I have difficulty to learn about ant colony algorithm (ACO), I have read about generating crossword game using (genetic algorithm) GA.I Know both of GA ant ACO usually used for optimization, but my question is Is possible if create crossword game generator using ACO and how it can work for optimization in this case?

  • 5
    Would you mind editing the question and providing a few links, for those of us unfamiliar with the Ant Colony Algorithm? – Mason Wheeler Mar 24 '13 at 17:53
  • What type of crosswords do you want to generate? Why do you want to use ACO? It's probably doable, but it doesn't sound easy. I wouldn't use that approach. – iCanLearn Mar 24 '13 at 18:16
  • @MasonWheeler: OP might be referring to this: en.wikipedia.org/wiki/Ant_colony_optimization_algorithms – FrustratedWithFormsDesigner Mar 24 '13 at 18:36
  • Like this website do crosswordpuzzlemaker.org And somebody have create it using python bryanhelmig.com/python-crossword-puzzle-generator. But, there is no body have create generator using ACO. – Benhard Jei Mar 24 '13 at 19:23
  • Sharing your research helps everyone. Tell us what you've tried and why it didn’t meet your needs. This demonstrates that you’ve taken the time to try to help yourself, it saves us from reiterating obvious answers, and most of all it helps you get a more specific and relevant answer. Also see How to Ask – gnat Aug 8 '13 at 21:32

The problem you're having with ant colony algorithms and the reason that no one has applied it to solving crosswords is because ACO poorly suited for the problem of crosswords.

Artificial intelligence is a MASSIVE field of study. There are a ton of methods and approaches. There are also quite a lot of problems out there that can have AI applied to them. Arguably, all of them per the definition of "problem".

For some problems, we have observed how nature solves them and we strive to emulate and improve on those methods. Things like genetic algorithms emulates the process of evolution. Neural networks emulates brains. ACO emulates, you know, ants. And ACO helps us figure out how to solve hard problems like finding the shortest path. But it's probably not the best method for object recognition or finding what words fit in a crossword puzzle. I can't really think of any way to correlate finding the shortest path with finding how to fit words together.

It's tempting to think that neural networks, genetic algorithms, machine learning, genetic programming, or even ant colony optimization are general solutions that can solve everything, given enough tuning, abstraction, and processing time. But that's just not the case.

tl;dr No.

| improve this answer | |
  • One big problem with some sorts of AI techniques modeled on nature is that they derive some of their power from massive parallelization. (This is true of neural networks and ant colony optimization, and to a lesser extent, genetic programming.) You completely lose that when you model it in a single process. – Gort the Robot Aug 8 '13 at 23:53

Not the answer you're looking for? Browse other questions tagged or ask your own question.