The following paper is also included in the Machine Learning special issue on “Inductive Logic Programming”:

Online Probabilistic Theory Revision from Examples with ProPPR, by Victor Guimarães, Aline Paes, and Gerson Zaverucha presents an algorithm for performing online theory revision from streams of (uncertain) open-ended data. To model the domain, the system uses ProPPR, a statistical relational language, able to represents uncertainty in the data. The problem of overfitting due to the continuous stream of new information is avoided by means of Hoeffding’s bound. The system has been tailored for real-world domains, such as social networks interactions and drug discovery, where data is heterogeneous and multi-related and may be uncertain. The paper also presents an evaluation of the system on two datasets modeling social networks.

All five papers comprising this special issue, as well as the guest editorial, can be found in this topical collection: https://rd.springer.com/journal/10994/topicalCollection/AC_071afc28e78ee62a6c512f8baf66a72c/page/1.

Nicolas Lachiche, Christel Vrain, Fabrizio Riguzzi, Elena Bellodi, and Riccardo Zese