Learning to reason with a restricted view

The Learning to Reason framework combines the study of Learning and Reasoni ng into a single task. Within it, learning is done specifically for the pur pose of reasoning with the learned knowledge. Computational considerations show that this is a useful paradigm; in some cases learning and reasoning p roblems that are intractable when studied separately become tractable when performed as a task of Learning to Reason. In this paper we study Learning to Reason problems where the interaction wi th the world supplies the learner only partial information in the form of p artial assignments. Several natural interpretations of partial assignments are considered and learning and reasoning algorithms using these are develo ped. The results presented exhibit a tradeoff between learnability, the str ength of the oracles used in the interface, and the range of reasoning quer ies the learner is guaranteed to answer correctly.