Last week in Venice, I presented our work on Social Circle polls at the hashtag#StatPhys29 satellite event on Complex Networks.
Come near, pollsters of the world! We have a recipe for guessing the next election result.
What if GPT-powered bots sneaked into our online conversations?
Could we spot them, or would we need more powerful tools?
Check out our new preprint to find out how social networks may change with LLMs!
@Gius_C@Daniebilearxiv.org/abs/2506.21620
Just presented our work on food pairing & recommender systems at Sapienza's Physics Dept! 🍽️👨🔬
Blind taste tests showed: fewer shared flavor compounds = worse combos 😬
Flavor chemistry isn't in recipes yet—but it should be.
Preprint 👉 arxiv.org/abs/2406.15533
What if, instead of asking "Who will you vote for?", we asked "What % of your social circle will vote for ...?"
This question can improve the predictions—but polarization can distort results.
🚀 The good news? We found a fix! Check out our preprint:
arxiv.org/abs/2503.23879
And it's 2! Finally, I was able to present our latest work at @ConfCompSys 2024 about social circle polls.
Sometimes, wisdom of the crowds may be misleading... but we have the fix!
@SonyCSLRome@SonyCSLParis@cref_official
Next up, @Gius_C talking to us about crises. But since the financial crisis, it’s been time to understand how contagion propagates on economic networks. We’ll see how models can measure systemic risk.
Opinion swings talk is gone! Great opportunity to share this work with the scientific community at @ConfCompSys
Don't miss my next talk about new polling methods on Friday (11,30 - Room 11)
@Gius_C@ric_dicle@_AnnaMancini_@ADTony_
PAPER OUT!
Have you ever wondered why prosciutto and melon are good eaten together? The reason might be "food pairing". That's how we created a new recommender system for matching new pairs of food... and much more! Check out our preprint!
arxiv.org/abs/2406.15533