Re: [web-adv] PELICAN: Private Learning and Inference for Causal Attribution

Hi All,

At the tail end of today's call, there was mention of sending out documentation prior to future presentations so just wanted to send this note out once more as it relates to next week's Pelican attribution presentation. 

Please note that you'll need to copy and paste the github link in your browser since I purposely sent this message in plain text. Reason being is that I don't think my original note made it out to the group last week (possibly due to a url defense service used on my company's end).

On a side note, we will have a few slides to present during next week's call (12/1) in support of the more detailed proposal directly on github.

https://urldefense.com/v3/__https://github.com/neustar/pelican__;!!N14HnBHF!oAyMGaTZW9e6upiH3AsloM36luwHwtWgqU4eTqMM0UtlV70-NB2t7NtM90Jk027QVSTX$ 

Thanks,
Ryan

From: "Avecilla, Ryan" <Ryan.Avecilla@team.neustar>
Date: Wednesday, November 18, 2020 at 12:05 PM
To: "public-web-adv@w3.org" <public-web-adv@w3.org>, Wendy Seltzer <wseltzer@w3.org>
Subject: [web-adv] PELICAN: Private Learning and Inference for Causal Attribution

Hi All,
                                                     
Please see below our proposal for private learning in the Chrome ecosystem: Pelican (PrivatE Learning and Inference for Causal AttributioN).
 
https://urldefense.com/v3/__https://github.com/neustar/pelican__;!!N14HnBHF!oAyMGaTZW9e6upiH3AsloM36luwHwtWgqU4eTqMM0UtlV70-NB2t7NtM90Jk027QVSTX$ 
 
The first part of the proposal lays out some industry terms and methods, and the second part outlines the importance of accurate measurement to different parties. The third section sets out a high-level framework and some core requirements for an accurate learning-based approach to measurement.
 
Based on conversations with the Chrome team earlier in the year, we’re starting off with high level requirements rather than suggesting specific solutions at this time.  We have more detailed proposals that we can add to or revise as appropriate once there’s been an opportunity for initial feedback.
 
As a FYI, I asked Wendy to add this topic to the agenda for the December 1st call in the event that some group members were taking extended time off during the US Thanksgiving holiday. In the meantime, feel free to provide feedback directly in the Pelican GitHub repo linked above.
 
Thanks,
Ryan