Soham De

sohamde@uw.edu

I’m a PhD student at the University of Washington and a part of the Center for Informed Public, advised by Martin Saveski. My research interests are broadly within Computational Social Science, which includes the study of misinformation and social media. I’m specifically interested in algorithms that help combat social harms and bring people together - let’s chat if this sounds exciting!

Previously, I interned at Microsoft Research India under Anirban Sen and at the University of Michigan under Joyojeet Pal. I graduated magna cum laude from Ashoka University with a Gold Medal in Computer Science. Beyond academics, I’m a competent artist, photographer, archer and an incompetent pianist.

personal updates

Oct 2024	We were selected as 1 of the 3 winning interventions in the Prosocial Ranking Challenge. Experiment is currently underway - stay tuned for an interesting paper!
Jul 2024	Heading to Philadephia to present our work at IC2S2 2024
Jul 2023	I’ve documented by experience applying to CS/IS PhD programs in this blog. Check it out and let me know if you find it helpful!
Jun 2023	I’m finally happy to share that I’ll be joining the PhD program at UW (iSchool) this Fall, and will be advised by `Martin Saveski`. Hence the color change!
Apr 2023	I visited Ann Arbor and Seattle in April, to attend PhD visit days and organise a conference. At some point, I will add these pictures in my gallery.

selected works

FeedSpan: A bridging-based prosocial feed reranking algorithm

Luke Thorburn, Soham De, Kayla Duskin, and 4 more authors

🏆 Winning Intervention (Top 3)

In The Prosocial Ranking Challenge 2024

Abstract Web

The algorithm detects civic content that doesn’t attract diverse engagement (not ‘bridging’) and replaces it with similar content that is expected to be more bridging, hopefully reducing divisiveness without decreasing civic content.
Enhancing Crowd-Sourced Fact-Checking Using LLM-Driven Consensus

Soham De, Michiel Bakker, and Martin Saveski

In The 10th International Conference on Computational Social Science 2024

Abstract PDF

One of the main limitations of Community Notes—Twitter’s crowd-sourced fact-checking system—is that it typically takes too long for raters to reach the consensus needed for a note to be attached to a misleading post. We propose speeding up the process by generating super-notes, LLM-generated fact-checking notes that (a) combine key information spread across several existing notes, (b) use language likely to drive consensus among diverse raters, and (c) follow first-order principles of effective fact-checking. Our contributions include a pipeline to generate such supernotes that involves a custom reward-model trained on historical Community Notes data that predicts how a user would rate any note with 70% accuracy. We also outline ongoing and future experiments using human-subjects to evaluate the effectiveness of such an intervention in the wild.