Analyzing and Improving the Network Flow Model in Hex Game System

Abstract

The key problem in determining Hex game ability is to design a satisfactory evaluation function. Since the network flow model can reflect the connectivity of the system, and Hex is a connectivity game, some previous researchers proposed the idea of applying the network flow model to Hex. However, in the actual game, it was found that the game ability was poor, and the satisfactory results were often not given. This paper, first of all, analyzes the shortcomings of the network flow model applied to Hex and summarizes three main reasons for its poor evaluation ability. In order to improve the game ability of the system, this paper proposes a multi-level priority queue evaluation model. The main idea is as follows: firstly, according to the status around the stones, the stones are divided into five kinds of priority and put into the corresponding priority queue. Then the stones of the same priority are sorted by the priority queue custom sorting function. Finally, the best decision is screened by the select function. The experimental and evaluation results show that the game ability of the multi-level priority queue model is significantly better than that of the network flow model. Besides, the new evaluation model provides a new idea for the design of the evaluation function of Hex.