Michael is going to be publishing a paper at the conference RecSys, or ACM Recommender Systems, this fall. His research deals with recommender systems, but even if you don’t study computer science, you probably deal with them.

Recommender systems are what their name indicates: systems that make recommendations. If you’ve purchased something online, a recommender system was probably used to encourage you to buy more by showing a few products for your consideration. Twitter offers suggestions for people to follow. If you read an article online, there is a good chance that it will suggest something else to see when you’re finished.

Some of these things can be done manually. I sometimes add recommendations for related items at the bottom of posts (like this one). If I didn’t, you could find similar results by looking at the tags for the post.

What a recommender system does is automate the process. Rather than me picking out a few articles based on what I remember having written and a quick glance through tags, a recommender system could select posts based on criteria such as tags and number of visits. With larger systems, manual recommendations become impractical, and the automated recommendations become better options.

Different systems use different information to make the decisions. A store might suggest items that people purchased when they purchased an item that you’re looking at. A media recommender might take your ratings and compare them to ratings made by others to suggest books, movies, or music that were rated highly by people who have similar taste. A social network might recommend following people followed by the people you follow. A news site might recommend headlines that are currently getting a high number of hits. Dating networks might make suggestions based on common interests and character traits.

Recommender systems can be useful. Good recommendations can be useful for both the person recommending and the person receiving the recommendation. If you buy something that is genuinely useful, it is good for both the buyer and the seller. However, you should still be aware of their impact. Do recommendations make you want items you do not need and cannot afford? Do they cause you to waste time on the internet?

Like all technology, recommender systems can be valuable or harmful. While you may not have control over who uses them, you can be aware of them and how they impact you.

Related:

Once Upon a Twitter

Five People to think about when considering privacy

Movielens is a movie recommender website which Michael’s lab uses for research (Michael isn’t directly involved).