Introduction: 

The MADCOW annotation system enables group-based annotation, allowing users to direct annotations toward communities focused on specific domain topics. In smart city environments, such groups may include citizens, urban planners, and domain experts collaborating on urban services, infrastructure, mobility, environment, and public safety. Existing recommendation approaches mainly rely on ontology-based semantic similarity, which limits their effectiveness in dynamic collaborative settings.

Methods: 

To address this limitation, this study extends ontology-based matching by incorporating behavioral and structural factors, where user activity is modeled using the number of posted annotations and joined groups, and group relevance is represented by group size and membership growth rate. These heterogeneous features are integrated with ontology-based semantic similarity using fuzzy logic operators to construct a more flexible and expressive ranking framework.

Results: 

Experimental evaluations demonstrate that the proposed approach improves the quality of user-to-group and group-to-user ranking compared to methods relying solely on ontology-based graph similarity measures. However, the model's performance may decrease in scenarios with highly sparse annotation data, indicating the need for further investigation into robustness under data-scarce conditions.

Discussion: 

The aggregation of annotations and of users according to topics of interests may facilitate focused discussions about issues related to city governance among concerned citizens. The identification of active groups and users may lead to more effective recommendations for user enrolment.

Title

Frontiers in Artificial Intelligence

Publisher

Frontiers Media SA

Publisher Country

Switzerland

Indexing

Scopus

Impact Factor

4.7

Publication Type

Online only

Volume

9

Year

2026

Pages

19