Abstract: Vector Space Model is a important model in information retrieval, traditional Vector Space Model take feature term frequence and document frequence into account, regardless of the location feature term appears, which is a signiifcant information. Considering the problem above, after turn document into Document Object Model, this paper add a ratio to weight of feature term based on different location it appears to inlfect different ability of feature term that appears in different location in expressing main idea of the document, thus improve ranking result of documents returned and users’ retrieving work. Simulation experiment manifests the advantage of the solution above over traditional VSM.