If you're building a predictive model, you need training data.  Hopefully  hundreds or more of job descriptions  and then you score your new 2 new college programs with that model that you built.   If you do no have training data. Try building user-defined topics that is based on your domain knowledge and use the topic assignment as your classification. SAS also has a product called Content Categorization that is explicitly designed for this.   Sorry, no immediate papers on "hierarchical classification", but you can google that  phrase to find the challenges of it and the approaches people use.     ... View more