explanations.
9
122
673
Mon May 10 2021
NLP
Inter-GPS: Interpretable Geometry Problem Solving with Formal Language and Symbolic Reasoning
Interpretable Geometry Problem Solver (Inter-GPS) parses the problem text and diagram into formal language automatically via rule-based text parsing and neural object detecting. Unlike implicit learning in existing methods, Inter-Gps incorporates theorem knowledge as conditional rules.
0
0
0