A method of representing algebraic age word problems for automated solving has been proposed. Schemas pertinent to the domain have been represented as logical predicates and functions and used as an intermediate representation. Though data-driven methods have dominated the paradigm for solving word problems recently, we demonstrate that representing these word problems formally has advantages. We illustrate that it outperforms a state-of-the-art method in terms of precision and accuracy on our newly curated and publicly available data set. Thus, we show how semantic information can improve some facets of natural language understanding. © 2019, Ohmsha, Ltd. and Springer Japan KK, part of Springer Nature.