Semantically Mapping the Web

The millions of web pages populating the internet seem to be unstructured and chaotic, but there are implicit semantic relations between them. In this paper we propose to make explicit the underlying semantic structure of the internet, by measuring joint keyword occurrences in web pages, around our notion of “Semantic Contexts”. As a result, we can draw a “map” of semantic clusters which can be used as a reference for situating individual web pages in a complex semantic space. Further, the methods we propose could be used for disambiguating and refining web search queries, for refining translations, for spam filtering, and in general for semantic-enabling many internet applications.