In this experiment, we build an automatic librarian that can classify books for students of different year levels, based on certain book characteristics.

When you describe a book to your machine learning model (e.g. how many pages, lines per page and pictures it has), it will try to predict who that book might be suitable for. You will teach the computer to recognise books of different reading levels by giving it examples of each. When you scroll through the input patterns, you see examples that we have already defined for you. For example, a book with few pages and few lines per page but with many pictures is suitable for the F-2 year level.

Before we can use our AI librarian, we need it to learn the patterns. Click on the 'Start Learning' Button to start the learning process.

Once learning is complete, drag the three sliders for pages, lines per page and pictures below. This creates new combinations for fictitious books. Do you agree with the recommendations by the AI librarian?

By dragging the three sliders you create all sorts of different combinations of the three input parameters that the AI librarian has never seen before. Most recommendations will make sense but some probably won't. Still, it is quite good, considering that you didn't have to write a single line of code.

This scenario has been inspired by and developed in cooperation between the Digital Technologies Institute, Australiaand Apps For Good, United Kingdom.