View Derivation Graph with Edge Fitting for Data Warehousing inDisconnected Environments

Data warehouse views cache summarized information from queries over one ormultiple distributed base sources. In this paper, we address the problem ofdata warehouse maintenance in an environment where base source data can bestored on mobile devices. The requirements of a partially disconnected worldlead to significant tradeoffs between the view update cost on one hand, and thelevel of freshness of the view on the other hand. Current decentralizedmaintenance methods that can be applied to a disconnected environment involveeither replicating of source data at the view, or computing view updatesthrough queries to all base sources. Replication may involve excessive storageoverhead, while view update may be blocked due to the disconnection of any ofthe base sources. We propose a view maintenance approach that stores pathinformation corresponding to data traditionally stored at the view, and offersseveral important advantages. From the maintenance point of view, our methodreduces local computation at the view as well as communication with the outsidesources, and lowers the execution load on the base sources. From the user\'sperspective, reducing the probability of executing view update queries at allbase sources leads to a more up-to-date state of the data warehouse view.