Research And Optimization Of The Computer-game System Of Connect6

Computer game is one of the most important and challenge subject of AI area, which contributes many important theories and methods to the research of AI and achieves so many production that wildly influences the society and science. Computer game is the drosophila in the AI field and the chess game is one of the standard problem of the computer game which contains the applications of all kinds of search algorithm, pattern recognition and intelligence method. So In the past half century, scholars from all over the world poured numerous time and energy into the research of computer game such as Othello,checker,chess,Chinese chess,Go-Moku and I-go, and get a good many of shocking results, for example the world chess game champion Kasparov lost his match in the competition with the super computer "Deep Blue" in 1997, which gave the world a big shock. Now the AI of many other chess games has already reached to the level of world champion.Comparing with chess like Go-Moku and Chinese chess, which have the mature computer-game algorithm now, there are few researches on connect6, a newly invented chess. Although the connect6 game is now appear in some forum and matches on the Internet, it's players are just people. Professor Yicheng Wu, the inventor of the connect6 has given the fairness problem's solution and threats-based winning strategy, but there is no more deep public research of the computer-game problem of connect6, neither a completely solution of the computer game problem of connect6 is given. This paper makes right a future study on the computer-game of connect6.Based on the former research, in which we divided the computer-game system of connect6 into four parts: searching engine, move generator, evaluation function and starting steps database, this paper is mainly about the farther optimization of the four parts and the whole system. In move generator parts, we combine heuristic information with classical search algorithm, then we limit the search area based on theory of "breakaway moves", as a result, the efficiency of search process are evidently improved. As for the searching engine part, by adding special strategy into the search progress, the search process is more intelligent. Last, as the import of the characteristic code of chess model, the applying of genetic algorithm on the evaluation function is come into ture. A lot of experts' chess examples are stored in the starting steps database to avoid the strategy mistakes caused by the insufficiency of searching depth, which also brings lots of efficiency in actual playing.As conclusion the testing and valuing of the computer game system of connect6 is given at last, include the precision of the evaluation function, the efficiency of the searching algorithm and the efficiency of the whole system.