in reply to Re: Improving this word search solver
in thread Improving this word search solver

The online tool is correct, especially since it outputs the right hidden message and mine is only close. I have no idea of their algorithm because it's closed source and the only other place I could find a similar algorithm was that codegolf stackexchange link.
  • Comment on Re^2: Improving this word search solver

Replies are listed 'Best First'.
Re^3: Improving this word search solver
by Anonymous Monk on Jan 21, 2015 at 02:03 UTC

    I don't see a description of their solver on their website at the moment, and my French is no good... If you really want your solving rules to match the "dCode" version exactly, you'll have to figure them out somehow - ask the people who wrote them, figure them out experimentally, ... in that respect my crystal ball is not any clearer than yours ;-)

    Remember, the "rules" for solving the puzzle are not necessarily the same as an algorithm. The rules will, in English (or French?), say things like "longer words match first", "the search proceeds from left to right, top to bottom", "each word matches only once", "matches across have higher priority than diagonal", etc. Maybe if you ask the people who implemented the solver, they can give you these rules, without revealing their code.

      I have implemented some rules, and you can see I look for longer words first. I have been tinkering with the code, but I can't replicate the results yet. I've also tried allowing only one match per word, but that didn't help this test case.
      I don't see a description of their solver on their website at the moment, and my French is no good...

      I looked at the French version, I haven't found any further detail on the detailed rules.

      Je suis Charlie.