As part of the Innovation Fellowship project, the Swiss National Library (SNL) worked with artificial intelligence from July 2022 to June 2023 to gain initial experience. We also applied design thinking and agile project management and promoted knowledge transfer.

Three ideas were explored in depth:

- Automatic classification of subject groups of e-dissertations: The SNL receives a copy of every dissertation produced in Switzerland from university libraries most of which are now digital. The dissertations have to be classified into one of about 100 subject groups. The aim of this project was to test open source algorithms that automatically classify the dissertations, anal-gous to similar projects in the German and Finnish National Libraries.

- Facial recognition for photo archives: Copyright law allows the SNL to publish thumbnails for catalogue entries. The SNL owns several photo ar-chives. Some of these photos are of people who are still alive and can be recognised in these thumbnails. For data privacy reasons these photos can not be published. A machine learning algorithm should be developed that can automatically search large photo archives and sort out photos with peo-ple in them. It should also be tested whether the age of the people in the photos can be automatically recognised.

- (Semi)automatic content indexing for the poster collection using existing methods: The SNL operates the collective catalogue of the Swiss Poster Collections. The effort for manual content indexing is high, so that it takes a relatively long time from digitisation to indexing. The content indexing of images was tested using commercial methods. This should provide initial indications as to whether such methods can be used to support content indexing with suggestions.