To Örebro University

The ability to solve probability word problems suchas those found in introductory discrete mathematicstextbooks, is an important cognitive and intellec-tual skill. In this paper, we develop a two-step end-to-end fully automated approach for solving suchquestions that is able to automatically provide an-swers to exercises about probability formulated innatural language.In the first step, a question formulated in naturallanguage is analysed and transformed into a high-level model specified in a declarative language.In the second step, a solution to the high-levelmodel is computed using a probabilistic program-ming system.On a dataset of 2160 probability problems, oursolver is able to correctly answer 97.5% of thequestions given a correct model. On the end-to-end evaluation, we are able to answer 12.5% of thequestions (or 31.1% if we exclude examples notsupported by design).