Signature word extracting research based on web metadata

Signature word of the text extracting is a useful technique which can abstract Web page text, as well as it provides technical support for text classification, information extraction and other related tasks. This paper attempts to partition document into a hierarchical structure by parsing the semantic distance between each adjacent paragraph in the web page content. On the basis of the hierarchical structure we use the metadata and special tags of the HTML to design a weighting function by considering the factor of the frequency, length and location of the word. Finally, various location factors on the system´s contribution are comparative analyzed.