I am doing a project for my CS class and I was wondering if the following would work.
I have 50 different people who have rated the same 50 books. The rating system is as follows:
negative 5 = hate it
negative 3 = dislike it
0 = I have not read it
2 = It is ok
3 = I like it
5 = I love it
My question is as follows. Can I replace the numbers into letters and calculate the relative entropy between the users to find the similarity and produce the best match ? I don't know enough about relative entropy but from what I have read so far it seems to be ok to do so, but just wanted more clarification over it.
Thanks for help!