Just as other AI computer games, the most critical function
of a Connect6 program is to search and determine the best
move to fight against the opponent player.
Generally, in a Connect6 AI program, three steps are
required to get the best move. Firstly, make a collection of
empty points of higher values. These points are called
candidate points. Secondly, construct candidate moves with
these candidate points. At last, use search engine to get the
best move from the candidate moves.
Because point evaluation directly affects the quality of
candidate points, and then the quality of candidate moves,
so the algorithm of point evaluation is critical to Connect6
AI programs.
By now, several methods have been reported about point
evaluation [1][2]. These methods generally present one or
more formulae with a lot of parameters, and these formulae
focus on calculating the distances between one center point
and other points of the same line.
Though these methods can finish point evaluation with a
relatively high accuracy, but from the perspective of
practice, the efficiency of these methods is dubious. It is
because that, these formulae are calculated in a real-time
manner without predefined knowledge. This would greatly
cut down the executing speed of the program. Another
shortness of these methods is that, the idea of segment
evaluation is not paid attention to enough, and the
relationship between point evaluation and segment
evaluation is rarely mentioned.
For convenience of the following context, several items are
presented and explained here.
Line, a straight line on the game board which both end
points are all on board edges. The stones on a line maybe of
different color.