While the connections it found made sense it was effectively no more than string matching in most cases. My hypothesis was that using something called a "dependency grammar" I could quickly extract relations between nouns in a sentence. My running every sentence in a document through that I could construct a map/reduce algorithm to build a map of entity relations in a document and further reduce that across a set of documents.

I used the Wikipedia article on World War II and every article linked to from there as my input set.

Though I'm done with the "thesis" portion of it, I'm working on ways to rewrite it. I have a few ideas involving machine learning that I think would significantly improve my results.

I wrote it all in Scala though Play was only used to display results from the database.