Handling uncertain information has been acknowledged as one of the most important research areas by the Principle of Data Management community. Missing information in databases is a four decades old problem and even if a lot of effort has been made to address the issues, only a few have been solved theoretically or practically. Uncertain information can emerge due to untruthful or incomplete sources but also from data interoperability. With the emergence of the Web, we have seen an overwhelming increase in the production of data generating uncertain information.
Models of uncertain information already exist, however, theoretical work needs to be done in order to give a complete picture of their relative expressive power. Moreover those models have to be linked with already existing technologies in order to offer practical results as soon as possible even if it is only approximations. However, first and foremost, the industrial and academic communities should agree on the proper metrics to evaluate solutions. Finally, designing new tools to address specifically the uncertain information issue might be a long term solution if no existing one is able to fulfil the users needs.
Previous: Typed meta-interpretive learning