Mustafa Bilgic

@ChessProblem And, one day, a pawn underpromotes to a bishop …

good morning

@berty38 Do you play at lichess?

I just played terribly and lost this chess game. Or did I? White to play.

Congratulations *Dr.* Ping Liu! 👏👏👏 #proudPhDadvisor #IllinoisTechGrad

@janadoppa Thank you, Jana!

@bilgicm Congratulations, Mustafa!

Happy.

@Sriraam_UTD Thank you, Sriraam!

@bilgicm Well deserved

@elenadata Thank you, Elena! 🙏

@bilgicm Congratulations, Mustafa! 🎉

@berty38 Or, investigating the effect of their tool’s impact on the society, where some of the findings are negative.

Is it a conflict of interest if your work is being reviewed by someone who works for a company that sells a product you're improving on? This must happen all the time now.

Update your review explaining your reasoning after the discussion period, especially if you do not change your rating, or if you lower your rating. We owe that to the authors. Thank you.

PS: If you are a reviewer reading my rant, please consider updating your ratings in the light of the new information (author response and reviews).

Now the discussion period is underway for #AAAI2022, I'd like to share my thoughts as an Area Chair who is overseeing reviews and discussions for a few dozen papers.

We are currently studying whether a) providing the user with further transparency into what they are shown and why and b) giving the users knobs to adjust these systems would alleviate/lessen the filter bubble problem or exacerbate it. (The end of the thread).

Clearly, recommender systems are contributing to and exacerbating filter bubbles through different mechanisms (language, popularity, etc.) and they are affecting different political typologies differently (pushing them to extremes and homogenizing the lean of the news they see).

This news is based on our recent WWW'21 paper (cs.iit.edu/~ml/pdfs/liu-w…). It was a very interesting project with some expected and a few unexpected results. (Thread).