Automation Strategies for Unconstrained Crossword Puzzle Generation

TL;DR

This paper presents algorithmic strategies for automatic unconstrained crossword puzzle generation, addressing grid design, word placement, and clue creation, resulting in efficient generation of well-packed puzzles of various sizes.

Contribution

It introduces a comprehensive end-to-end algorithm that combines multiple strategies for unconstrained crossword puzzle generation, including grid resizing and clue generation, which is novel in this domain.

Findings

Strategies enable quick generation of large, well-packed puzzles

Permutation of word sequences significantly affects grid fitting

The combined algorithm produces high-quality puzzles efficiently

Abstract

An unconstrained crossword puzzle is a generalization of the constrained crossword problem. In this problem, only the word vocabulary, and optionally the grid dimensions are known. Hence, it not only requires the algorithm to determine the word locations, but it also needs to come up with the grid geometry. This paper discusses algorithmic strategies for automatic crossword puzzle generation in such an unconstrained setting. The strategies proposed cover the tasks of selection of words from a given vocabulary, selection of grid sizes, grid resizing and adjustments, metrics for word fitting, back-tracking techniques, and also clue generation. The strategies have been formulated based on a study of the effect of word sequence permutation order on grid fitting. An end-to-end algorithm that combines these strategies is presented, and its performance is analyzed. The techniques have been found to be successful in quickly producing well-packed puzzles of even large sizes. Finally, a few example puzzles generated by our algorithm are also provided.