Reversi/Othello Evaluation Function

[Artificial]

Hey,

I while ago I made a rather crude Reversi AI engine in PHP located here (it's actually very rudimentary - from memory I wrote it in one sitting and didn't really apply any thought to it).

Anywho, I'm taking an AI course this semester, and one of the assessments is to produce a piece of simple AI software. I like working with algorithms, which is why I decided to go with writing an intelligent game algorithm. I was thinking for a while what game to go with (thought about Chess, Poker, Snakes and Ladders, etc.) and then remembered I'd already had a tiny bit of exposure to Reversi, so decided to go with it (though I'm very novice at the game in terms of strategy). Perhaps the most important part of the agent is writing an intelligent evaluation function (I'm confident in my ability to write an alpha-beta pruning algorithm to utilise the evaluation function, and perhaps a dictionary of starting moves), which is why I'm trying to take my time and really nail it.

Brian Rose (2001 world Othello champion) has written a pretty good book about game strategy (found here) which I'm slowly working my way through. However, I remember there were a few people who had a crack at the problem on Gaia (and in so doing would have written an evaluation function), and perhaps there are members (out of the 10 that are still active haha) who are good at the game. In either of those two cases, is there anyone who has any sort of advice to give? In the event I don't get anything, I'm thinking of perhaps writing an agent to create/refine the evaluation function. I don't really want to just search Google for a solution, as I'd like to do all of this without looking at 3rd party code (as that would defeat the purpose - I know if I saw code it'd influence me right away).

Subject coordinator suggested I brainstorm with likeminded individuals, which is why I decided to give it a go here, haha. We'll see.

[The Unintelligible]

Unfortunately I don't have much advice to give you. I agree however that viewing external code happens to be both consciously and subconsciously influential, so Google solutions is an obvious no-go.

While my problem solving is still quite unpolished, I love algorithms. Especially ones that involve more logic than direct math/computation. AI is a pretty interesting topic in Computer Science, though I've never really done more than dabble in it. Playing around with Cleverbot always puts that fact in perspective for me.

But perhaps I'll have more sentiments to share after doing more research myself. At any rate, good luck and have fun.