Nate Kushman, Yoav Artzi, Luke Zettlemoyer, and Regina Barzilay:

We present an approach for automatically learning to solve algebra word problems. Our algorithm reasons across sentence boundaries to construct and solve a sys- tem of linear equations, while simultane- ously recovering an alignment of the vari- ables and numbers in these equations to the problem text. The learning algorithm uses varied supervision, including either full equations or just the final answers. We evaluate performance on a newly gathered corpus of algebra word problems, demon- strating that the system can correctly an- swer almost 70% of the questions in the dataset. This is, to our knowledge, the first learning result for this task.