Skip to content

Better implementation of JsonVectorCollection than the "fake words" approach #1890

Open

Description

@lintool

To index JsonVectorCollection sparse vectors, we currently use the "fake words" trick - just duplicate the word X times, where X is the score. This might be a better solution: https://lucene.apache.org/core/8_11_0/analyzers-common/org/apache/lucene/analysis/miscellaneous/DelimitedTermFrequencyTokenFilter.html

Activity

  1. tteofili commented on May 30, 2022

    @tteofili
    Collaborator

    I did an experiment for this a while ago, the behavior for some reason is not exactly the same (expected) as the fw one. I can dig into this.

  2. JMMackenzie commented on Jun 1, 2022

    @JMMackenzie
    Contributor

    Possibly different due to #1843?

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

Metadata

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