A Retrieval Model for Automatic Resolution of Crossword Puzzles in Italian Language

In this paper we study methods for improving the quality of automatic ex- traction of answer candidates for an ex- tremely challenging task: the automatic resolution of crossword puzzles for Italian language. Many automatic crossword puz- zle solvers are based on database system accessing previously resolved crossword puzzles. Our approach consists in query- ing the database (DB) with a search engine and converting its output into a probability score, which combines in a single scoring model, i.e., a logistic regression model, both the search engine score and statisti- cal similarity features. This improved re- trieval model greatly impacts the resolu- tion accuracy of crossword puzzles.