Dr.Fill: Crosswords and an Implemented Solver for Singly Weighted CSPs
This work addresses the problem of automated crossword solving for AI and puzzle enthusiasts, representing an incremental advancement with domain-specific applications.
The paper tackles solving American-style crossword puzzles by converting them to weighted constraint satisfaction problems and introduces novel heuristics, a variant of limited discrepancy search, and postprocessing techniques, achieving performance that ranks among the top fifty crossword solvers globally.
We describe Dr.Fill, a program that solves American-style crossword puzzles. From a technical perspective, Dr.Fill works by converting crosswords to weighted CSPs, and then using a variety of novel techniques to find a solution. These techniques include generally applicable heuristics for variable and value selection, a variant of limited discrepancy search, and postprocessing and partitioning ideas. Branch and bound is not used, as it was incompatible with postprocessing and was determined experimentally to be of little practical value. Dr.Fillls performance on crosswords from the American Crossword Puzzle Tournament suggests that it ranks among the top fifty or so crossword solvers in the world.