-
Notifications
You must be signed in to change notification settings - Fork 1.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Implement FeedFeedback API #3498
Conversation
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
react parts make sense to me, didn't take a super close look at the product logic.
…refreshes, and modify the api design a bit
Your Render PR Server URL is https://social-app-pr-3498.onrender.com. Follow its progress at https://dashboard.render.com/web/srv-cognenmd3nmc73f62650. |
this seems to be working, can we ship it behind a statsig gate for now? |
Actually, before shipping, we should make sure that when this api reports a post as seen, the user has definitely seen it. |
@whyrusleeping done. I now make sure the post is substantively visible for at least 3 seconds |
The Pull Request introduced fingerprint changes against the base commit: e264dfb Fingerprint diff[{"type":"file","filePath":"package.json","reasons":["expoConfigPlugins"],"hash":"972bc4be2c765e4da03b73ced5fb523694b13879"}] Generated by PR labeler 🤖 |
* Updated as PR on Plural is merged (bluesky-social#3882) * Feedback API (bluesky-social#3498) etc.
* Implement onViewableItemsChanged on List.web.tsx * Introduce onItemSeen to List API * Add FeedFeedback tracker * Add clickthrough interaction tracking * Add engagement interaction tracking * Reduce duplicate sends, introduce a flushAndReset to be triggered on refreshes, and modify the api design a bit * Wire up SDK types and feedContext * Avoid needless function allocations * Fix schema usage * Add show more / show less buttons * Fix minor rendering issue on mobile menu * Wire up sendInteractions() * Fix logic error * Fix: it's item not uri * Update 'seen' to mean 3 seconds on-screen with some significant portion visible * Fix non-reactive debounce * Move methods out * Use a WeakSet for deduping * Reset timeout * 3 -> 2 seconds * Oopsie * Throttle instead * Fix divider * Remove explicit flush calls * Rm unused --------- Co-authored-by: dan <dan.abramov@gmail.com>
* Update Japanese translations * Updated as PR on Plural is merged (#3882) * Feedback API (#3498) etc. * Improve localization marks (#3285) * Update Lightbox.tsx * Change strings for easier localization * Update DeleteAccount.tsx * Update LabelsOnMeDialog.tsx * Update FeedCard.tsx * Update index.tsx * Update LabelsOnMeDialog.tsx * Update index.tsx * Update FeedCard.tsx * Update SelfLabel.tsx * Update Hashtag.tsx * Update index.tsx * Update Hashtag.tsx * Update ChangeHandle.tsx * Update index.web.tsx * Update index.web.tsx * Update index.tsx * Remove unnecessary `<Trans>` tags * Update Drawer.tsx * Finnish translation update (#3755) * Update messages.po Translated new strings (GIF's etc) and cleaned up deprecated ones. * Update messages.po resolve conflict * Update messages.po removed double quotes. Thanks @lapanti * fix bad bool check in action (#3885) * fix bad bool check in action * add `fetch-depth` so we can get the commit hash * `.env` should be in `.easignore` 🙃 * Updated to follow #3285 etc. * The translation of 'place' and 'apply' of labels has been unified into '適用する'. * Fix translation of "repost" * Updated * Updated Japanese translation * Updated Japanese translation - #3816 * Fixed translations * Updated translation : #3995 * Updated Japanese translation * Updated Japanese translation (ref. #3962 #4028) * Updated Japanese translation * Updated Japanese translation * Updated Japanese translation --------- Co-authored-by: Minseo Lee <itoupluk427@gmail.com> Co-authored-by: Jan-Olof Eriksson <jan-olof.eriksson@iki.fi> Co-authored-by: Hailey <me@haileyok.com>
Hi! Sorry if it's rude to comment on an old PR, but wondering if there's somewhere I can follow along with the plans to generalize this to custom feeds. It seems necessary for any good algorithmic feeds. |
This PR introduces a system for feedback to make it's way back to feeds. This feedback lets the feed know which posts have been seen (so it can stop showing the same stuff twice), requests for more/less of a kind of post, and interaction signals to help train the feed's model.
The permissions model for this is going to be important. For now, we're hard-coding it to the Discover feed only. Once we introduce a permission framework for asking user permission, we'll be able to send this feedback to other feeds as well.
Most of this is invisible, except for the show more / less menu items:
Tasks