4 ms·
Explicit control can work for ranking but can’t scale to the retrieval step of search - you aren’t going to sift through thousands of thousands of pages of pote
by mattnewton 3y ago
Explicit control can work for ranking but can’t scale to the retrieval step of search - you aren’t going to sift through thousands of thousands of pages of potential results to give feedback on a meaningful portion of them, there must be some method of pulling pages to be ranked. This is where the gaming happens, with basically limitless spam domains that need to be filtered out. There just is no way Kagi is 100% relying on self reports for this.
- asdff 3y agoThe whole idea of the model is to fit it on some reference so you can project it onto new data you haven't classified yet. So your first few weeks, you might get some spam sure. But eventually if you are optimizing the system by e.g. only ranking technical blogs with no seo bullshit, then technical blogs with no seo bullshit is what would fit best with that model among the new data and what you see on your results. That can't be gamed without knowing the sort of model fit you are generating and this fit is defined by the user.
- bayindirh 3y agoThe ranking is only relevant to you. It's part of Kagi's lenses and search tailored for you, by you approach. I guess their "Universal Summarizer" is a part of their indexer, so they try to understand what you want and provide you more relevant results. They probably have other signals and systems to detect spam and other unwanted things.