Finding Steiner trees for web information retrieval

Web pages and hyperlinks in a web site are represented graph theoretically to generate optimal structured web objects for more convenient and effective searching of relevant web contents. Mathematical models based on Linear programming for web site are developed to structure web objects so that most relevant web contents can be reachable more efficiently with the optimal-structured web objects. In the web structuring model, arc weights that represent weights for hyperlinks are used to rank the degrees of the relevance among web objects. We also developed a tree-filtering algorithm and a top_k Steiner tree algorithm that can be used to provide search recommendations in practical applications. With real web sites´ information, some examples are given for more detail explanations of the structured web search mechanism.