In this paper we represent a new idea of searching related text information on web. The syntactical characters of related keywords and texts are described detail, which do not involve semantics of the keywords. So computers can find related keywords and texts without understanding their meanings. Based on these, an intelligent expansion of query of searching related text information is proposed. Algorithms for such an expansion are provided. The algorithms are effective. The results of experiments of this method we made on internet before show that the algorithms in the paper are powerful and satisfying. The paper concludes with a comprehensive analysis on the results of the experiments.