## GroupLens NSF GRFP Awards!

We’d like to congratulate two GroupLens Ph.D. students – Hannah Miller (@hannahjean515, website) and Taavi Taijala – on being awarded National Science Foundation Graduate Research Fellowships!

Hannah’s proposal focused on strengthening a social network user’s experience by bootstrapping from previously established social network presence.

Taavi proposed work on better explaining recommender systems output, focusing on the use of analogies to describe recommendations.

There were more than 16,000 applications for the Graduate Research Fellowship nationwide from all areas of science and engineering. The National Science Foundation (NSF) only awarded them to 2,000 students (of which GroupLens members were 0.1%!). The awards financially support 36 months of research and include opportunities for international research experiences.

Congratulations again, Taavi and Hannah!

## Onboarding New users in Recommender Systems

According to the song, “Getting to know you…. getting to know all about you…” is a lot of fun. But when you go to a new doctor, and fill out a 10-page patient intake form so the doctor can get to know you, it’s not so much fun.

The same has been true for recommender systems. A typical experience for new users is to rate a bunch of items to let the system know their preferences. For example, in the MovieLens film recommender, first-time visitors had to rate 15 movies (see the screen below). This process usually required paging through multiple screens, took over 5 minutes, and discouraged some people so much they dropped out before ever making it to the site home page! (more…)

## Congratulations to Ed Chi and Patrick Baudisch!

GroupLens would like to congratulate Ed Chi (a Ph.D. graduate from our lab) and Patrick Baudisch (a former visiting graduate student to our lab) on being named ACM Distinguished Scientists! We wish them all the best and are very proud of their continued accomplishments!

## Preparing for a Google technical interview

GroupLens students and alumni successfully interview at Google on a regular basis. Several current GroupLens students have interned at the company, and our alumni have become Google research scientists and software engineers. I collected the following technical interview preparation tips from Google recruiters and engineers. Please ask your recruiter if you need confirmation of anything below, as the interview structure changes over time. This advice applies to other companies that are interested in similar problems or which hire the same types of engineers. And if you’re interviewing for an internship instead of a full-time job, there will be a different standard; not necessarily harder or easier, just different, based on the type of student you are and your other interests.

Technical interviews are each about 45 minutes long. There is no dress code. You will code on a whiteboard, showing the interviewer your thought process by talking through decisions and assumptions. Occasionally a video chat and collaborative document allows you to interview from a distance, or a piece of paper substitutes for the whiteboard during in-person interviews. Interview topics may cover anything on your résumé, especially where you claim expertise. Fundamental computer science knowledge is required for all engineering roles at Google and will form the basis for almost all interview questions. Google wants to see if you can take a hard, big problem for which you don’t know an obvious solution, and break it down into manageable solvable parts for which you can provide reasonable runtime and space bounds.

## Recommending for new users is surprisingly difficult

For years MovieLens has required users to enter 15 ratings before they are allowed to get personalized recommendations. This design makes sense: how can we make recommendations for a user we know nothing about? That said, we don’t know if this provides users with the best experience. Why should users have to enter fifteen ratings, why not ten or five? What would happen if we let users into the system without any ratings? To answer these questions we need to understand how our algorithms behave for users with very few ratings.

To understand how algorithms behave for users just joining the system, we looked at historic MovieLens ratings. We trained three popular recommender algorithms: ItemItem, UserUser, and SVD on this rating data. While training, we limited some users to have only a small number of ratings. We used the ratings that were not given to the algorithm to measure several things:

• How accurate are the predictions? Can the algorithm accurately predict the user’s future ratings?
• How good are the recommendations? Does the algorithm suggest movies for the user that the user would like?
• What type of recommendations does the algorithm generate? Is there a good diversity of movies? Are the movies popular, or more obscure?