A knowledge-based recommender for the Rijksmuseum artworks

Knowledge based recommender systems are well suited for the exploration of complex knowledge domains without having domain knowledge.

In this use case presentation for the Cultural Heritage subtopic we demonstrate a recommender system for historical artworks of the Rijksmuseum.
To help museum visitors understand and make decisions for navigation within the collections, we provide recommendations based on a semantic description of the content.

We show how to combine a knowledge graph model with statistical scoring to provide navigation options on different semantic levels.

By interlinking multiple Cultural Heritage vocabularies we can automatically expand the semantic content descriptions of the artworks.
The interlinks are then used to calculate information content scores that represent the relevancy of concepts regarding the artwork descriptions.

The scores contribute to the similarity calculation when comparing the semantic descriptions.
Artworks are more similar if they share more important concepts with higher relevancy scores.

The result is a recommender that improves the quality of the knowledge-based recommendations based on the most relevant parts of the content.
This helps museum visitors to explore and learn about the Rijksmuseum collection and the domain of art history.