This is the final project for CSSE474 - Hidden Markov Models. The idea for this project is to use HMMs to cluster relations. A relation is a triple, of the form (object, relation, object). The data set we will use is a set of extractions from web text using the ReVerb relation extractor. Our system will use HMMs to group (or cluster) these relations together by meaning.

For example, a file with relations might look like the following:

Abe Lincoln	      became	   president
Facebook	      is	   a waste of time
Lincoln	      was	   a president
Abraham Lincoln      was elected  president
Facebook	      is clearly   a huge time-waster

The output of our system might look like:

Abe Lincoln	      became	   president
Lincoln	      was	   a president
Abraham Lincoln      was elected  president

Facebook	      is	   a waste of time
Facebook	      is clearly   a huge time-waster