- From: Martin Thomson via GitHub <sysbot+gh@w3.org>
- Date: Wed, 22 Jun 2022 04:53:31 +0000
- To: public-patcg@w3.org
When you define "breakdown keys" you don't mention the site on which source event occurred. Did you mean to make that one of the factors that might be used to determine a breakdown key? I've a few quibbles with your Mathy bit, but mostly presentation-wise. Let's concentrate on what I think your intent is. In my slides for tomorrow, I had something like: 1. For each conversion on site X, find all ad events that: * are from the same ~~browser~~ person, * precede the conversion, and * meet some additional conditions 2. For those events, calculate a function over associated values 3. Aggregate the results of each calculation against the site in the ad event using a second function You have described concrete versions of both the per-conversion and aggregation functions that I think could (with appropriate values for $\varepsilon$) meet our goals. I hope that we will be in a position to swap those functions out as we learn more, but as a first attempt, these seem plausible to me. Though I have to point out that my interests are in making this work while maintaining privacy. I don't have to live with the utility this produces, so my opinion here means very little. -- GitHub Notification of comment by martinthomson Please view or discuss this issue at https://github.com/patcg/private-measurement/issues/16#issuecomment-1162638446 using your GitHub account -- Sent via github-notify-ml as configured in https://github.com/w3c/github-notify-ml-config
Received on Wednesday, 22 June 2022 04:53:32 UTC