I want to understand difference between
Creating simple vector search by importing model (may be from hugging face)
versus
Using ELSER
Any guidelines or recommendations on choosing approach? What are the use cases for each approaches? What are limitations for each approaches?
Hello @9d224d4833094bd482cf,
ELSER is an out-of-domain model trained by Elastic that enables you to perform semantic search (search based on the intent and contextual meaning of a search query).
It expands the searched passages into collections of terms, and these expanded terms are weighted as some of them are more significant than others (learned associations capturing relevance).
It's a more understandable search experience compared to dense vectors.
This blog post talks about learned sparse models and if you want more details and examples, you can check:
Note: ELSER is recommended for English language documents and queries.
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.