Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Printed Work Classification - Automation #107

Open
emilyyan04 opened this issue Nov 3, 2023 · 3 comments
Open

Printed Work Classification - Automation #107

emilyyan04 opened this issue Nov 3, 2023 · 3 comments

Comments

@emilyyan04
Copy link
Collaborator

For 11/9:

  • Make best guess at threshold (+ reasoning)
  • Use pillow library to manually label correct category
  • After manual labels assigned, loop through dataset; adjust thresholds by increments, calculate error rate for each iteration, determine actual best cutoff
  • Evaluate highest-performing cutoff to determine viability of this process
  • Optional: relate general trends in scores to characteristics of posters vs prints
@emilyyan04
Copy link
Collaborator Author

Application of #108

@emilyyan04 emilyyan04 changed the title Printed Work Classification Printed Work Classification - Automation Nov 11, 2023
@emilyyan04
Copy link
Collaborator Author

emilyyan04 commented Nov 19, 2023

Dataframe schema:

  • accession_number: accession number
  • qid: QID
  • locale: language (from text detection)
  • description: text output (from text detection)
  • label_en: title of work (manipulated for string matching)
  • partial_ratio, ratio: scores from string matching
  • prop: ratio/partial_ratio
  • category: classification of work based on prop
  • title: title of work (official)
  • type: actual classification of work (print, poster)

@emilyyan04
Copy link
Collaborator Author

  • Data: gallery_buchanan/text_works/ratio_scoring.csv
  • Dataset: 619 English works (206 prints, 413 posters)
  • Max accuracy: 68.3%

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
Development

No branches or pull requests

1 participant