I have a MongoDB Atlas Search index with the autocomplete type on a field name and the following attributes :
maxGrams : 15
minGrams : 2
tokenization: edgeGram
fold diacritics: true
The collection contains the following names :
The American
The American Equity Underwriters
The American Prairie Foundation
The American Conservatory Theater
The American Club
With this autocomplete query :
{
autocomplete: {
query: 'The American',
path: 'name'
}
}
I'm able to retrieve all the names above with the following searchScore :
The American Conservatory Theater -> 15.474836349487305
The American Equity Underwriters -> 15.379003524780273
The American Prairie Foundation -> 15.379003524780273
The American Club -> 15.271049499511719
The American -> 13.68109130859375
Even if The America is the exact match it gets a lower score.
Why does the exact match searchScore is below the others ?
As a workaround I found something like:
{
"compound": {
"should": [
{
"phrase": {
"query": "The American",
"path": "name",
"score": { "boost": {"value": 5}}
}},
{
"autocomplete": {
"query": "The American",
"path": "name"}
}
]
}
}
The score boost on the phrase match will handle exact match and make the exact match score higher than others.