Mine

basic idea is:
Take 2 0r more word documents as input(any format that is simple to work with) and compare for word matching or check similarity(Cosine similarity is being considered as an option But i have to look into further detail as i dont have much idea rite now.) 

Then represnt these word documents as vertices of graph  and assign weights to edges according to the similarity between two docs.
Visualize uing JUNG.

Visualization is easy as it is done by JUNG API.