04/01: Ana-Andreea Stoica

[ehuppert]

Comment below with questions or thoughts about the reading for this week's workshop.

Please make your comments by Wednesday 11:59 PM, and upvote at least five of your peers' comments on Thursday prior to the workshop. You need to use 'thumbs-up' for your reactions to count towards 'top comments,' but you can use other emojis on top of the thumbs up.

[weijiexu-charlie]

Thanks for your presentation. I was wondering what are the potential applications of this research project?

[97seshu]

Thank you for your presentation. I have a similar question as my other classmates: is the glass ceiling effect common in most social media platforms, or the effect is only detected because of the exclusive recommendation algorithm of Instagram?

[yierrr]

Hi Dr. Stoica, thanks so much for sharing your research! I’m also interested in how the study may be applied to social media. Thanks!

[afchao]

Thank you for presenting for our group! Regarding the work described in your 2020 paper, I wonder the extent to which different social networks may exhibit different properties that may interfere with your theoretical conclusions. For example, I imagine (perhaps naively) that different demographic groups would exhibit different internal network structures. If this was the case, I'd be curious whether testing this model on different minority groups within the DBLP network would affect the results at all. Similarly, I imagine that the structure of these networks vary from platform to platform e.g. Facebook networks may not resemble TikTok networks or something to that effect. How general can we interpret your diversity seeding findings to be?

[NikkiTing]

Thank you for sharing your work! I would like to ask if your findings apply to a mix of sensitive attributes. For instance, does diversity seeding that accounts for both race and gender (e.g., women of color) reduce inequality?

[Panyw97]

Thanks for sharing! As you mentioned that the recommendation algorithm furtherly promoted the spread of inequality in social network, is there any solutions towards this issue based on the lack of incentives for developers (who mainly aim at realizing efficiency and profits)?

[wanxii]

Could you share some experiences about how lifetime questions could motivate theoretical studies and how to select data resources to help testify and validate theories? Many thanks!

[Yutong0828]

Hi Dr. Stoica, I really think that your work can have significance meaning for improving equality in social networks! My question is about the motivation behind companies or users to adopt more diverse recommendation system rather than the system more guided by homophily. According to my knowledge, people would prefer recommendation system that could meet their needs, which in this case, may be people who share similar interests or have other similarities with them. So, if people would naturally prefer similar people, how would they accept more diverse recommendation system? In my view, the algorithm could either recommend underrepresented users who shares similar interests with someone, or recommend users who are quite different from someone in an independent section, as some people would have special interest in learning about others who have a very different background.

[k-partha]

Thank you for your presentation! This is an interesting area of research in algorithmic biases. My question relates more broadly to how we can address inequities. Are there ways in which incentives can be built into networks that drive persons to choose socially equitable outcomes, without corrections to preexisting algorithms?

[timqzhang]

Thank you for your presentation ! I wonder how you raise the efficiency of model estimation, as the network analysis always requires huge computing power.

[xxicheng]

Thanks for sharing your work with us.

[FrederickZhengHe]

Thank you for this wonderful presentation! I am wondering if the replication data and code are available for further studying.

[bjcliang-uchi]

Very interesting research! I am curious about whether there are substantive differences in types of opinion leaders. Is it possible that say, those with the highest degree centrality (popular within a community) tend to be men and those with the highest betweenness centrality (bridging communities) tend to be women. If this is true, then how should we conclude about the bias?

[minminfly68]

Thanks for the presentation. Could you kindly elaborate more its applications in the social media perspective? Thanks!

[fyzh-git]

Thank you Dr. Stoica for presenting this highly relevant topic with rigorous mathematical construction. The extended form of the basic model to reconcile the seeming discrepancy between the outcomes of mathematical prediction and the empirical results is heuristic. The seed setting process, which proved to be possible to promote diversity without sacrificing efficiency under certain conditions, is very similar to the ideas behind the First and the Second Welfare Theorems in Economics. So, this opportunity to hear about these ideas from a non-econ perspective is a great experience to me. Thank you again for bringing this topic from the computational view.

[chuqingzhao]

Thank you Dr.Stoica for sharing your work. I have a similar question with my peer: could you please elaborate more on how the network model can be applied to real-world settings? Thanks

[hihowme]

Thanks a lot for your representation! How do you see your research to apply into the real-world settings? Thanks a lot.

[YijingZhang-98]

Thanks for your sharing. Your work is so insightful! I used to think. the recommendation could actually increase the economic efficiency since it would lower the searching cost of suppliers and consumers. You provide a new perspective. Hope to learn more in your lecture!

[siruizhou]

Thank you for your sharing. I am interested in learning more about the mathematical derivation related to this inequality issue.

[Yiqing-Zh]

Thank you for the presentation! I am wondering how you define efficiency. Is it defined from the perspective of society or other entities?

[ziwnchen]

Thanks a lot for the presentation! The topic of algorithmic bias and possible solutions is an important topic today. Could you share some of your thoughts about the possibility that stops companies from adopting such solutions? For example, how does diversity promotion related to platform profit?

[goldengua]

Thanks for your presentation! I was wondering what is your future direction?

[Yaweili19]

Thank you for sharing your work! Looking forward to your presentations tomorrow.

[Raychanan]

Hi Stojka, I don't have a specific question about your two papers. But they do remind me of the disappointing vaccine distribution algorithm developed by Stanford. Really glad your work is contributing to solving this kind of problem!

[XinSu6]

Thank you so much for your work! I am wondering why did you choose dynamic modeling?

[bazirou]

Thank you for sharing your work! Looking forward to your presentations tomorrow.

[bakerwho]

Thank you for presenting your work, Dr. Stoica. I'm interested in hearing more about the setup of your experiments, like many others! I'd also love to hear your recommendations for government or corporate policy to mitigate some of the issues you've discussed.

[wu-yt]

Thanks for sharing! This research is very interesting. What do you think the role of big tech firms is in creating this type of inequality? Because the recommendation system is basically created by them to keep our attention.

[chentian418]

Thanks for sharing your work, Dr Stoica! I was inspired by your notion of "Algorithmic Glass Ceiling", and would you mind elaborate a bit on how you integrate the model of algorithmic growth under a recommendation algorithm with the empirical finding and data? Thanks!

[zixu12]

Thanks for sharing your work! I think many students have raised similar questions that I am interested in. I am glad that your research seems to help solve some inequality issues.

[hesongrun]

Thank you for sharing your work! Looking forward to your presentations tomorrow.

[anqi-hu]

Thank you for sharing your work with us. I'd like to hear more about the application of your framework to other areas.

[YuxinNg]

Thanks for sharing your work. My question is that why you think a dynamic model is better for this topic than a a static model.

[qishenfu1]

Thank you for sharing! I am also very interested in this topic. I am curious about that will recommendation algorithms further aggravate the problem of sexism and racial discrimination?

[ginxzheng]

I'm interested in the explanation of the polarization created by algorithms. Thanks so much for your work!

[ttsujikawa]

Hi Dr. Stoica, Thank you for your wonderful work. The relationship between diversity and efficiency is pretty interesting. I have a question about the diversity behind the recommendation system. Since the recommendation is meant to be guided by the preferences of those who have a similar behavioral tendency, then why would you seek the diversity on the recommendation? Thank you!

[SoyBison]

Hello, welcome to our workshop! I'm excited to watch your talk asynchronously even though it's admittedly far outside my field. I'm curious about how different kinds of algorithms yield different results in your data.