Human-in-the-Loop Crowdsourced Annotation Dataset for Ukrainian Folk Art with Reproducible Jupyter Notebooks

This record contains the open datasets of the pilot "Human-in-the-Loop Crowdsourced Annotation of Ukrainian Folk Art", developed by Web2Learn.

Particularly, it contains: 

Explore the full pilot on GitHub, with a workflow layout designed to support reproducibility, with helpful notes, detailed README files for each step, machine-readable data output files and end-to-end executable code provided through Jupyter Notebooks: https://github.com/Web2LearnEU/AISTER-Crowdsourcing-Pilot

The pilot serves as an open repository for digital humanities research, freely available for reproduction, adaptation, and reuse by scholars, students, and educators. It is also open for creative reuses.

The pilot is implemented within the framework of the AISTER project. It operationalises and analyses a human-in-the-loop (HITL) crowdsourcing framework for metadata enrichment in Europeana collections. The objective is to enhance the accessibility and discoverability of Ukrainian ethnographic heritage by improving the quality of descriptive metadata combining artificial intelligence tools (natural language processing, computer vision) and human participation, while contributing to a better understanding in HITL approaches to AI-assisted metadata generation in cultural heritage. 
The pilot includes a crowdsourcing campaign, set up on the CrowdHeritage platform, maintained by Datoptron. Participants are invited to browse images from the ethnographic collection of the Krovets Online Museum of Traditional Art of Ukraine on Europeana, which includes more than 300 folk art paintings depicting scenes from everyday rural life and religious themes. By reviewing keywords automatically generated with computational methods, participants corrected terms, rejected inaccurate ones, and added additional keywords by recognising scenes, objects and figures.

The repository contains the complete workflow with all activity captured, including almost 55,000 annotation marks of AI-generated and crowdsourced content of 70 contributors:

1. Automatically generated annotations (description tags) for artefacts on Europeana using AI tools (natural language processing, computer vision) and Europeana APIs, and
2. Human-in-the-loop crowdsourced annotations on the CrowdHeritage platform to validate the AI-generated content, also enabling participants to contribute additional user-generated annotation.

Disclaimer
Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Education and Culture Executive Agency (EACEA). Neither the European Union nor EACEA can be held responsible for them.