Incremental maintenance of multi-source views

In recent years, numerous algorithms have been proposed for incremental view maintenance of data warehouses. As a matter of fact, all of them follow almost the same general approach, namely they compute the change of a multi-source view in response to an update message from a data source, following two steps: (i) issue a set of queries against the other data sources, and (ii) compensate the query result due to concurrent updates interfering with the first step. Despite many recent improvements, the compensation approach needs precise detection of interfering updates occurring remotely in autonomous data sources and the assumption that messages are never lost and are delivered in the order in which they are sent. However, in real networks, loss and misordering of messages are usual. In this paper, we propose a maintenance algorithm that does not need the compensation step and that applies to general view expressions of the bag algebra, without any limit on the number of base relations per data source.