This paper researched key techniques of topic-focused web crawler at first, then designed and implemented a crawler system by using improved slef-adapted vector space model. It analysised documents both in text and links. As the same time, this paper also comed up with a web search stategy based on gene factor combined with manully control. This strategy can solve the problem of searching path blocked. In the end, we provide some experiment results to prove the feasibility and advantages of our system from recall ratio and precision ratio.