GSM: A generalized approach to Supervised Meta-blocking for scalable entity resolution

Luca Gagliardelli, George Papadakis, Giovanni Simonini, Sonia Bergamaschi, Themis Palpanas

Abstract: Highlights•Formalization of meta-blocking as a probabilistic classification task.•A supervised meta-blocking algorithm that requires only 50 examples for training.•Four new weighting schemes that enhance the meta-blocking performance.•Extensive experimental evaluation on 9 real-world datasets and 5 synthetic ones.•Comparison with state-of-the-art recently published blocking solutions.