This talk discusses recent research related to managing data quality for Knowledge Graphs and also some applications related to the biomedical domain. First, the talk will show how to deal with noisy labels in datasets which are used to train machine learning models. Then, the completeness dimension of knowledge graphs will be discussed. The focus will be on work aiming to estimate the expected number of instances for a class in order to measure the level of data completeness. Related to this, the talk will show how crowdsourced knowledge graphs receive contribution towards increasing completeness and how di erent people contribute at di erent levels over time. Finally, it will be demonstrated how human bias re ected in the contributed data can be represented in the knowledge graph and surfaced to users. The second part of the talk will discuss a couple of application scenarios in the biomedical domain where knowledge graph are used, including entity extraction and information access.
i been Program Committee member for several conferences including WWW, SIGIR, KDD, AAAI, IJCAI, ISWC, and ICWSM. Before joining the University of Queensland, he was Lecturer at the University of She eld in UK, post-doctoral researcher at the eXascale Infolab at the University of Fribourg in Switzerland, visiting researcher at UC Berkeley, junior researcher at the L3S Research Center in Germany, and intern at Yahoo! Research in Spain. In 2011, he obtained a Ph.D. in Computer Science at the Leibniz University of Hanover focusing on Semantic Search.