Report Open Access

Visitor Interaction and Machine Curation in the Virtual Liverpool Biennial: A Towards a National Collection COVID-19 Project Final Report

Impett, Leonardo; Cetinic, Eva; Krysa, Joasia

This report summarises the AHRC-funded Towards a National Collection COVID-19 Project, Visitor Interaction and Machine Curation in the Virtual Liverpool Biennial, which ran from 1 January 2021 to 31 August 2021. The project was based at the Department of Computer Science, Durham University, in collaboration with the Liverpool School of Art and Design, Liverpool John Moores University and the Liverpool Biennial.

A summary of our research project and its main questions is provided in the abstract. The aims and objective section outlines our three principal research ambitions: to prototype a different use for machine learning in virtual exhibitions (as co-curators, not search engines); to understand how visitors might interact with such a system; and to look at the bias present in the machine learning algorithms that power it. We then report details of our administrative structures (partnerships, staffing, and timetable).
Our research methodologies and results are summarised in the Research Approach, Research Results, and Project Outputs section. We have three principal project outputs, tied to our three research questions: an online machine curation prototype hosted by the Liverpool Biennial, ai.biennial.com, and associated open-source codebase to reproduce it on other collections; a visitor interaction dataset, which is freely viewable inside our online browser, and an associated anonymous survey which is not public; and an open-source toolkit for examining bias in the major machine learning algorithm used in the project.

The report concludes with our project’s recommendations for the Towards a National Collection program, in the context of our unusual status as an AHRC project in a computer science department.

Files (1.4 MB)
Name Size
Leo Impett TANC Final Report.pdf
md5:7138f70f7efb54b8f9f0d3bdf5c55bff
1.4 MB Download
107
43
views
downloads
All versions This version
Views 107107
Downloads 4343
Data volume 61.9 MB61.9 MB
Unique views 9797
Unique downloads 3535

Share

Cite as