Add query preprocessing before hybrid retrieval - #175
Conversation
Thanks for creating a PR for your Issue!
|
|
@Shalini828 |
|
I've updated the PR according to the latest architecture and resolved the CI issues. All checks are now passing. Please let me know if anything else needs improvement. |
|
Hi @Shalini828 Shalini, thanks for the update and for fixing the CI issues! Just a heads-up: I merged another PR recently which caused some new conflicts in this branch. Since today is the final day of the program, I'm trying to wrap everything up quickly. If you're able to resolve these new conflicts right away, please go ahead and I’ll check it. But if it’s taking too much time, no worries at all—you don't have to stress about it since we're on a tight deadline. Let me know what you think! |
|
@Shalini828 Now the conflicts are resolved great job. But some checks failed please check that if you are free. |
|
I've checked and the failing tests are resolved now. All checks are passing. |
Great job Shalini |
|
By the way @Shalini828 can you please link the issue number too. |
Summary
This PR introduces a query preprocessing step before hybrid retrieval.
Changes
engine/preprocessing.pypreprocess_query()to normalize user inputsearch_pdfs()before retrieval logicWhy
This ensures consistent query formatting and improves retrieval quality by:
Impact
Improves robustness of hybrid search pipeline without affecting existing logic.