<Product Initiative>Image Suggestions Dataset
Closed, ResolvedPublic
Actions

Assigned To

Authored By

	• SWakiyama
	Jan 28 2022, 4:40 AM

Description

Request Status: APP Request
Request Type: project support request
Related OKRs: P-PPL KR1

Request Description: The algorithm to match images with unillustrated articles generates a dataset. This dataset needs to be stored with write access for the team. The algorithm will be run once on articles and then updated weekly. Many to many article image relationship. The model for the algorithm needs to be maintained and updated, based on user feedback data to improve the article-image match. The Image Suggestions dataset will require updates.
Indicate Priority Level: High - supporting Newcomer Pilot and SDAW
Main Requestors: Structured Data Team: Content Data Products
Ideal Delivery Date: End of Q3
Stakeholders: Structured Data, Growth, Inuka:Wikistories

Document Type	Required?	Document/Link
Related PHAB Tickets	Yes	https://phabricator.wikimedia.org/T299781
Product One Pager	Yes	https://docs.google.com/presentation/d/1W_6f5BN-l26wAO8LQsgbQjT6n5jETv7r3RGVoPV_8w0/edit#slide=id.g103d38f7ef2_0_0
Product Requirements Document (PRD)	Yes	https://docs.google.com/document/d/1mpxJAHiBrCnsA6D8alrRMthHo3z7aTeDSnUMuTBKCmI/edit?usp=sharing
Product Roadmap	No	<add link here>
Product Planning/Business Case	No	<add link here>
Architecture	Yes	https://miro.com/app/board/uXjVOUYPg6M=/
Data Requirements	Yes	https://docs.google.com/presentation/d/1W_6f5BN-l26wAO8LQsgbQjT6n5jETv7r3RGVoPV_8w0/edit#slide=id.g103d38f7ef2_0_68

		Status	Subtype	Assigned	Task
		Resolved		lbowmaker	T300327 <Product Initiative>Image Suggestions Dataset
		Resolved		lbowmaker	T292661 <Product Initiative> Image Suggestions API v1.0

• DAbad updated the task description. (Show Details)

January 28, 2022 - Product Steering Committee

Progress: significant progress on alignment of back-end design for data persistence and pipelines
Open Questions: working with ML/AI team to review how we can eventually support algorithm in Liftwing
Future Work: changes to Image Recommendations API, move off of MVP and deprecate WMCS version

Update as of 2022/02/14

the team has reviewed code and started work on any required refactoring to run via Airflow
The team continues to iterate on a data model for image suggestions that incorporates some of the new data fields as part of the MVP
We have met with Chris Albon's team to look at future support in LiftWing

Summary as of May 10th

SD in final stages of creating the Airflow job - data writes to Cassandra schema
On target to deliver data gateway service to production by end of May (~75% done)
Schema and stream for feedback is live - waiting on data being written

• DAbad changed the task status from Open to In Progress.Jun 8 2022, 3:24 PM

• DAbad triaged this task as High priority.

Summary as of June 8th

Growth has switched to use the new API, resolving ticket