| Abstract |
I will present an algorithm for estimating the trustworthiness of crowdsourced museum annotations, based on a combination of semantic similarity measures, provenance and subjective logic.
The algorithm shows satisfactory results both with its reputation- and its provenance-based variant.
I will show the results obtained from two evaluations on cultural heritage datasets, provide an indication about how to use semantic similarity to improve the computation cost of the algorithm, and discuss possible future directions for this work. |