Skip to content

Review 'default' retrieval algorithm #161

Open
@renekrie

Description

Chorus let's the user choose from a list of pre-configured search algorithms that each yield different result sets and rankings.

The default algorithm is a pretty bad one that we would always de-recommend in e-commerce search, eg. using mm<100%. This is problematic from a number of perspectives:

  • As Chorus is meant to be an example application for a good e-commerce search, the default algorithm should be a reasonably good one
  • As we are adding vector search, we default increasingly serves as a reference for comparison, which would not make sense, if it is so bad that nobody would use it in practice. More concretely, we have a search algorithm that uses vectors for boosting but keywords for retrieval. Which algorithm would we compare it to? Should we turn on mm=100% for hybrid search? - In the context of telling the story - default vs. hybrid search - we should keep it off. But in a more general e-commerce search context, it doesn't make sense.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions