I am requesting an improvement to Kagi's ranking algorithm to better handle mixed-quality UGC (User Generated Content) platforms, specifically "note.com" in Japan (similar to Medium).
Currently, while Kagi excels at blocking pure spam domains, it struggles with individual low-quality AI articles posted on high-authority domains like Note. These AI-generated posts often rank high in search results due to the domain's reputation. Since Note also hosts many valuable human-written articles, I cannot simply "Block" or "Lower" the entire domain without losing useful information.
I envision this as a backend improvement to the "quality filtering" mechanisms.
When I search for topics (often technical or hobby-related), I want Kagi to scan the content of pages even from trusted domains like note.com. If a specific page exhibits strong signs of being low-quality AI generation (SEO spam), it should be downranked or filtered out, while leaving the genuine human-written articles from the same domain visible.
This would allow users to keep trusted UGC domains in their results without being flooded by the increasing amount of AI content hosted on them.
