Learning to Rank for multi index

Versions (relevant - OpenSearch/Dashboard/Server OS/Browser):

Describe the issue:

Can we define a ltr model that would work for multi index. Every index would have a different structure, different attributes. Can a single model work for multi index?


Relevant Logs or Screenshots:

Thanks for posting a question! The model is pretty tied to the fields in the index. But, making it easier if there is overlap in field names could be something to try. Can you say more about the challenge you have? Also, feel free to open an issue here: Sign in to GitHub · GitHub

Don’t worry about filling in all the information in the proposal. It’s a guideline, not a requirement.

@markcohen Thanks for the reply. Well, different customer will have different index and every index will have different attributes and structure. I was thinking if we have a separate model for every index, it would be expensive, rather if one single model can work for multiple indices.

I think this is an interesting idea. Would you have a mapping between customer/user concepts and the internal model somehow? Can you lay out an example or two that might illustrate how you’re thinking about this? I do think if you do have these ideas, then creating a proposal in Github would be the best way forward for visiblity.

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.