Search is useful when we know what to ask for. Exploring a collection can begin differently: one object catches our attention, and we want to discover where that interest might lead. Personalised recommendations offer one way to support that journey.
Recommendations grounded in collection knowledge
A DHBenelux 2026 poster by Xuemin Duan, Ruben Peeters and Anastasia Dimou introduces a recommender system built on a cultural heritage knowledge graph. Its stated aim is to support personalised artefact discovery and address the cold-start problem.
Cold start describes the difficulty of making useful suggestions when a system has little interaction history to learn from. A newly available collection or a new visitor can create this situation.
Looking beyond a history of clicks
A knowledge graph represents objects through connections. To illustrate the idea, an object might be related to a maker, a place or a historical period. Those connections provide possible routes for exploration even when an object has not attracted many previous interactions.
Consider a visitor interested in one historical garment. A useful recommendation could invite them to explore a related design tradition or another object with relevant contextual links. This example illustrates the purpose of knowledge-based discovery; it is not a reported result of the poster.
Why this matters for REEVALUATE
For REEVALUATE, recommendations offer a way to think about collections as starting points for exploration. The question is how a digital service can help someone take a meaningful next step after finding an object.
The poster introduces a research direction towards that goal. Its openly available record gives readers a route to the work and its authors, without treating an initial research contribution as proof of improved visitor engagement.
Read the publication
Knowledge-Graph-Driven Personalised Recommendation for Cultural Heritage Artefacts
Xuemin Duan; Ruben Peeters; Anastasia Dimou
DHBenelux 2026 poster, available through Zenodo.
REEVALUATE is funded by the European Union under Grant Agreement No. 101132389. Views and opinions expressed are those of the authors and do not necessarily reflect those of the European Union or the European Research Executive Agency. Neither the European Union nor the granting authority can be held responsible for them.





