For the final project of a search engines class, I developed a method for evaluating webpages based on their semantic relevance. Together with my two partners we built a system that used WordNet to cluster sentences into semantically related categories. The categories represented sentences conveying the same information. The more sentences in any give category, the more support those sentences have across WebPages, and the more likely they will be important sentences. Then, webpage ranking is simple a matter of choosing the most well supported documents -those with the highest concentration of well supported sentences. Unfortunately, this process was computational very demanding and isn't a viable option for search engines at this time. However, it was well worth exploring as a possible avenue for research as computation power may soon catch up.