- From: Charlie Harrison via GitHub <sysbot+gh@w3.org>
- Date: Mon, 09 May 2022 16:52:30 +0000
- To: public-patcg@w3.org
If we are allowed to set `group_id` after the fact (i.e. at query time) it resolves my concern, i.e. that it should be possible to make complicated queries from arbitrary data at query time. This is one of the downsides of the Attribution Reporting API in that you need to decide at report-creation time what the query needs to be, and the architecture of IPA seems well suited to solve that problem. I will need to think a bit (and understand more about IPA) if there's any benefit to multiple `group_ids` per record in a single query. I do believe there will be a large benefit to batching queries, so if that can be supported in an efficient way it might not be necessary. For example, it would be great to have an answer for how we would support the [Criteo experiment](https://medium.com/criteo-engineering/results-from-the-criteo-adkdd-2021-challenge-50abc9fa3a6) in IPA (even theoretically). In that setting, we have each attributed event contributing separately to 171 buckets. -- GitHub Notification of comment by csharrison Please view or discuss this issue at https://github.com/patcg/private-measurement/issues/9#issuecomment-1121342044 using your GitHub account -- Sent via github-notify-ml as configured in https://github.com/w3c/github-notify-ml-config
Received on Monday, 9 May 2022 16:52:32 UTC