Could anyone advise regarding the proper settings of search hit ranking?
Based on my experience YaCy is mostly able to find the right pages when searching the right way but the rankig (order of hits) seems to be huge mistery to me.
It’s great that ranking could be adjusted under Ranking and Heursitcs but could anyone advise if there are any seetings that are proven to work well?
I assume the large commercial search portals add a little salt to search ranking by using ML or profiling users based on internet fingerprint besides deciding purely based on SEO data, but how does it work for YaCy? Is there any way to tweak ranking to make YaCy fit for daily use?
Let’s put it another way?
Is there anyone who ever adjusted the Ranking settings and achieved better search hits or anyone uses the default settings?
In all these years I adjusted it here and there. There is one important thing to know: we have different ranking strategies in YaCy for different use cases:
- p2p mode: very complex. Mixed from Solr ranking, RWI ranking and who-comes-first-gets-better-places
- portal/intranet mode: only solr ranking
The most important method is the solr ranking. It works using boosts on search fields. Fields are i.e. title, url, sections, general text and so on. And on every field there is a lucene ranking, which is a special version of TF*IDF.
All details would fill a many-hour lesson series (which I actually gave as assistant professor some years ago).