Maintaining the consistency of large-scale knowledge graphs (KGs) like Wikidata requires both automated methods and human expertise. In this paper, we address the task of recommending users best suited to repair a given inconsistency in a KG. Our approach leverages textual entity abstracts to compute sentence embeddings, which are clustered to identify semantically coherent regions of the KG. We introduce a framework that combines unsupervised clustering with 10-fold evaluation to test user recommendation strategies. Repair histories are linked to users, and test inconsistencies are assigned to clusters using approximate prediction. We evaluate two strategies: (i) frequency-based assignment, which recommends users based on how often they have edited entities in the predicted cluster, and (ii) embedding-based similarity, which compares the test inconsistency to past user-edited items via cosine similarity. Preliminary results show a cluster silhouette ≥ 0.5, membership hit rate of 80%, with the frequency-based approach achieving a Hits@3 of 60%. Our findings suggest that lightweight unsupervised methods can effectively recommend users, showing promise for semi-automated KG maintenance.

Abstract Every year, new ontology matching approaches have been published to address the heterogeneity problem in ontologies. It is well known that no one is able to stand out from others in all aspects. An ontology meta-matcher combines different alignment techniques to explore various aspects of heterogeneity to avoid the alignment performance being restricted to some ontology characteristics. The meta-matching process consists of several stages of execution, and sometimes the contribution/cost of each algorithm is not clear when evaluating an approach. This article presents the evaluation of solutions commonly used in the literature in order to provide more knowledge about the ontology meta-matching problem. Results showed that the more characteristics of the entities that can be captured by similarity measures set, the greater the accuracy of the model. It was also possible to observe the good performance and accuracy of local search-based meta-heuristics when compared to global optimization meta-heuristics. Experiments with different objective functions have shown that semi-supervised methods can shorten the execution time of the experiment but, on the other hand, bring more instability to the result.