I was planning on experimenting with a similar thing for a university assignment. Though my main focus was going to be indexing links that are posted. Ie, on twitter if someone links to an article on continuous integration with teamcity, but just says 'great CI server' in their tweet, my search engine would still return resutls for 'teamcity' or 'continuous integration' as its indexing the actual links as well.

Perhaps Greplin could consider adding similar functionality?