ajayaks opened a new issue, #64178:
URL: https://github.com/apache/doris/issues/64178
Hi,
I was trying to run the hybrid search via below route
- Create a view with structured filter
- Then text search + semantic search
- RRF
I m using 4.1.0 version
High level:-
The query works correctly when selecting columns from the View directly.
However, when score() is added to the query, Doris returns the ranking score
but does not return values for the projected columns such as hotel_id and
hotel_name.
Details:-
//View
CREATE or REPLACE VIEW hotel_candidates AS
SELECT
hotel_id,
city,check_in_window,hotel_name,nightly_price,hotel_description,hotel_tags,
description_embedding
FROM hotels
WHERE
city = 'Seattle'
AND check_in_window = 'Next weekend'
AND nightly_price BETWEEN 150 AND 500;
select hotel_id,
hotel_name, city, hotel_description from hotel_candidates; [Working
fine]
``<html xmlns:o="urn:schemas-microsoft-com:office:office"
xmlns:dt="uuid:C2F41010-65B3-11d1-A29F-00AA00C14882"
xmlns="http://www.w3.org/TR/REC-html40";>
<head>
<meta name=ProgId content=OneNote.File>
<meta name=Generator content="Microsoft OneNote 15">
</head>
<body lang=en-US style='font-family:Calibri;font-size:11.0pt'>
<!--StartFragment-->
<div style='direction:ltr'>
hotel_id | hotel_name | city | hotel_description
-- | -- | -- | --
H003 | Pike Place Palace | Seattle | Historic luxury hotel overlooking
Pike Place Market with premium spa services
H008 | Azure Heights | Seattle | Premium luxury hotel with rooftop
amenities and personalized concierge services
H001 | The Grand Seattle | Seattle | A luxurious boutique hotel with
stunning ocean views and world-class amenities
H012 | Capitol Hill Quarters | Seattle | Hip and cozy hotel in the heart
of Capitol Hill with local coffee culture vibes
</div>
<!--EndFragment-->
</body>
</html>
// Below is the ISSUE as no hotel_id and hotel_name in output with score()
function.
SELECT
hotel_id, hotel_name,score() AS text_score
FROM hotel_candidates
WHERE hotel_description MATCH_ANY 'luxury boutique'
ORDER BY text_score DESC
LIMIT 1<img width="1123" height="240" alt="Image"
src="https://github.com/user-attachments/assets/8341672a-9d3e-46f8-a2d9-e1f2f057a661";
/>->
Its only showing the text_score not the other columns data while using the
score().
hotel_candidates view to represent the structured filtering stage of Hybrid
Search. However, Doris appears to have a limitation (or optimizer issue) when
using the full-text ranking function score() on top of a view. While MATCH_ANY
works correctly through the view, score() does not properly return the
associated row columns (hotel_id, hotel_name), so for the text-search stage we
queried the base table directly while keeping the same structured filters.
Note:- The same query works fine when run directly on the table instead of
the view.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
