AI in relation to GLAMs

About

GLAMs have started to experiment with different AI techniques applied in the different areas of their operation ranging from data analysis to enrichment, or even generation of new information. AI technologies are not new but recent innovative developments have spurred a proliferation of applications and experimentations in different areas making AI the next big thing in many areas of activity and nurturing great expectations about its potential impact.

This Task Force focused on AI techniques used in the analysis and enrichment of digital collections resulting in improved search and browsing functionalities for the users. These techniques include but are not limited to machine learning, natural language processing including translation, OCR, HTR, statistical analysis, computer vision and text analysis algorithms and are used to extract information like entities, features, colors, composition and patterns, do similarity matches, image classification & tagging. A set of data is used to train the algorithms in a semi or even unsupervised way.

The purpose of the Task Force was to do a horizon scanning exercise and to start investigating the expected role and impact of AI in the digital cultural heritage domain especially with regards to collections analysis and improvement. Furthermore, the Task Force looked for opportunities for the Europeana Foundation and the Europeana communities to benefit from and utilise this research and knowledge being developed across Europe (and the world). Particular issues investigated were, for example:

1. What are the key projects and institutions making strides investigating and presenting AI’s R&D impact on cultural heritage, specifically matters related to (meta)data enrichment.
2. What are the key techniques and focuses related to AI that are most relevant for the DCH community?
3. What kind of datasets are used for training algorithms and what validation techniques and metrics are being used by key cultural heritage projects.
4. R&D tools and methodologies being implemented by different institutes and their efficacy.
5. Identify key challenges in regards to negative consequences and possible misuses of AI as applied in heritage. These challenges concern biases, deep-fakes and source criticism, and environmental impact.
6. Identify the major obstacles that CHIs that wish to take advantage of AI face: technical and organisational i.e. financial costs, personnel, technical specifications.
7. Domain specific breakdowns of the above information.

The EuropeanaTech AI for GLAMs Task Force also ran a survey to help gain an understanding of who is working with AI, the different types of projects being run, the methodologies being used, the challenges faced, the success granted and the resources applied. 

Expected outcomes

The final outcome of the Task Force are a report that delivers an eagle-eye view of AI in the European digital cultural heritage R&D projects especially with regards to use cases including metadata/content enrichment. It includes include a set of recommendations for cultural heritage institutions to consider when setting up an AI project within their organisation. 

You can download the full Task Force report and the interim report below.