Developer forum

When a field is analyzed boosting does not work as we want for some fields - i.e. product number.

For a number like ABC123 it is not an issue because the analyzed and non-analyzed version would give the same 'term'. But if the number is ABC-123 an analyzed would give 2 terms and a non-analyzed would give one term. And then boosting would be different - i.e. if searching for 123 would boost high in the analyzed but find nothing in the non-analyzed version.

So we use the nonanalyzed version of the field to ensure high boosting when you enter a precise SKU to ensure that product is on top - i.e. in express buy. But if you search for random things and press enter, we might want to fin ABC123 when searching for AB - and still if you have a product with SKU AB put that up higher.

By default in Swift product number is part of free text summary field that is analyzed and used on a contains expression (which will not add scoring) and then also on the product numnber search (non-analyzed) using Equals which will score the document... And in this way both scoring for precise searches and partial searches for SKU would come out fine.

I see.

>>and then also on the product numnber search (non-analyzed) using Equals which will score the document

But that requires you to force to lower case on the search term or the non-analyzed product number wouldn't match and the boost would still not work, no?

Imar

Or just add an equals on the analyzed product number would do the same in 99% of cases. Most customers have sane product numbers.

Then crazy customer have SKUs like '"!#¤ % & / 213 ___ØÆÅÅLK ```!RRS(' and they will run into issues no matter what.... :-)

>>Or just add an equals on the analyzed product number would do the same in 99% of cases. 

Yeah, that's what I've been doing so far. Think I'll stick with that.

Thanks!

Imar