Hello, i have successfully created my schema and started to index some documents, so far so good. However i would like to sort my search results based on weights, so i came across query by weights. Say that i have those two fields: title, city. I want to give a greater weight to the tile than city so that i give priority to the titles. I am using query_by_weights=2,1. I would expect that results with the keyword in title would come first but then i saw that results are sorted by default with text_match. What is the purpose of the weights if they dont override sorting? How can i achieve what i want?

👋 can you post an example query, an example record that you were expecting first, and the record that actually shows up?

sure i have the following structure: Documents: {title: 'Eat Croissant in Nice', food: ['Socca', 'Ratatouille'} {title: 'Stroll through Montmartre's cobblestone streets', food: ['Baguette', Croissant] query_by=title,food query_by_weights=2,1 the second one comes first when i search for 'Croissant'

Can you try setting:

prioritize_exact_match: false

The exact match with the word

croissant

on the

food

field is giving it a higher text match score. Weight is used to influence the ranking when match score is the same.

is there a way to sort the results based on the weight of the fields?

There is also the

text_match_type

parameter which can help with the behavior you want. Here's what the docs say:

In a multi-field matching context, this parameter determines how the representative text match score of a record is calculated. 

Possible values: `max_score` (default) or `max_weight`. 

In the default `max_score` mode, the best text match score across all fields are used as the representative score of this record. Field weights are used as tie breakers when 2 records share the same text match score.

In the `max_weight` mode, the text match score of the highest weighted field is used as the representative text relevancy score of the record.

nice let me check!

It's max_weight

you are right! thank you so much, worked as well