Why use BM25 to prerank? The Elastic Search default scoring method is BM25.

koursaros-ai / nboost

NBoost is a scalable, search-api-boosting platform for deploying transformer models to improve the relevance of search results on different platforms (i.e. Elasticsearch)

Apache License 2.0

674 stars 69 forks source link

Why use BM25 to prerank? The Elastic Search default scoring method is BM25. #48

Closed caipengbo closed 3 years ago

caipengbo commented 4 years ago

I do not know why implement a PrerankPlugin use BM25 algorithm, the ES use BM25 algorithm to scoring the results.

pertschuk commented 4 years ago

NBoost works with any JSON based system, not just Elasticsearch.

Also it works directly within python: e.g.

from nboost.plugins.models import resolve_model

reranker = resolve_model(data_dir, model_dir, model_cls)

The preranker is for cases such as these, and is turned off by default.