Inclusion of Time-Varying Measures in Temporal Data Warehouses

Data Warehouses (DWs) integrate data from different source systems that may have temporal support. However, current DWs only allow to track changes for measures indicating the time when a specific measure value is valid. In this way, applications such as fraud detection cannot be easily implemented since they require to know the time when changes in source systems have occurred. In this work, based on the research related to Temporal Databases, we propose the inclusion of time-varying measures changing the current role of the time dimension. First, we refer to different temporal types that are allowed in our model. Then, we study different scenarios that show the usefulness of inclusion of different temporal types. Further, since measures can be aggregated before being inserted into DWs, we discuss the issues related to different time granularities between source systems and DWs and to measure aggregations.