Learning algorithms applied to digital marketing

The online recommendation mechanisms used by firms like Amazon and Netflix are among the most common examples of the usability of learning algorithms. Using data collected from millions of buyers and users, learning systems are able to predict items you might want, according to your previous purchases or viewing habits.

In addition, learning algorithms may act directly on search engines. Google, Microsoft Bing and other search engines use learning algorithms to improve their minute-by-minute capabilities. They are able to collect and analyse data on which links users click in response to queries to improve their results to posterior queries.

According to Pedro Domingos in the book “The revolution of the master algorithm”, advertising on the web is just the beginning of a much larger phenomenon. In all markets, producers and consumers have to establish a connection before the transaction may take place. In pre-internet times, the main obstacles were physical. We could only buy books at our local bookstore, and it had limited shelf space. But when we can download any book to our e-reader whenever we want, the problem becomes the overwhelming number of choices.

How to look up on the shelves of a bookstore that has millions of book titles for sale? The same applies to other items such as videos, music, news, tweets, bloggers or simple web pages. And It may also be applied to all goods and services that may be made available at a distance: shoes, flowers, appliances, hotel rooms or classes. How do we find and compare them? This is one of the problems that define the information age, and machine learning may a big part of the solution. The learning algorithms may guarantee clicks, which are potential business opportunities and thus growth and visibility for a company.

Another potential of using automatic learning is the personalization of educational services. Web-based learning systems still offer educational resources in the same way for students with different profiles. The personalization of e-learning, generally, relying on explicit information reached based on the recent browsing history of browsers, allows exploring similarities and distinctions between user preferences and between the contents of learning resources.