Create and evaluate algorithms to match Internships to Interests

[philipmjohnson]

Please do work for this task in a branch called issue-765.

We need to determine which Interests are associated with a given Internship. This can be complicated to do well.

One way to proceed is to create a standalone JSON file containing a few thousand listings, and another file with the current set of interest descriptions in RadGrad, and then build algorithm(s) to map internship listings to one or more RadGrad Interests.

We'll start with a default, "dumb" algorithm which just does some glorified regex matching, and then allow folks to try to beat it with NLP or other classification algorithms. We'll grade the results by running the algorithms over a test set which used an "oracle" (basically some humans) to establish the ground truth mapping.

This needs to work for both Computer Science and Computer Engineering internships and interests.