I'm planning my next project which will be a tycoon like strategy game. Had my first contact with prolog last week and really liked what i saw. Since i like to try new stuff i would like to use prolog or a similar tool in my project. During my research i also found out about clips, soar, jess... Apparently Prolog uses backward chaining, clips and soar use forward chaining. I'm not quite sure i understand the difference. If i'm correct, with backward chaining you try to find the best solution to a problem. With forward chaining you trigger events that change the data and end up getting one of many possible solutions. Is that correct? In my case(a strategy game) what would be the best solution? Have you ever used these kind of tools in more complex games? Would like to know about your experiences with prolog or other logic programming languages.

Depends on the kinds of problems your AI has to solve. There are a lot of things you can look at. Could it be that a simple optimized minimax type thing is what you're looking for?

Forward chaining means that you try to combine all known facts to derive sub-goals, which allow you to derive even more sub-goals. You do this until one of the derivatives is your goal. You go from the question to the conclusion.

Backward chaining works backward. You start with the problem and look at the prerequisites. Then you look at how to satisfy those requirements. The methods to do so have prerequisites of their own, etc. You keep finding all necessary prerequisites until you have found all prerequisites to be known facts. You work from the conclusion back to the question.