Microsoft Download Center Archive

Diverse Algebra Word Problem Dataset with Derivation Annotations

  • Published:
  • Version: 0.7
  • Category: other
  • Language: English

This dataset provides training and testing examples for solving algebra word problems automatically.

This is a public release of the dataset corresponding paper "Annotating Derivations: A New Evaluation Strategy and Dataset for Algebra Word Problems". It consists of over 2000 algebra word problems. Each word problem is annotated with the full derivation (template + alignments) of the relevant equations from the word problem. Please refer to the paper for details.

Contents

  1. 1000 new training/testing data with diverse templates and narratives crawled from algebra.com. (DRAW dataset in the paper)
  2. Word problems from http://groups.csail.mit.edu/rbg/code/wordprobs/ annotated using our proposed schema. (Alg-514 in the paper)
  3. Word problems from linear-T2 subset from http://research.microsoft.com/en-us/projects/dolphin/ annotated using our proposed schema. (Dolphin-L in the paper)

Templates annotated in all the above datasets were globally reconciled. The cross validation splits and train-test splits used in the papers are also provided (Thanks to the respective authors for sharing the splits).
If you found the dataset useful, please support our work by citing our paper. Please email the authors if you find any problems.

Files

Status: Live

This download is still available on microsoft.com. The downloads below will come directly from the Microsoft Download Center.
Files
0.7.zip

System Requirements

Operating Systems: Windows 10, Windows 7, Windows 8

  • Windows 7, Windows 8, or Windows 10

Installation Instructions

  • Click Download and follow the instructions.
This page was generated from a snapshot of the Microsoft Download Center made on .
Report a problem