Solving arithmetic word problems by scoring equations with recursive neural networks

作者：

Highlights：

• We introduce an algorithm to solve arithmetic word problems.

• We propose a tree-based neural model to encode arithmetic expressions.

• Tree-based approach outperforms current state-of-the-art by 3%.

• Our approach outperforms current state-of-the-art by 15% on complex problems.

• Tree-LSTM outperforms linearly-structured LSTM on complex problems.

摘要

•We introduce an algorithm to solve arithmetic word problems.•We propose a tree-based neural model to encode arithmetic expressions.•Tree-based approach outperforms current state-of-the-art by 3%.•Our approach outperforms current state-of-the-art by 15% on complex problems.•Tree-LSTM outperforms linearly-structured LSTM on complex problems.

论文关键词：Arithmetic word problems,Recursive neural networks,Information extraction,Natural language processing

论文评审过程：Received 8 November 2019, Revised 6 January 2021, Accepted 8 February 2021, Available online 20 February 2021, Version of Record 5 March 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.114704