Data Bazaar

Event Goals

• Curate high-quality, well-documented datasets that will be shared openly and used by researchers around the world for downstream machine learning and modeling applications

• Edit and improve the Rosetta Commons Data Curation Standards Guide and Decision Tree based on experience during the Data Bazaar

Schedule

*For more details on timing and meals see Schedule 

Monday (3/2): 

Hotel check-in and group dinner

Tuesday (3/3):

Guided, Hands-On Dataset Curation

• Participants will be split into teams to work on easy, starter datasets: Day-1 Dataset Teaming & Assignments

• Teams will follow the Data Curation Guide & Molecular Dataset Curation Guide on Rosetta Common's HuggingFace 

Wednesday (3/4):

Best Practices for ML-Ready Dataset Curation + Custom Dataset Development

• Gina El Nesr will give a lecture on dataset curation for downstream machine learning applications

• Teams will reform and apply workflow from Tuesday to participant's datasets or more challenging datasets from Data Bazaar list. 

Thursday (3/5):

Dataset Finalization, Presentation, & Standards Synthesis

• Teams will finalize curation, complete uploads to HuggingFace and add collaborators

• Teams will give mini-presentations (slide deck here) on 

  • Dataset purpose and intended downstream ML application

  • Key curation decisions and tradeoffs

  • Dataset split strategy and evaluation considerations

  • Known limitations, open questions, or future extensions

• Feedback Form

• Group discussion to improve data standards guide and recap lessons learned

Friday (3/6) - Sunday (3/8):

Enjoy San Juan! 

Questions?

For questions please email Hope Woods (hope.woods@omsf.io) or Ashley Vater (awvater@ucdavis.edu)