​In October 2011, Bamshad Mobasher (professor, CDM) and Robin Burke (professor, CDM) organized and chaired the fifth ACM International Conference on Recommender Systems (RecSys).* The College of Computing and Digital Media and DaMPA (DePaul Center for Data Mining and Predictive Analytics) were among the event’s sponsors. RecSys is an annual meeting of researchers and practitioners who develop and apply software applications that “recommend” or suggest choices based on preferences a user has already expressed, explicitly or implicitly.

Mobasher, Burke, and Jonathan Gemmell (PhD ’12) discuss DePaul’s prominent role in this burgeoning field.

Mobasher:  In search engines and on websites, recommender systems help people navigate the volume and complexity of information on the Web to find content, people, items, and events of actual or potential interest. The bigger idea behind recommender systems is personalization: over time, an intelligent system can learn from a user’s behavior, and then adapt to become relevant.
In the last few years, recommender systems have become very popular, and everyone’s familiar with their use on websites selling books or music. Amazon actually uses some technology we developed in 2000. Now, the new thing in this area is to make systems more intelligent and to make recommendations more useful by including a context. Say you want to go to a restaurant. Why are you going? Would you go to the same place for a date and for a business meeting? The next generation of recommender systems would “get” context. That’s a difficult problem to solve, but one of the things we’re working on.
Burke: Not a lot of people do research on recommender systems; we’re a small community. One of DePaul’s strength in this area is perspective. We bring different skills to problem solving — Bamshad is a data mining guy; I’m a knowledge base guy. In being able to look at challenges from multiple angles, we can create hybrid recommender systems that sometimes draw on data patterns, sometimes draw on information embedded in the software, or sometimes draw on algorithms.  We’ve found we get better results that way — systems that are capable of more sophisticated and subtle processing.
Mobasher: Another challenge we’re working on is the introduction of novelty. 
Wonder if a recommendation could be not just useful (and often obvious), but surprising? Wonder if the system could suggest something that the user wouldn’t think of on his own but would still like? A system with serendipity is really hard to achieve. We’re also exploring ways to factor in elements of social media — including “hot topics” that emerge, evolve, and dissipate over time — to increase or improve personalization.
Burke: This is definitely an applied field. Recommendations don’t happen in the abstract; they’re driven by consumption. The ultimate test of a system’s success is its use by real people.  Also, every recommendation system is a little different — books are not sweaters are not recipes are not romantic partners.  An endless number of problems for us to solve! It’s not surprising that, since DePaul leads in doing the cutting-edge research, our graduates end up working in the best companies.
Gemmell: At this year’s RecSys Conference, 300 computer scientists from academia and industry presented papers and shared their research, talking about what they’re doing to recognize user preferences. The last line of each presentation was “We’re hiring!”  Each year, the use of recommender systems just increases: it’s a good time to be in this field. At the same time, our recommender systems research group is among the top five internationally. We’re recognized leaders – and that makes it a great time to be in DePaul’s graduate program.”

* ACM (Association for Computing Machinery) is the premier computing professional organization in the world, and RecSys is the top international conference on recommender systems. The Chicago conference attracted participants from such companies as Google, Microsoft, Yahoo, Netflix, Pandora, Motorola, IBM, Comcast, Linkedin, Facebook, Twitter, Telefonica, Technicolor, and many others. Topics at this year’s conference covered ground-breaking work in music recommendation, movie recommendation, “context aware” recommender systems, and personalization in mobile applications.