brave-browser
brave-browser copied to clipboard
Implement text embedding processing for ad matching MVP
Removing from the 1.43.x
milestone as https://github.com/brave/brave-core/pull/14393 is still opened and hasn't been merged into master
yet. We're releasing 1.43.x
this week so there's no way this is making it. We only move issues into milestones once they've actually landed in the target milestone and https://github.com/brave/brave-core/pull/14393 is still opened.
@ptjames Adding QA/Blocked
label as test plan mentions building locally which QA does not do, please provide additional test plan, can work with @btlechowski on this as needed.
Verification passed on
Brave | 1.45.75 Chromium: 106.0.5249.65 (Official Build) beta (64-bit) |
---|---|
Revision | 3269dc3633cdd2ab94546fdbe54962e45b17a6e0-refs/branch-heads/5249@{#580} |
OS | Ubuntu 18.04 LTS |
Verified test plan from https://github.com/brave/brave-core/pull/14393
Verified text embedding is working:
[5051:5051:1009/140553.354039:VERBOSE3:text_embedding_processor.cc(72)] Successfully logged text embedding HTML event
[5051:5051:1009/140553.373045:VERBOSE3:text_embedding_processor.cc(80)] Successfully purged stale text embedding HTML events
Verified embeddings being stored in table text_embedding_html_event
Verified embeddings are not duplicated for the same id:
Before page reload:
After page reload:
Verified when embedding is not possible, proper message is shown:
[6876:6876:1009/144455.972530:VERBOSE1:text_embedding_processor.cc(53)] No text available for embedding
Logged https://github.com/brave/brave-browser/issues/25878 for failed embedding on wikipedia pages