Within Azure Cognitive Search scoring profiles, you can use boosting functions. Here is one I'm working with:
{
"type": "magnitude",
"fieldName": "TasksCompleted",
"boost": .75,
"interpolation": "linear",
"magnitude": {
"boostingRangeStart": 1,
"boostingRangeEnd": 30,
"constantBoostBeyondRange": true
}
I chose .75 for boost because I'm trying to have the boost be pretty small compared to some other boosts that are present. However, I noticed that the boost seemed to be working backwards, where 1 was the most boosted document and 30 was the least. This is opposite of how the other magnitude boosts were working.
Are scoring profile function boosts with a boost value below 1.0
reversed for some reason? Or do they somehow reduce the total boost for matching documents (even though my function aggregation is set to SUM
)? I can't find any documentation around using numbers below 1.
As per Adding scoring profiles to a search index this is documented for functions freshness
property, but in general, using a boosting factor of less than 1
will reverse the boosting.