Skip to content

kwx 0.0.2.2
Compare
Choose a tag to compare
@andrewtavis andrewtavis released this 31 Jan 13:25
· 169 commits to main since this release
v0.0.2.2
d763e82

The minimum viable product of kwx:

  • Users are able to extract keywords using the following methods

    • Most frequent words
    • TFIDF words unique to one corpus when compared to others
    • Latent Dirichlet Allocation
    • Bidirectional Encoder Representations from Transformers
    • An autoencoder application of LDA and BERT combined

  • Users are able to tell the model to remove certain words to fine tune results

  • Support is offered for a universal cleaning process in all major languages

  • Visualization techniques to display keywords and topics are included

  • Outputs can be cleanly organized in a directory or zip file

  • Runtimes for topic number comparisons are estimated using tqdm

Assets 2
Loading