Abstract
The Crozzle, a complex crossword variant, is described.
The Crozzle consists of entering words from a given lexicon
into an empty grid such that a total score based upon, (i)
the individual scores for the letters at which words intersect and (ii) the total number of words entered, is maximised. This process also determines the number and location
of black squares in the grid. An algorithm, using depth first
recursion and a 2-level look-ahead, to generate solutions to
the unconstrained crossword puzzle problem is modified
and applied to the Crozzle. The solution scores generated
are compared to the score of the solutions produced by
humans. Three data sets are presented as benchmarks for
future research efforts.
The Crozzle consists of entering words from a given lexicon
into an empty grid such that a total score based upon, (i)
the individual scores for the letters at which words intersect and (ii) the total number of words entered, is maximised. This process also determines the number and location
of black squares in the grid. An algorithm, using depth first
recursion and a 2-level look-ahead, to generate solutions to
the unconstrained crossword puzzle problem is modified
and applied to the Crozzle. The solution scores generated
are compared to the score of the solutions produced by
humans. Three data sets are presented as benchmarks for
future research efforts.
| Original language | English |
|---|---|
| Pages (from-to) | 41-48 |
| Journal | Australian Computer Journal |
| Volume | 25 |
| Issue number | 2 |
| Publication status | Published - 1993 |
| Externally published | Yes |