== Background
We are [[ https://phabricator.wikimedia.org/T373039 | running a survey ]] against the output of three different recommendation engines:
- morelike
- vectorsearch
- listreader
Assuming that none of these recommendation sources are ideal, I'd like to run a follow-up survey with the "winner" of the first experiment and two other recommendation sources:
# The "See Also" links from the articles
# The [[ https://superlative-jalebi-8c2fbb.netlify.app/ | Most Linked API ]]
The goal is to see how the performance of a recommendation engine based on an article's position in the knowledge network and hand-curated recommendations from community contributors compares to the article content based APIs we tested in the first experiment.
== User story
As a member of the web team, I want to know if any other sources of recommendations outperform the 3 APIs in the non-UI experiment. will affect the end user
== Requirements
APIs to use:
- MoreLike
- Vector Search
- See Also
- @Jdlrobson's idea
We would like:
- 3 responses from each API (Minimum of 4 responses, maximum of 12)
- 6 questions? (randomize if we can?) (maybe 5?)
Criteria:
More wonders of the world?
=== BDD
- For QA engineer to fill out
=== Test Steps
- For QA engineer to fill out
== Design
n/a
== Acceptance criteria
- Add acceptance criteria
== Communication criteria - does this need an announcement or discussion?
- Add communication criteria
== Rollback plan
- What is the rollback plan in production for this task if something goes wrong?
//This task was created by Version 1.2.0 of the [[ https://mediawiki.org/w/index.php?title=Reading/Web/Request_process | Web team task template ]] using [[ https://phabulous.toolforge.org/ | phabulous ]] //