New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Implement a SQL, Qdrant SQL? #4026
Comments
Hey @raulcarlomagno, thanks for bringing that up! I recognize the problem of JSON filters being complicated to write sometimes, especially with explicit pydantic types in python. I, however, don't think sql-like language is a right answer to this problem, for the following reasons: Original SQL was designed for a very different purposes
Instead of doing this, I would rather:
Please feel free to share any additional arguments why would you think the sql is a better alternative |
thank you for answering @generall if i have time i will try to make an sql parser for the python client, but my initial thought was about having this parser on server side, to send sql queries through http or grpc |
I think it could be very useful to have a QSQL, or being optimist it could become an standard for vector databases, VSQL?
Instead of dealing with JSON in qdrant http api or all the methods in Python API for example. Queries can be written easily in the same way for scroll api and search api, in plain SQL. It can be useful for the qdrant dashboard or even for production queries.
The generated abstract syntax tree (AST), can match all qdrant features for searching and scrolling.
field score is a virtual calculated field that is returned when using LIKE for vectors
in the case of having named vectors, it would be like:
limit, offset and filtering payload
all payload and filtering
complex filtering
scroll api (no vector similarity search). Filtering fulltext, values count and range, plus sorting
geo bounding box filtering
The text was updated successfully, but these errors were encountered: