Learning from Explicit and Implicit Supervision Jointly For Algebra Word Problems

@inproceedings{Upadhyay2016LearningFE,
  title={Learning from Explicit and Implicit Supervision Jointly For Algebra Word Problems},
  author={Shyam Upadhyay and Ming-Wei Chang and Kai-Wei Chang and Wen-tau Yih},
  booktitle={EMNLP},
  year={2016}
}
Automatically solving algebra word problems has raised considerable interest recently. Existing state-of-the-art approaches mainly rely on learning from human annotated equations. In this paper, we demonstrate that it is possible to efficiently mine algebra problems and their numerical solutions with little to no manual effort. To leverage the mined dataset, we propose a novel structured-output learning algorithm that aims to learn from both explicit (e.g., equations) and implicit (e.g… CONTINUE READING

Figures, Tables, Results, and Topics from this paper.

Key Quantitative Results

  • By leveraging both implicit and explicit supervision signals, our final solver outperforms the state-of-the-art system by 3% on ALG514, a popular benchmark data set proposed by (Kushman et al., 2014).

Citations

Publications citing this paper.
SHOWING 1-10 OF 18 CITATIONS

The Gap of Semantic Parsing: A Survey on Automatic Math Word Problem Solvers

  • IEEE transactions on pattern analysis and machine intelligence
  • 2018
VIEW 9 EXCERPTS
CITES BACKGROUND
HIGHLY INFLUENCED

Learning Fine-Grained Expressions to Solve Math Word Problems

VIEW 4 EXCERPTS
CITES BACKGROUND & METHODS
HIGHLY INFLUENCED

A Neural Semantic Parser for Math Problems Incorporating Multi-Sentence Information

  • ACM Trans. Asian & Low-Resource Lang. Inf. Process.
  • 2019
VIEW 1 EXCERPT
CITES BACKGROUND

References

Publications referenced by this paper.
SHOWING 1-10 OF 31 REFERENCES

Parsing Algebraic Word Problems into Equations

  • Transactions of the Association for Computational Linguistics
  • 2015
VIEW 1 EXCERPT