Deprecate the LegacyBM25Similarity class and default to BM25Similarity #17315
Description
Is your feature request related to a problem? Please describe
Coming from LegacyBM25Similarity and from @msfroh comment here #17241 (comment), the class already includes a note advising users to use Lucene's BM25Similarity. Since the work for the 3.0.0 release has begun, this is the perfect opportunity to default the scoring to BM25Similarity.
The change is focused on moving from a legacy implementation to the current Lucene BM25Similarity, which provides a cleaner and more standardized way scoring. The switch from LegacyBM25Similarity to BM25Similarity doesn't change the core ranking/scoring behavior in a way that would significantly impact search results.
More details on apache/lucene#9609.
Describe the solution you'd like
Default to Lucene's BM25Similarity, while allowing users to choose LegacyBM25Similarity if required.
Example
curl -X PUT "http://localhost:9200/test-index" -H 'Content-Type: application/json' -d'
{
"settings": {
"index": {
"similarity": {
"default": {
"type": "LegacyBM25",
"k1": 1.2,
"b": 0.75
}
}
}
},
"mappings": {
"properties": {
"content": {
"type": "text"
}
}
}
}
'
Related component
Search:Performance
Describe alternatives you've considered
No response
Additional context
No response
Metadata
Metadata
Assignees
Labels
Type
Projects
Status