Machine Learning algorithm for predicting a user's rating on an item?

I’m trying to build an supervised learning algorithm that deals with the regression problem of predicting a user’s rating of a new incoming item based on the user’s previous ratings on items. The ratings will be in the range of 1 – 5 (1 = best, 5 = worst).

Features a item will probably have:

• author (nominal)
• release channel (nominal, can be empty)
• length in minutes (most likely something between 2 and 15, but can be higher)
• release year (nominal)

The outcome of the algorithm should be a rating for an previously unrated item. (Should be a decimal value – for ordering the predictions, like something between 1.0 and 5.0, or normalized 0.0 to 1.0)

What type of algorithm would suit this prediction problem best and how would i implement it?