Résumé
With the success of machine learning (ML) techniques, ML has already proved a tremendous potential to impact the foundations, algorithms, and models of several data management tasks, such as error detection, data quality assessment, data cleaning, and data integration. In Knowledge Graphs, part of the data preparation and cleaning processes, such as data linking, identity disambiguation, or missing value inference and completion could be automated by making a ML model "learn" and predict the matches routinely with different degrees of supervision. This talk will survey the recent trends of applying machine learning solutions to improve and facilitate Knowledge Graph curation and enrichment, as one of the most critical tasks impacting Web search and query-answering. Finally, the talk will discuss the next research challenges in the convergence of machine learning and management of Knowledge Graph evolution and preservation.