Page MenuHomePhabricator

Preprocess unmatched pages
Closed, InvalidPublic5 Estimated Story Points

Description

Context

Structured Data is expecting to provide the Image Suggestion API POC to bot writers. Because the POC provides random results but does not guarantee pages for a requested offset, the API provides fewer results, sometimes none. One way we could address this is pre-processing the unillustrated pages with no ImageMatchingAlgo matches, query against MediaSearch and determine which pages have neither IMA nor MS results.

Acceptance Criteria
  • As a bot writer, when I request n number of pages from image suggestion API for a given wiki, I expect the response to have n number of pages**
    • When querying the production dataset, I expect to see a boolean value whether an unillustrated page has an image suggestion from MediaSearch so that the ImgSugAPI can disregard those results

**This is relative to when preprocessing occurs. MediaSearch might update its index after unmatched pages are processed