Learning to Automatically Solve Logic Grid Puzzles

Logic grid puzzle is a genre of logic puz-zles in which we are given (in a natural language) a scenario, the object to be de-duced and certain clues. The reader has to figure out the solution using the clues provided and some generic domain con-straints. In this paper, we present a sys-tem, LOGICIA, that takes a logic grid puz-zle and the set of elements in the puz-zle and tries to solve it by translating it to the knowledge representation and rea-soning language of Answer Set Program-ming (ASP) and then using an ASP solver. The translation to ASP involves extrac-tion of entities and their relations from the clues. For that we use a novel learn-ing based approach which uses varied su-pervision, including the entities present in a clue and the expected representa-tion of a clue in ASP. Our system, LO-GICIA, learns to automatically translate a clue with 81.11 % accuracy and is able to solve 71 % of the problems of a corpus. This is the first learning system that can solve logic grid puzzles described in natu-ral language in a fully automated manner. The code and the data will be made pub-licly available a