Designing a live recommendation engine

One widely used application of recommendation engines is to give an e-commerce site's visitor, as he/she performs various actions, such as clicking on products, buying goods, or performing searches, some relevant advice about items they have not visited yet but would be interested in. This way, e-commerce site owners increase the odds of purchases and drive more sales.

There are various ways of designing recommendation engines. In technical literature, we often find them classified in two great families. First, the engines that take into account the nature of the items being visited, that is, they use knowledge about the very products to recommend something - for instance, one would recommend batteries to a user who has bought a drummer bunny toy. These recommenders are labeled as content based and require specific knowledge about the products they deal with, and so, while providing a rather good outcome, they could not be easily adapted to new...