Love or Money? (in collaborative filtering)

Gotta smile at his succinct summary: “Love is proxy for matrix vector” — Ashish Goel of Stripe, formerly Twitter at ICME

Ashish was wondering how to personalize content on Twitter, lacking much original content, intent or demographics. To parse the social graph for attention allocations, he considered several frameworks. “Love” is short hand for undiluted attention; like “loving all of your kids equally”, having more kids does not diminish the love for each. “Money” is like a finite resource to be allocated. When comparing the two, he found that “money” was 25x more useful than “love” as a framework for how to associate attention and association within the social graph.

From the AI in Fintech Forum, hosted by the Stanford Institute for Computational & Mathematical Engineering (“We do big math”).