Can anyone shed any light on how the search page uses the relevance weightings to decide which assets have most relevance?
I set the defaults in the search manger so that the asset name had a weighting of 100, the short name a value of 50 and the contents 1.
I then re-indexed the relevant assets.
I'm not seeing the results I expected, though.
I was expecting that the search engine would award the value of the weighting each time it came across the term, such that if the term appeared once each in the asset name and asset short name, the asset would receive a value of 150 (100+50), making it easily more relevant than an asset that had a single occurence in the contents (earning it 1 point by my reckoning).
The search terms I'm using are longer than those that are ignored by the indexing routine, and they are not in the white noise list.
The search page is set to return results in reverse order based on relevance, and the logic to match all.
We do have caching switched on, but I'm assuming that searches are not covered by this?
What am I doing wrong?
Could anyone offer advice on this, please?
I would recommend contacting Squiz Support directly to arrange one of the Implementation or Development staff to take a look at your database. It should work exactly as you describe.
Okay - thanks, Avi.