There is a thesis in my University, in which the object is creation of a Semantic Database for the University (and addition to DBPedia I guess). Knowing some basic stuff about how the Semantic Web works, and some basic stuff about OWL, this sounded like something not easy, but also not really hard. But my professor insisted that it will take a lot of time. If someone has knowledge on Semantic Web I would like to hear an opinion. I'm not looking for details and such. Just an opinion on whether you think something like this would be easy or hard based on your experience. Thanks for reading.
Back in my university days I worked together with the Research Group Multimedia Information Systems that focuses on Semantic web.
I will try to answer your question from one perspective, but there are many others possible. I assume that you want to convert some non-semantic texts the University has produced over the years into something more structured. RDF is good for that.
There are many tools for NLP that will help you parse the text. The next step would be to convert the output of the NLP to some form of RDF. That should also be fairly simple. Writing a prototype is as you've put it "not really hard". Why your professor believes that it is lots of work is because it is hard to get useful triples out of text. In DBPedia you can extract the verb of the sentence and make it the predicate. The outgoing link is your object and the referencing page is your subject. Without links that uniquely identify a concept you will fail on homonyms. Somehow you need to bind your text to RDF URIs.
It will take lots of time to clean the data.