My problem is that the search is a small addition to my Application, and I really do not want to spend so much time digging the whole idea. Looking at my search results - this is a very common picture in which I get very good matches (7+) and very very bad matches, witch rating - 0.10. If I wanted to sort the results using any criteria other than evaluation, this would make very little sense, since 0.10 has almost nothing to do with the query and may appear first in the list.
Seriously, it seems that cutting everything below a score of about 3 will make my results more consistent, and sorting will make more sense.
Now, after some basic research, it seems like many people think that filtering Solr results by result is a really bad idea. There are some hits on how to do this, but so far I have not found a working solution.
The proposed ideas using frange (both for the correct q-query and qf) do not really work. Throwing results with a low result in the application itself seems rather boring, as it will break the pagination, slow down the work and generally give a lot of unnecessary work.
After about an hour at Google, I found out that many people really want this solution, although I could not find anything that works for me.
So, is there any way to reduce the results of poor results on the solr side? Are there any custom filters for this?
Edit:
Due to a lack of results, for some reason there is a significant difference in level. For example, the last corresponding result will be, say, 4.5 points, and there are several more results with the next highest at 0.12 ... Maybe I'm doing something wrong at the index level? Is there an easy way to push these non-local results to a hash of the result? After some more research would look like this, I would become less confident after 1 grade ...